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

AMERICAN MUSEUM Novitates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 3092, 34 pp., 9 figures, 1 table February 17, 1994

Basicranial Anatomy and Phylogeny of Prinmitive Canids and Closely Related Miacids (Carnivora: Mammalia)

XIAOMING WANG' AND RICHARD H. TEDFORD2

ABSTRACT

Selected fossil carnivorans are studied in an at- internal carotid artery in an extrabullar position tempt to bridge the gap between caniform miacids between the entotympanic and the petrosal. Early and early canids, and to identify character trans- canids are further distinguished by their posses- formations within the clade leading to canids. Sev- sion of a shallow suprameatal fossa. Canids suc- eral basicranial and dental features are important cessively acquired the following dental characters: in characterizing the cladogenetic events that oc- presence ofposterior accessory cusps on upper and curred and culminated in Hesperocyon. lower third premolars, loss of M3, reduction of Basicranial transformations from miacids to the the parastyle on M1, and enlargement ofthe meta- first recognizable canids involved the formation conid on m2 so that it equals the protoconid in ofa rigid middle ear chamber, i.e., the ossification size. of an entotympanic bulla and development of a Although the middle ear structure of canids is low septum from the in-bent edge of caudal en- relatively stable when compared with that ofother totympanic. The middle ear region -s further families of caniform carnivorans, it does change strengthened by medial expansion of the petrosal in at least three aspects. Firstly, the low entotym- into contact with the basioccipital and basisphe- panic septum varies in size and extent, and tends noid, the ossification of the tegmen tympani, and to be confined to the anteromedial corner of the the closure of piriform fenestra. Another impor- bulla in derived forms. Secondly, the anterior loop tant character complex consists ofthe medial mi- of the internal carotid artery changes from intra- gration of the internal carotid utery, She loss of cranial to extracranial in position. Evidence for the stapedial artery, and the entrapment of the this latter feature is the loss of a bony signature of

' Frick Postdoctoral Fellow, Department of Vertebrate Paleontology, American Museum of Natural History. 2 Curator, Department of Vertebrate Paleontology, American Museum of Natural History.

Copyright © American Museum of Natural History 1994 ISSN 0003-0082 / Price $3.60 2 AMERICAN MUSEUM NOVITATES NO. 3092 this loop in the basioccipital-an embayment on bony tube for the external auditory meatus de- the lateral wall of the basisphenoid anteroventral velops in later canids. to the middle lacerate foramen. Finally, a short

INTRODUCTION Among caniform carnivorans, the basicra- does not prove that the reverse process may nial anatomy ofcanids has long been consid- not exceptionally take place, whether by re- ered relatively stable, and canid character- version or otherwise." Otocyon was therefore istics in the middle ear region are identifiable removed from the basal stock in Scott's phy- in the earliest canid Hesperocyon (Hough, logeny which still recognized Miacis as an- 1948; Tedford, 1976). This apparent stability cestral to canids. of the middle ear of canids, together with The miacid ancestry of canids has since their rather uniform dental patterns, makes been accepted by most students of carnivor- canids a morphologically conservative group. ans (e.g., Matthew, 1930; Clark, 1939; Ted- Basicranially, canids can be distinguished ford, 1978; Flynn and Galiano, 1982; Gus- from other families of carnivorans by pos- tafson, 1986). Various authors, however, session of an inflated bulla, formed mostly adopted different ways of defining the Cani- from an inflated caudal entotympanic; a low dae-some favored the inclusion of certain septum across the inner surface of the bulla; miacid genera in the Canidae consistent with a fully functional internal carotid artery in an their philosophy of a vertical classification extrabullar position; an external auditory (see Flynn and Galiano, 1982, for a recent meatus lacking a prominent bony tube (ex- review). As may be expected from such a cept derived forms); and a small mastoid pro- basal group, miacoids could be claimed as cess. ancestral to every modern family of carni- Cope (1877, 1883) first proposed that the vorans. Indeed, Flynn and Galiano (1982) ancestry ofcanids was to be found among the associated certain miacoids with the phylog- forms placed in his Creodonta. More specif- eny ofprimitive feliforms, and the rest ofthe ically, the Miacidae was postulated as the miacoids were set aside in an unresolved probable ancestral stock (Cope, 1880: 81). Caniformia. In a more recent analysis (Wyss The central theme of Cope's phylogeny re- and Flynn, 1993), however, the Miacidae was volved around a linear sequence of dental removed from direct relationship to the liv- reduction from four upper and lower molars ing families of Carnivora, and placed one in the African bat-eared fox (Otocyon mega- branch below the common ancestor of Viv- lotis), to a more "normal" dental formula in erravidae and Carnivora. Amphicyon (three upper and three lower mo- Among the various North American Eo- lars), and to variously reduced dentitions (e.g., cene miacids, Miacis, with its tendency to loss of pl, M3, and sometimes M2) in later elongate the carnassial shear and reduce the true canids. Even though the number of up- M3 (a "typical cynoid" character according per molars in Miacis was not known to Cope to Matthew, 1909: 362), appears morpholog- (1883), he correctly predicted that it would ically closest to the canids. Clark (1939), and be the same as in Amphicyon. Cope thus pre- later Gustafson (1986), went further and sug- dicted discovery of Otocyon-like fossils pre- gested that "Miacis" gracilis was closest to ceding other extinct dogs. Hesperocyon. The proposed affinity between Cope's theme of progressive dental reduc- "M. "gracilis and Hesperocyon was primarily tion was soon challenged by Scott (1895: 74) based on postcranial evidence, i.e., the ad- who realized that the supernumerary teeth in vanced canidlike limb bones of "M. "gracilis; Otocyon could be a result of character rever- dentally, however, "M. "gracilis was little dif- sal: "I am by no means convinced of the ferent from other species of Miacis. impossibility of the addition of new teeth to Apart from the abovementioned attempts the molar series. That modification in the to link a particular taxon to the ancestry of mammalian lines is very generally by way of the Canidae, the relationships between var- reduction in the number ofteeth, is true, but ious species of miacids on the one hand, and 1 994 WANG AND TEDFORD: CANIDS AND MIACIDS 3 most primitive canids on the other, remains of upper fourth premolar and lower first mo- unresolved. Tedford (1976: 364) remarked: lar. Miacoids are otherwise more or less de- "The fact that very few miacoids have been fined by the paucity of derived characters identified as phyletically related to members shared with advanced (mostly extant) fami- of the modern superfamilies only increases lies of carnivorans. The taxon Miacoidea is the isolation of the archaic and modern car- therefore one of convenience and has long nivore families. In other words, there are few been a taxonomic wastebasket waiting to be miacoids that possess derived characters emptied. Recent phylogenetic treatments uniquely shared with representatives of the range from dividing miacoids into feliform modern families. This represents one of the and caniform components (Flynn and Galia- largest gaps in our knowledge of the phylog- no, 1982), to placing them within a multi- eny of the Carnivora." chotomy of Feliformia + Viverravidae + Recent discoveries offorms from the Vieja Miacidae + Nimravidae etc. (Flynn et al., Group of Texas (Chadronian) intermediate 1988), and to postulating a sister-group re- between miacids and canids has significantly lationship between Miacidae and a Viverrav- bridged this gap in our knowledge of canid idae + Carnivora clade (Wyss and Flynn, ancestry (Gustafson, 1986). The Texas ma- 1993). In this discussion, the Miacidae in- terial includes the last known representatives cludes such archaic caniforms as Miacis, Uin- of Miacis in North America, "M." cognitus tacyon, Vulpavus, etc. (roughly equivalent to and M. australis. Their dental morphology is Miacinae of Simpson, 1945: 108), whereas also the most derived among all Miacis. An- the Miacoidea includes the caniform Miaci- other important taxon, also from the Vieja dae and the feliform Viverravidae. The pres- Group, is "Hesperocyon" wilsoni Gustafson ent study considers only a few putative mia- (1986), which reduces, in yet another way, cids, which share several derived characters, the morphological gap between the well- such as reduced or absent P4 parastyle, re- known White River H. gregarius and various duced P4 protocone, and internal cingulum Miacis. "H. " wilsoni has a fully ossified bulla of upper molars surrounding the protocone, with a high degree ofinflation resembling that in addition to various postcranial characters of H. gregarius. However, the dental mor- (Flynn and Galiano, 1982). Preservation of phology of "H." wilsoni is still primitively the basicranium is also important to the scope similar to that of miacid carnivorans. It re- of this study. tains a primitively short shearing blade on As discussed in the introduction, of the the ml with a relatively high trigonid and known miacid genera Miacis seems in general narrow talonid, and has a relatively large closest to canids because of its tendency to parastyle on the Ml. The Texas material dis- elongate the carnassials and to reduce the M3. plays an interesting combination of charac- The genus was established by Cope (1872) on ters, which not only offers an example ofwhat the type species M. parvivorus from the Bridg- a transitional canid may have looked like, er Formation (Bridgerian), Wyoming. Nearly but also raises new questions regarding the two dozen nominal species of Miacis have origin of canids. been described from the Eocene of North The present paper describes in detail the America and the Old World. Considerable basicranial anatomy of selected primitive confusion exists regarding the taxonomic sta- caniform carnivorans that led to the emer- tus of these forms. It is beyond the scope of gence ofthe Canidae and places these taxa in this study to evaluate the taxonomic com- a phylogenetic framework. plexities among these species, and only a selected few North American taxa which have reasonably complete basicranial and dental SCOPE, DEFINITION, TERMINOLOGY, materials are analyzed. AND ABBREVIATIONS Unless otherwise noted, we follow Flynn The term "miacoid" is usually applied to et al. (1988) for taxonomic usages of major Paleocene to late Eocene carnivorans united, groups ofcarnivorans. We will enclose within among other features, by their common pos- quotation marks taxa that are clearly para- session ofa carnassial shear fixed on the locus phyletic in this analysis, and restrict the use 4 AMERICAN MUSEUM NOVITATES NO. 3092 ofgeneric names to the type species, i.e., Mia- mp mastoid process cis = M. parvivorus, and Hesperocyon = H. mt mastoid tubercle gregarius. Fortunately, these type species also oc occipital condyle contain the best known basicranial materials. pa promontory artery pf piriform fenestra ( promontory foramen) For Vulpavus, we assumed that the type of pgf postglenoid foramen V. profectus, AMNH 12626, is morphotypic pgp postglenoid process for the genus. For basicranial terminology, plf posterior laterate foramen we follow MacPhee (1981) on eutherian pp paroccipital process mammals in general, and Evans and Chris- pr promontorium tensen (1979) on living Canis in particular. ps pneumatic sinus re rostral entotympanic ABBREVIATIONS sa stapedial artery Institutions sf suprameatal fossa sica sulcus for internal carotid artery AMNH American Museum of Natural History, smf stylomastoid foramen New York smp sulcus for major petrosal nerve (palatine ra- CMNH Carnegie Museum of Natural History, mus of facial nerve) Pittsburgh spa sulcus for promontory artery FMNH Field Museum of Natural History, Chi- spe suture of promontorium/entotympanic cago sq squamosal F:AM Frick Collection, American Museum of srsa superior ramus of stapedial artery Natural History ssa sulcus for stapedial artery KUVP Museum of Natural History, University of Kansas, Lawrence MCZ Museum of Comparative Zoology, Har- DESCRIPTION OF BASICRANIAL vard University, Cambridge ANATOMY TMM Texas Memorial Museum, University of Vulpavus profectus Matthew Texas, Austin Figures IA, 2A Anatomical Terms SPECIMEN ExAMINED: AMNH 12626, ho- ac alisphenoid canal lotype, complete skull and mandible with al alisphenoid partial skeleton, from unit B, Bridger For- apa ascending pharyngeal artery mation, Bridger Basin, Wyoming, early bo basioccipital Bridgerian. bs basisphenoid Both sides of the basicranium in AMNH cc condyloid canal 12626 are excellently preserved, affording ce caudal entotympanic confident interpretations ofmost osteological cf condyloid foramen structures. Matthew's (1909) original de- cfn canal for facial nerve scription of the basicranium is brief, and he cica canal for internal carotid artery (or prom- has been credited with the first proposal that ontory artery) eam external auditory meatus two separate branches of the internal carotid ec ectotympanic artery (medial and lateral) occurred in the Ec Eustachian canal middle ear region (Presley, 1979; see descrip- en entotympanic tion below). Vulpavus profectus is used here er epitympanic recess as an outgroup for determination ofcharacter fc fenestra cochleae (fenestra rotunda) polarities among the ingroups. fo foramen ovale BULLA: No bulla is known for this genus. fs fossa for stapedius muscle Neither is there any indication that any bullar ftt fossa for tensor tympani muscles element may have been in firm contact with fv fenestra vestibuli (fenestra ovalis) gf glenoid fossa the bones which make up the roofofthe mid- Gf Glaserian fissure dle ear cavity. The lateral edge of the basi- ica internal carotid artery occipital and basisphenoid lacks the thick- ips inferior petrosal sinus ening or formation of a bony ledge usually irsa inferior ramus of stapedial artery associated with sutural contact with the cau- mlf middle lacerate foramen dal entotympanic. The surface ofthe petrosal 1 994 WANG AND TEDFORD: CANIDS AND MIACIDS 5 is smooth toward the basioccipital side rather foramen and piriform fenestra. On the pos- than roughened, as it is where the bulla at- terior edge ofthe fused basisphenoid and ali- taches in many carnivorans. sphenoid plate is a small protuberance PETROSAL AND SURROUNDING BoNEs: The pointing toward the petrosal, which has a cor- ventral surface ofthe basioccipital is flat and responding bulge closing in on the projection broad. A broad basioccipital is also present (better seen on the left side of the skull of in ursids and amphicyonids, and may be a AMNH 12626). These two opposing projec- primitive character for caniform carnivor- tions give hint to the initial separation ofthe ans. The lateral border of the basioccipital is middle lacerate foramen, or anterior carotid constricted at the level of the sutural line be- foramen (here continuous with the basicap- tween the basioccipital and basisphenoid, and sular fenestra, i.e., the open space between has a sharp edge (i.e., it lacks the thickened petrosal and basioccipital) from the piriform and grooved edge for the inferior petrosal fenestra (fig. IA). In the fetal stage of living sinus that occurs in all canids). Anterior to eutherians, the anterior carotid foramen is the basioccipital, the lateral edge of the ba- separated from the piriform fenestra by an sisphenoid similarly lacks an embayment near alicochlear commissure, a cartilaginous bridge the middle lacerate foramen, which houses a continuous with the soft tissue complex in- convoluted internal carotid artery in canids side the suture ofthe basisphenoid-alisphen- and arctoids. oid (MacPhee, 1981: 21 and fig. 1). The ex- The petrosal is globular in outline and is pansion of the petrosal and basisphenoid not expanded medially. This is in contrast to replaces this commissure in the adult and the viverravid Protictis in which there is a ensures the complete separation of the an- rugose articular surface on the medial edge terior carotid foramen (middle lacerate fo- of the promontorium (Gingerich and Wink- ramen) and promontory foramen (piriform ler, 1985: fig. l1; personal observation of an fenestra). The lack of ossification in this area epoxy cast). The location of this rugose area in some fossil taxa (e.g., Miacis parvivorus, is identical to that ofa rostral tympanic pro- see description below) precludes the separa- cess of petrosal in most insectivorans and tion of these two foramina. primates (MacPhee et al., 1988). In aeluroids, The surface between the fenestra vestibuli the anteromedial corner ofthe petrosal is fur- and the groove for the promontory artery is ther expanded medially to form a ventral pro- not clearly marked by an impression for a montorial process, an important synapo- stapedial artery branching offthe internal ca- morphy for aeluroid carnivorans (Hunt, rotid artery, although Matthew (1909) 1989). claimed to have observed one. The only in- The petrosal is completely separated from dication of a sulcus for the stapedial artery the adjacent basioccipital and basisphenoid is, on the right promontorium, a slight inter- by a 1 mm wide space (basicapsular fenestra ruption of the lateral border ofthe sulcus for of MacPhee, 1 981: fig. 1). This space allows internal carotid artery just opposite to the ample room for a large inferior petrosal sinus. fenestra vestibuli (see more below). The only bony signature of this sinus on the The tegmen tympani area descends rather petrosal is a broad and shallow depression suddenly from the posterior edge of the al- dorsal to a rather prominent medial ridge (fig. isphenoid and squamosal, leaving a large free 2A, above the ips on cross-sectional view). space as the piriform fenestra. There is there- The ventral surface of the promontorium fore little continuity between the tegmen is somewhat rugose, as compared to Hesper- tympani area and tympanic processes of ali- ocyon. It has a shallow groove (sulcus for the sphenoid and squamosal. This is in contrast promontory artery) toward its lateral aspect. to Hesperocyon gregarius, "Miacis " cognitus, The groove begins posteriorly just in front of and others, which have a fully ossified tegmen the fenestra cochleae. Traveling anteriorly, tympani smoothly articulated with the bones the groove slightly descends ventrally and anterior to them. The opening for the facial then turns medially to lead to the free space nerve (foramen faciale) on the ventral surface between the petrosal and alisphenoid, here of the petrosal faces anterolaterally and is the common opening for the middle lacerate slightly enlarged to house the genicular gan- 6 AMERICAN MUSEUM NOVITATES NO. 3092

.s'a

,]l~~~~~~~~~~

.w;

*I..

Fig. 1. Basicranium of primitive caniform carnivorans. Ventrolateral view of left middle ear region. All drawn to approximately the same size. A, Vulpavus profectus, AMNH 12626; B, Miacis parvivorus, USNM 214706, reversed from right side; C, Hesperocyon gregarius, F:AM 76163 (an internal view of the dissected ventral portion of the bulla is shown above with a low septum). See Definition section for abbreviations. In Vulpavus and Miacis, the piriform fenestra (pf) encompasses the middle lacerate foramen (anterior carotid foramen of MacPhee, 1981) because of the lack of preservation of the cartilaginous alicochlear commissure. 1994 WANG AND TEDFORD: CANIDS AND MIACIDS 7 glion. Leading from this ganglion are two well- Medial to the postglenoid process is a deep marked grooves. One leads anteriorly toward cleft-the Glaserian fissure. The well-devel- the piriform fenestra as the palatine ramus oped Glaserian fissure ofcarnivorans has been of the facial nerve (major petrosal nerve in considered suggestive of an Insectivora-Car- Evans and Christensen, 1979), which may nivora sister-group relationship (Wyss and exit the middle ear cavity through the Eus- Flynn, 1993). Immediately medial to the tachian canal as in Canis (Evans and Chris- Glaserian fissure is the sutural contact be- tensen, 1979: figs. 15-16). The other traverses tween the squamosal and alisphenoid. The posterolaterally to circle around the dorsal basisphenoid/alisphenoid suture is not clear- rim of fenestra vestibuli and contains the ly visible anterior to the Eustachian canal. hyomandibular ramus of the facial nerve, The grooving on the alisphenoid leading to which ultimately emerges from the middle the Eustachian canal is divided by a promi- ear cavity through the stylomastoid foramen. nent ridge and a minor ridge. In Canis the The entire course of the anterior and poste- lateral groove carries the Eustachian tube, rior branches of the facial nerve inside the whereas the medial groove transmits the ten- middle ear cavity is thus exposed without a don of the tensor veli palatini muscle (and bony covering. Lateral to the piriform fenes- the accompanying nerve), the palatine ramus tra and along the petrosal/squamosal suture, ofthe facial nerve, and the nerve to the tensor there is a small canal excavated on the squa- tympani muscle (Evans and Christensen, mosal side of the suture, possibly for the su- 1979: figs. 15-16, 19-5). perior ramus ofthe stapedial artery (see more INTERNAL CAROTID ARTERY (fig. 2A): Mat- below). thew (1909: 384) described the internal ca- The epitympanic recess is deeply excavat- rotid artery of V. profectus as follows, "The ed laterally above a squamosal shelf, and is grooves for the tympanic and stapedial walled laterally by the squamosal and roofed branches of the internal carotid artery are by the petrosal. The petrosal also sends out distinctly shown; beginning just anterior to a lateral process, better developed in the right the fenestra rotunda the former passes for- side, toward the postglenoid foramen. This ward on the outer slope ofthe auditory prom- process articulates with the squamosal near inence to the foramen lacerum medius while the posterior base ofthe postglenoid process. the latter descends to the fenestra vestibuli. On the dorsolateral part of the epitympanic The inner branch of the entocarotid artery is recess, this sutural contact between the squa- not so certainly recognizable but appears to mosal and petrosal is rather loose and filled pass into the otic depression over the lip of with matrix. Toward the posterior end ofthe the basioccipital a little in front of the fora- recess, there is an enlargement of this suture men lacerum posterius." This short descrip- to form a canal with a diameter near that of tion is usually considered to be the origin of the fenestra vestibuli. This canal possibly the concept ofthe division ofinternal carotid provides the exit ofthe superior ramus ofthe into a two-artery system in the middle ear stapedial artery. region ofprimitive eutherians (Presley, 1979; The tip of the mastoid tubercle is broken Wible, 1983: fig. 1). Instead ofan entocarotid on both sides ofAMNH 12626. Judging from artery coexisting with a promontory artery, the relative contribution of the squamosal current consensus regards the groove which and petrosal, the squamosal, rather than the Matthew interpreted as "entocarotid artery" petrosal, makes up the main part ofthe mas- as an impression left by the inferior petrosal toid tubercle as primitively occurs in Car- sinus. nivora (Flynn and Galiano, 1982: 19). The A groove on the ventrolateral surface of mastoid process is less inflated than in Hes- the promontorium, however, clearly indi- perocyon, and there is no suprameatal fossa cates the presence ofa promontory artery. As in front of the process. The paroccipital pro- for the "distinct" impression for the stapedial cess is large and projects posteriorly. artery indicated by Matthew (1909), we noted The postglenoid process is high, and the only a faint interruption of the lateral border postglenoid foramen is large and rounded. of the sulcus for internal carotid artery, in- 8 AMERICAN MUSEUM NOVITATES NO. 3092

srsa

Fig. 2. Ventral (upper) and cross-sectional (lower) views of four caniform carnivorans with the hypothesized position of the internal carotid artery. A, Vulpavus profectus, AMNH 12626; B, Miacis parvivorus, USNM 214706, reversed from the right side; C, "Hesperocyon" wilsoni, TMM 40504-126, reversed from the right side, the internal structure of the posterior half of bulla not prepared, dashed 1 994 WANG AND TEDFORD: CANIDS AND MIACIDS 9

dicating a possible branch toward the fenestra ofthe epitympanic recess as in leptictids (No- vestibuli. We thus tentatively assume that vacek, 1986). The other exits anteriorly there is a stapedial branch ofthe internal ca- through a canal lateral to the piriform fenes- rotid artery because of its common associa- tra posterior and dorsal to the Glaserian fis- tion with a promontory artery in a number sure as in erinaceid insectivorans and tree of insectivorans, primates, and other primi- shrews (Cartmill and MacPhee, 1980). We tive eutherians (MacPhee, 1981; Novacek, tentatively choose the former alternative be- 1980, 1986), as well as in primitive carni- cause that seems the case in Miacisparvivorus vorans, e.g., Miacis parvivorus (see descrip- (see below). tion below), Viverravus minutus (Matthew, 1909), and Protictis schaffi (Gingerich and Winkler, 1985). Furthermore, the presence of foramina near the epitympanic recess and Miacis parvivorus Cope piriform fenestra, as well as a Glaserian fis- Figures IB, 2B sure, also suggests that the superior and in- SPECIMEN EXAMINED: USNM 214706, par- ferior rami of the stapedial artery were pres- tial skull and mandible with fragments ofthe ent, although the Glaserian fissure could be postcranium, from Pinnacle Rock, unit B, solely occupied by the chorda tympani nerve Bridger Formation, Bridger Basin, Wyoming, as in living dogs (Evans and Christensen, early Bridgerian. 1979). The genotypic species, M. parvivorus, was The internal carotid artery (fig. 2A) pre- founded on a jaw fragment with a single m2 sumably first enters the tympanic cavity in (AMNH 5019). The present description is the vicinity ofthe posterior lacerate foramen, based on USNM 214706, which has a well- as in insectivorans (MacPhee, 1981; Nova- preserved right ear region. USNM 214706 cek, 1986). Near the fenestra cochleae, the was briefly mentioned by Gustafson (1986) internal carotid artery gives rise to two in connection with his description of "M." branches, the promontory artery and the sta- cognitus. The specimen has since been further pedial artery. After following the groove on prepared, revealing more detail in the area of the promontorium, the promontory artery tegmen tympani. A complete description of enters the brain cavity through the anterior its basicranium presented below. carotid foramen (middle lacerate foramen, BULLA: Like Vulpavus, M. parvivorus does here not fully separate from the piriform fe- not preserve an ossified bulla. Neither is there nestra). After penetrating the obturator fo- any indication of strong bulla attachment on ramen ofthe stapes, the stapedial artery prob- the petrosal, basioccipital, basisphenoid, ali- ably split into two branches. The ramus sphenoid, or squamosal. inferior abruptly turns anteriorly and exits PETROSAL AND SURROUNDING BONES: The the tympanic cavity through the Glaserian basioccipital is still essentially a platelike bone fissure. There are, however, two possible as in Vulpavus; no thickening of its lateral routes for the superior ramus ofthe stapedial edge is observed. This thin-edged plate lacks artery. One enters the brain cavity laterally a longitudinal groove for the inferior petrosal through a small canal on the posterior aspect sinus as in canids. The lateral edge of the

line indicating outline of bulla (as also in D); D, Hesperocyon gregarius, F:AM 76163. All specimens are drawn to approximately the same size. Actual arterial positions may vary from the drawings except in those segments where a bony impression is available, i.e., sulci on promontorium, Glaserian fissure, and the foramen for superior ramus of stapedial artery. Cross-sectional views are diagrammatic, and their locations of sections are indicated by short horizontal arrows to the left of midline of basioccipital and to the right oflateral border ofsquamosal. Diagonal hatch lines indicate areas offree spaces between bones or missing bones. In ventral views, top is anterior and right is lateral; in cross sectional views, top is ventral and right is lateral. See Definition section for abbreviations. As in fig. 1, the piriform fenestra (pf) encompasses the middle lacerate foramen in Vulpavus and Miacis. 10 AMERICAN MUSEUM NOVITATES NO. 3092 basioccipital and basisphenoid is straight, small and less inflated than in Hesperocyon. lacking the constriction at the juncture ofba- There is no suprameatal fossa. The part of sioccipital/basisphenoid seen in Vulpavus. the basioccipital (mostly paroccipital process The petrosal of M. parvivorus is primitive and the posterior rim of the middle ear re- as in Vulpavus (figs. iB, 2B). The promon- gion) behind the petrosal is missing along the torium is more medially extended toward the sutural contact of the two bones, leaving the lateral edge ofbasioccipital and basisphenoid posterior face of the petrosal mostly intact. compared to that of the latter. In fact, the In this unobstructed view of the posterior medial apex of the promontorium touches promontorium and at the horizontal level of the lateral border of the basioccipital. The the inferior petrosal sinus, a small opening rest of the promontorium, however, is still for the cochlear canaliculus is clearly seen free of any contact with the surrounding facing posteromedially; this canal contains bones. On the steep, nearly vertical medial the perilymphatic duct connecting the coch- face of the promontorium, there is a con- lea and semicircular canals in Canis (Evans spicuous groove slightly dorsal to the level of and Christensen, 1979: 1070). the basioccipital. This groove is inferred to The degree of ossification of the tegmen have contained the inferior petrosal sinus. tympani is intermediate between that of Vul- The surface texture of this groove is rather pavus and Hesperocyon. Anteriorly, the pe- uneven, possibly mirroring the irregular shape trosal ofM. parvivorus is in complete contact of the vein. As in Vulpavus, the large size of with the tympanic process of squamosal lat- the groove for the inferior petrosal sinus and eral to the anterior tip of the canal for the the wide space between the promontorium facial nerve. Medial to the canal, however, and the basioccipital/basisphenoid suggest a there is still a large gap, the piriform fenestra, relatively large vein. between the petrosal and alisphenoid. The The promontorium is slightly less globular tegmen tympani floor is well-ossified so that than in Vulpavus, especially in its less inflated the canal for facial nerve is mostly embedded anterolateral corner. The ventral surface of in bone instead of being in a groove on the the promontorium is covered with irregular tegmen tympani as in Vulpavus. On the other low ridges. Lateral to these ridges is a shallow hand, the ventral side ofthe facial nerve canal but clear groove for the promontory artery. is covered by bone only at the anteriormost As in Vulpavus, the groove begins just an- tip (the segment of the canal for the palatine terior to the fenestra cochleae and extends ramus of the facial nerve, or major petrosal anteriorly and medially toward the large open nerve), and the posterior part of the canal space for the combined anterior carotid fo- (containing the hyomandibular ramus of the ramen (middle lacerate foramen) and piri- facial nerve) remains open, in contrast to the form fenestra. A transverse sulcus branches complete bony coverage in "M. " cognitus and off the postenror end of the sulcus for the some Hesperocyon. The greater ossification promontory artery. This sulcus, which must of the tegmen tympani in M. parvivorus also have been occupied by the stapedial artery, helps define a rounded fossa for the tensor ends at the ventral rim of the fenestra ves- tympani muscle. The site for the genicular tibuli. ganglion can be seen through the open facial Immediately posterior to the fenestra coch- nerve canal along the lateral border of pro- leae on the posterolateral aspect of the promontorium. montorium, there is a rounded process with The epitympanic recess is slightly less ex- a flat top, presumably attached to the tym- cavated than in Vulpavus. As in Vulpavus, panohyoid cartilage. Lateral to this process however, the recess is laterally walled by the is a deep, rounded fossa for the stapedius squamosal and roofed by the in-bent lateral muscle, partially covered by the overhanging margin of petrosal. On the posterior aspect mastoid tubercle. The main body ofthe mas- of the epitympanic recess, there is a canal toid tubercle is made up of the petrosal, ex- through the suture between the petrosal and cept for a small contribution from the squa- squamosal. The opening of this canal faces mosal at the base. The mastoid process is directly opposite the fenestra vestibuli, and 1 994 WANG AND TEDFORD: CANIDS AND MIACIDS I1I its diameter is approximately the same as that The dentition of AMNH 129284 compares of the latter. This combination of geometric favorably with two referred specimens features suggests that this canal forms the exit (AMNH 13071 and 13073) from the Bridger ofthe superior ramus of the stapedial artery. Formation referred to "Miacis" sylvestris by In contrast to the more slitlike Glaserian Matthew (1909: fig. 15). Marsh's (1872) type fissure of Vulpavus, that of M. parvivorus is ofHarpalodon sylvestris (also from the Bridg- a broad trough with a gently curving floor. er Formation, Wortman, 1901: fig. 38) con- Immediately medial to the Glaserian fissure sists of a left ramal fragment with the p3-4 is the essentially anteroposteriorly oriented and the talonid of ml. The only diagnostic alisphenoid/squamosal suture. The post- character ofthe holotype is the rather narrow glenoid process is mostly broken off; the part talonid on ml (as opposed to the wider, more that remains indicates a gentle crest rather basined talonid in M. parvivorus), and with than a sharp ridge as in most carnivorans. this AMNH 129284 agrees. INTERNAL CAROTID ARTERY (fig. 2B): The BULLA: No bulla element is preserved in course ofthis artery is nearly identical to that AMNH 129284. A slightly elevated area along in Vulpavus. As in Vulpavus, the internal ca- the suture between the squamosal and alis- rotid artery, after entering into the middle phenoid, medial to the postglenoid process, ear cavity, branches into a promontory artery suggests the site ofattachment ofthe anterior and a stapedial artery. The promontory ar- crus of the ectotympanic. tery makes an arc along the contour of the However, on the ventral surface of the pe- promontorium and enters the brain cavity trosal is a broad roughened area, approxi- through the anterior carotid foramen (here mately 2 mm wide, along the medial side of continuous with the piriform fenestra). The the promontorium (spe offig. 3). This rugose stapedial artery branches into superior and band is laterally delineated by a distinct bony inferior rami after passing through the ob- ridge, and lateral to this ridge, the ventral turator foramen of the stapes. The course of promontory surface is smooth. Such a su- the superior ramus of the stapedial artery is turelike surface texture along the medial pro- indicated by the presence ofa canal posterior montorium is invariably left by the impres- to the epitympanic recess, i.e., a straight lat- sion of the in-bent medial edge of the bulla eral exit into the brain cavity before turning in caniform carnivorans. The rather ventrally forward. The inferior ramus leaves the mid- extended attachment area ofthe anterior crus dle ear region via the Glaserian fissure. of the ectotympanic suggests that the ectotympanic ring is in a rather erect position (large angle with the horizontal plane), and "Miacis" sylvestris (Marsh) such an erect ectotympanic is often the result Figure 3 of inflation of the bulla due to an expansion SPECIMEN EXAMINED: AMNH 129284, par- ofthe caudal entotympanic. We thus assume tial skull and mandible with left P2-M3, right that the rugose surface on the promontorium P2-M2, left p3-ml m3, right p3-m3, and indicates the presence ofthe caudal entotym- fragments of postcranium, from south of panic. Wamsutter, northeast Washakie Basin, Wy- It is puzzling that such a large area for bulla oming, Adobe Town Member of Washakie attachment on the promontorium is not ac- Formation, Bridgerian. companied by the preservation of the bulla. This specimen was recently donated to the The lack of any bullar elements in miacids is AMNH by Mr. Joel H. Curran of Jackson, presumably due to lack of ossification with Wyoming. Except for the slightly collapsed the surrounding bones. In AMNH 129284, appearance of the basisphenoid region, the the entotympanic attachment to the petrosal right basicranium of AMNH 129284 is es- is certainly more extensive than in Hesper- sentially intact; the left basicranium is miss- ocyon, whose caudal entotympanic has only ing. A complete description of the cranium a narrow contact with the petrosal (see de- and dentition is the subject of a separate re- scription below). port; only the basicranium is described here. PETROSAL AND SURROUNDING BONES: Al- AMERICAN MUSEUM NOVITATES NO. 3092

Fig. 3. Ventral view (in stereophoto and line drawing) of right basicranial region of Miacis sylvestris (AMNH 129284). Top is anterior and right is lateral. See text for descriptions and Definition section for abbreviations. 1 994 WANG AND TEDFORD: CANIDS AND MIACIDS 13 though it lacks a securely attached bulla, the Anterior to the petrosaValisphenoid suture basicranium of AMNH 129284 is advanced is a deep, oval fossa in the tympanic wing of in many aspects. The lateral edge of the ba- alisphenoid. This fossa is variably developed sioccipital is fully exposed on the left side in Hesperocyon (e.g., F:AM 76163 and and exhibits a small, dorsolaterally extended 129176) and is often surrounded anteriorly flange to enclose the inferior petrosal sinus. by the dorsal edge of the ectotympanic. The This bony flange, more extensively devel- alisphenoid/squamosal suture lies medial to oped in its posterior segment near the pos- the Glaserian fissure as is common in the terior lacerate foramen, is in full contact with other taxa examined. The fissure continues the medial face ofthe petrosal. The cross sec- posteriorly by a distinct groove, directed to- tional area of the inferior petrosal sinus is ward the epitympanic recess; the groove small and probably did not house a double- probably contained the inferior ramus of the looped internal carotid artery as in ursids and stapedial artery. The postglenoid foramen is presumably amphicyonids. large. There is a shallow depression just an- Contact between the petrosal and the other terior to the mastoid process in the postero- surrounding bones is similarly extensive, al- dorsal wall of the external auditory meatus, though the sutures are still wide, i.e., the pe- which appears to be an incipient suprameatal trosal is no longer isolated from the surround- fossa. The mastoid process is composed of ing bones as in Vulpavus, and to a lesser ex- the petrosal and squamosal, and is moderate tent in Miacis parvivorus. In addition to the in size. The tip of the paroccipital process is broad rugose area described above, the pe- broken; the remaining part suggests the pres- trosal is more flattened than in other miacids ence ofa distinct ventral ridge, which is vari- described above, in which it is more globular. ably developed in many early caniforms. On the ventrolateral face of the flattened pe- INTERNAL CAROTID ARTERY: As described trosal surface is a shallow but distinct groove above, the internal carotid artery is medial indicating the presence of a promontory ar- to the presumed contact ofthe entotympanic tery. For its entire course, the groove is lateral and petrosal, i.e., retaining the primitive pat- to the presumed contact zone (rugose area) tern of the transpromontorial (intrabullar) of the entotympanic. The sulcus of the prox- position. However, AMNH 129284 suggests imal stapedial artery is less distinct and that the artery had an anterior loop as in all marked by a groove only near the ventral rim canids and many primitive arctoids. Instead of the fenestra vestibuli. Aside from the of directly entering the brain cavity through grooves left by arteries, the surface of the the anterior carotid foramen (middle lacerate promontorium lateral to the rugose band is foramen) as in some miacids, the promontory rather smooth. Immediately behind the fe- artery passes forward of the middle lacerate nestra cochleae is a prominent process, which foramen, into a pit (cica of figs. 3 and 6) in is more distinct than in M. parvivorus, and, the basisphenoid 2 mm anterior to the mid- together with ,a columnlike mastoid tubercle dle lacerate foramen. It presumably loops (composed of the squamosal), forms the at- backward within this pit, and finally turns tachment ofthe cartilaginous tympanohyoid. dorsally to enter the brain cavity through the The tegmen tympani area is well-ossified. middle lacerate foramen. This looped ar- The facial nerve is enclosed by bone ven- rangement is further indicated by the rather trally, and the fossa for the tensor tympani shallow sulcus for the promontory artery near muscle is well-formed. The piriform fenestra its anterior tip. A direct entry into the brain is closed and only a small promontory fo- cavity by the artery would have left a much ramen is left. Lateral to the tegmen tympani, deeper impression on the anterior edge ofthe the epitympanic recess is shallower than in promontorium leading into the middle lac- Vulpavus and is not excavated laterally into erate foramen as seen in Vulpavus and M. the squamosal wall. A small canal appears to parvivorus. be present between the petrosal/squamosal The presence of a groove on the ventral suture in the posterodorsal corner of the re- rim of the fenestra vestibuli suggests a prox- cess for the transmission of the superior ra- imal stapedial artery. The course ofthe artery mus ofstapedial artery. This suture along the is identical to that in Miacis parvivorus. After epitympanic recess is otherwise tightly closed. penetrating the obturator foramen ofthe sta- 14 AMERICAN MUSEUM NOVITATES NO. 3092 pes, the proximal stapedial artery branches established by the presence of the embay- into two rami: the superior ramus runs lat- ment on the basisphenoid, the unambiguous erally and enters the brain cavity through a turn-around point of the artery. There are no canal in the squamosaVpetrosal suture near grooves on the promontorium for branches the posterior end of the epitympanic recess; of the internal carotid artery. the inferior ramus turns forward and exits the middle-ear cavity through the Glaserian "Miacis" gracilis Clark fissure. SPECIMEN EXAMINED: CMNH 11900, holotype, crushed skull and mandible with most "Miacis" cognitus ofthe limb bones. CMNH 12063, partial skull Gustafson and mandible with most of the upper teeth SPECIMEN EXAMINED: TMM 40209-200, except incisors and less well-preserved lower holotype, complete skull, from upper Cham- teeth; an epoxy cast ofthe dentition is avail- bers Formation, Reeves Bone Bed locality, able for examination. Both specimens are Little Egypt Local Fauna, Presidio County, from Horizon C of Uinta Formation, Uinta Texas, medial Chadronian. County, Utah, late Uintan. In the following description, we will outline The following brief discussion of the bas- only features relevant to the phylogenetic dis- icranium was based on a stereophotograph cussion; other features have been adequately of the type. described and figured by Gustafson (1986). BULLA: Clark (1939: 352) described the ho- BULLA: No ossified bulla has been de- lotype (CMNH 1 1900) of "M." gracilis as scribed in "M." cognitus, although an ecto- having "a large, extraneous piece ofbone ad- tympanic ring was reconstructed by Gustaf- hering to the right alisphenoid just anterior son (1986: fig. 24B). Medial to the left to the inner part ofthe glenoid fossa," which postglenoid process, there is a low platform he interpreted as "a portion ofthe rim of the of bone on the tympanic process of squa- tympanic." This he cited as an important mosal; this bone is clearly part of the squa- characteristic linking miacids, without an os- mosal rather than a floating piece ofectotym- sified bulla, and Hesperocyon, with its fully panic. ossified entotympanic. Clark's observation PETROSAL AND SURROUNDING BoNEs: Aside cannot be confirmed in our photograph. from its lack of an ossified bulla, the overall However, a distinct band of rough area on structure of the basicranium is more ad- the medial edge of the petrosal is present, vanced than that of the miacids described indicating the attachment area of the ento- above. The promontorium is flattened and tympanic, as also occurs in "M. " sylvestris. considerably expanded so that it is in contact PETROSAL AND SURROUNDING BONES: The with the basioccipital, basisphenoid, alis- promontorium is generally flat on its ventral phenoid, and squamosal, although there is surface and it is in contact with the lateral still a large, triangular space at the antero- border ofthe basioccipital. The tegmen tym- medial corner of the petrosal that forms the pani area appears to be well-ossified. A shal- middle lacerate foramen. The tegmen tym- low suprameatal fossa may have been present pani is completely ossified and the facial nerve on the anterior face of the mastoid process. is enclosed by bone which forms the floor of INTERNAL CAROTID ARTERY: Presence of a the fossa for the tensor tympani muscle. An- transpromontorial artery cannot be ascer- terior and medial to the middle lacerate fo- tained from the available photographs. A pit ramen, there is a clear impression ofa looped in the basisphenoid for the turn-around of internal carotid artery on the thickened lat- the internal carotid artery, however, can be eral wall of basisphenoid. No suprameatal observed although the pit is partially ob- fossa is present. The mastoid tubercle is scured by the lateral edge ofthe basisphenoid formed from petrosal. in the photograph. INTERNAL CAROTID ARTERY: A medially COMMENTS: In addition to his single cranial positioned internal carotid artery is firmly character (bulla ossification), the remaining 1 994 WANG AND TEDFORD: CANIDS AND MIACIDS 15

cica

--.A. w M pr * r.", .'. .II I.,.

re Fig. 4. Internal view (bulla partially removed) of the anterior part of right middle ear region of "H. " wilsoni, TMM 40504-126, showing the rostral entotympanic, the internal carotid canal, and the promontory foramen. Anterior is toward the left and lateral is toward the bottom of the illustration. The breakage line on the right side corresponds to a natural break of the middle ear region (see fig. 2C, ventral view) and thus only the anteromedial corner of the promontorium is shown in the figure. See Definition section for abbreviations.

"cynodictoid" (canid) characteristics (five) of netic losses, particularly on the left side, the "M. " gracilis cited by Clark (1939: 360) are overall shape of the bulla is still intact be- entirely postcranial, and are mainly related cause of the well-preserved endocast. The to its slender, straight limb bones that con- bulla wall is generally thinner than those in trast with the more robust limbs ofmost mia- H. gregarius and some portions of the bulla cids. The postcranial adaptation toward in- are nearly transparent. The bulla in "H. " wil- creased cursoriality is one of the important soni has a canidlike shape, although the su- transformations from miacids to primitive ture between ecto- and entotympanic is not canids (Tedford, 1978; Wang, 1993). The readily discernible externally. The bulla is poorly known postcrania of miacids, how- highly inflated (presumably mainly by the en- ever, are of limited use in a detailed phylo- totympanic) and anteroposteriorly elongated. genetic analysis. The anterior apex of the bulla extends anterior to the posterior wall ofthe glenoid fossa. To accommodate the anterior expansion of "Hesperocyon" wilsoni Gustafson the bulla, the external opening ofthe foramen Figures 2C, 4 ovale is shifted forward toward the alisphe- SPECIMEN EXAMINED: TMM 40504-126, noid canal. There is no trace of a bony ex- holotype, partial skull and mandible, from ternal auditory meatus. Airstrip Local Fauna, Capote Mountain Tuff, There is a diagonal break across the middle in southern part ofthe Sierra Vieja area, Pre- section of the right bulla, affording an op- sidio Co., Texas (Wilson et al., 1968), early portunity to examine the cross section of the Chadronian. bulla wall. No trace of a low septum can be Gustafson's (1986) original description detected on either side of the break, which dealt with only the external morphology of crosses the region where a low septum is found the bulla. Further preparation during the in H. gregarius. Further preparation poste- present study reveals details inside the bulla riorly did not reveal a septum, although a described below. significant amount of matrix was left intact BULLA: "H." wilsoni is the most primitive for fear of damage to the extremely delicate canid with a fully ossified bulla. Although bulla wall. In the anterior half, no septum is areas of the bulla wall have suffered diage- present at the junction between rostral (see 16 AMERICAN MUSEUM NOVITATES NO. 3092

Fig. 5. Ventral views (in stereo) of dissected left basicranial region in A, "Mesocyon "josephi (MCZ 2102) and B, Enhydrocyon pahinsintewakpa (CMNH 13588), showing developments of bullar septa. Arrows in the line drawings indicate the extent of septum. Top is anterior and right is lateral. See Definition section for abbreviations. 1 994 WANG AND TEDFORD: CANIDS AND MIACIDS 17

panic: the rostral and caudal entotympanics (Hunt, 1974). The rostral entotympanic in living Canis, which ossifies only after birth (Wible, 1984), is usually a small triangular bone at the anteromedial corner of the bulla that forms the roof of the internal carotid artery (Hunt, 1974: pl. 7). In the anterior half of the right bulla of "H." wilsoni, there is a slender bone that is nearly continuous with the in-bent edge ofthe caudal entotympanic, immediately anterior to the promontory foramen and along the lateral edge of the internal carotid canal (fig. 4). This small bone is here interpreted as a rostral entotympanic. The contact zone between the rostral and caudal entotympanic is a narrow area ofoverlap, slightly wider than the thickness of the individual bones. The rostral entotympanic is covered slightly on its ventrolateral edge by the caudal entotympanic. Although not fused, the rostral/caudal entotympanic suture is not filled with dark-colored matrix as occurs between more loosely attached bones, and this suture is best seen near the posterior segment ofthe rostral entotympanic. The topographic position ofthe rostral entotympanic is rather typical ofcanids as seen in living Canis (Hunt, 1974), i.e., the internal carotid artery lies near the junction of the caudal and rostral entotympanic. Near the posterior lacerate foramen, the opening for the internal carotid artery is defined by a small groove on the external surface of the entotympanic. More anteriorly, near the petrosal/basisphenoid suture, the course of the internal carotid artery is recognized as a canal beneath the in- bent edge of the caudal entotympanic. PETROSAL AND SURROUNDING BoNEs: Much ofthe relevant information is obtained from the cross section afforded by the breakage of the right auditory region, and can be partially verified by the exposed anterior half of the bulla. The lateral border of the basioccipital in "H." wilsoni is thick and deep, instead of being thin-edged as in Vulpavus and M. par- Fig. 5. Continued. vivorus. This lateral edge is divided into dorsal and ventral lips, thus enclosing a groove below) and caudal entotympanic. We thus that housed the inferior petrosal sinus. The infer that "H." wilsoni lacks a septum. thickened basioccipital is filled with a small All living carnivorans have at least two pneumatic sinus isolated from the inferior separate ossification centers for the entotym- petrosal sinus by a thin bony wall (fig. 2C). 18 AMERICAN MUSEUM NOVITATES NO. 3092

alisphenoid and squamosal. The piriform fenestra is fully closed between these bones, leaving only a small promontorium foramen at the petrosal/alisphenoid suture directly above the internal carotid canal (fig. 4). Another derived character of "H." wilsoni is a small suprameatal fossa in front of the mastoid process. It is a shallow, rounded pit on the squamosal halfofthe mastoid process, just like that of H. gregarius. The mastoid tubercle is broken, but judging from the remaining basal portion, it was composed of the petrosal as in H. gregarius. INTERNAL CAROTID ARTERY (figs. 2C, 4): pgf fv sf Although slightly more laterally positioned than in H. gregarius, particularly in the an- Fig. 6. Ventrolateral view of the anterior part terior segment, the internal carotid artery in of right middle ear region of H. gregarius, F:AM "H." wilsoni is in a truly medial, extrabullar 76160. All bullar elements are removed, showing the impression of a loop of the internal carotid position. The internal carotid canal is floored artery on the basisphenoid. Anterior is toward the by the dorsal flange of the caudal entotym- left side and lateral is toward the down side. See panic posteriorly and by the caudal and ros- Definition section for abbreviations. tral entotympanic anteriorly (thus extrabullar in topography). It is roofed by the medially expanded petrosal. Entry into the internal ca- The promontorium of "H. " wilsoni is can- rotid canal is marked by a small indentation idlike in its low and flattened appearance, as in the caudal entotympanic immediately be- opposed to the primitive globular promon- low (outside) the posterior lacerate foramen torium of miacids. The promontorium has as in H. gregarius (see below). After entering expanded medially and its medial edge reach- the internal carotid canal, the artery travels es to the lateral border of the basioccipital. forward following the contour of the ventral However, the medial edge ofthe petrosal does surface of petrosal. As in "M. " cognitus and not appear to fully surround the lateral side H. gregarius, it probably makes a U-turn ofthe inferior petrosal sinus. Instead, the en- within the bony confinement ofbasisphenoid totympanic forms a thin strip of the ventro- and caudal entotympanic (not exposed in this lateral wall ofthe sinus. (TMM 40504-126 is area in TMM 40504-126) and finally enters slightly distorted in this area and there is a the brain cavity through the middle lacerate possibility that the original relationship ofthe foramen. bones is obscured.) As in the lateral edge of basioccipital, the medial extension ofthe petrosal is also pneumatic with a small sinus inside the medial Hesperocyon gregarius (Cope) just edge. 2D, 6 Due to its medial expansion, the ventral Figures IC, surface of petrosal in "H." wilsoni is more SPECIMENS EXAMINED: AMNH 39442, skull flattened than in Vulpavus and M. parvivorus. without bullae, from south of Scenic, Pen- In cross section, there is an oval space (some- nington County, South Dakota, Brule For- what more compressed in the anterior seg- mation, Orellan; F:AM 50345, complete skull ment due to crushing) between the overlap- and mandible, left bulla removed, from Quinn ping the entotympanic and petrosal. This Draw, Washington County, South Dakota, space is for the medially positioned internal near top of lower nodular zone, Brule For- carotid artery (fig. 2C). mation, Orellan; F:AM 63942, skull, right The petrosal is also anteriorly expanded bulla dissected, from middle fork of Lone and in contact with the tympanic process of Tree Gulch, Natrona County, Wyoming, 10 1994 WANG AND TEDFORD: CANIDS AND MIACIDS 19

ft below ash F, White River Formation, encompasses the entire medial halfofthe bul- Chadronian; F:AM 76160, basicranial part la. A Chadronian individual (F:AM 63942) of skull without bullae, from Morton Ranch shows an appreciably smaller bulla than those highway area, 6/2 mi. southeast of Douglas, in the Orellan sample. Its caudal entotym- Converse County, Wyoming, middle nodular panic, whose lateral edge can be observed to zone, Brule Formation, Orellan; F:AM 76163, terminate in a low septum (see below), is less partial skull, left bulla dissected, from W. R. expanded on the medioposterior aspect ofthe Silver Ranch, Converse County, Wyoming, bulla, and consequently its ectotympanic ring middle part of a nodular zone, Brule For- is inclined toward the medial side. No defi- mation, Orellan; F:AM 129176, endocast plus nite trend of increased inflation of the bulla basicranium without bullae, from west ofthe through time can be established because of Anthill, Niobrara County, Wyoming, 40 ft the relatively small Chadronian sample of above the purplish white layer, Brule For- Hesperocyon compared to that from the Or- mation, Orellan; FMNH UC495, complete ellan. skull and mandible, left bulla dissected, from In F:AM 76163, a triangular rostral ento- Cedar Creek, Sioux County, Nebraska, Brule tympanic, similar to that in "H." wilsoni, is Formation, Orellan. visible on the dorsal side of the internal ca- Various aspects of the skull, dental, and rotid canal (figs. 1C, 2D). This rostral ento- postcranial morphology ofHesperocyon have tympanic is slightly posterior of the anterior been well-known from descriptions in Scott end of the bulla, and straddles the petro- (1898), Matthew (1901), Scott and Jepsen saValisphenoid suture. Anterior to the rostral (1936), and Wang (1990, 1993, MS). Abun- entotympanic, there is no bullar element on dant specimens are available for examination the dorsal side of the internal carotid canal, ofnearly every detail ofthe basicranium and in contrast to living Canis, in which the ros- their individual variations. Our knowledge tral entotympanic reaches the anterior end of of the Hesperocyon middle ear region is thus the bulla (Hunt, 1974). Although no suture relatively secure compared to that of other is seen, the rostral entotympanic presumably fossil carnivorans that are known from more joins the caudal entotympanic along the in- limited materials. Existing descriptions ofits ternal carotid canal, as in "H. " wilsoni (fig. basicranium (e.g., Hough, 1948; Mitchell and 4). A similarly shaped, narrow rostral ento- Tedford, 1973), however, are inadequate for tympanic can be seen in FMNH UC495 and the present comparisons, and a detailed de- F:AM 63942, and is similarly positioned scription is thus necessary to identify the rel- across the petrosaValisphenoid suture. In evant features. F:AM 63942, the internal carotid canal is BULLA: The external morphology of the broken near the middle of the rostral ento- bulla of H. gregarius is little different from tympanic region, affording a limited view of that of "H." wilsoni other than the slightly the relationship between rostral and caudal greater inflation of the latter. No bony exter- entotympanic; the ventral portion ofthe bony nal auditory mneatus is present. The bulla is canal (presumably mostly made ofcaudal en- made up of the ectotympanic, caudal ento- totympanic as in "H." wilsoni) appears con- tympanic, and rostral entotympanic com- tinuous with the more dorsal rostral entotym- ponents as it is in living canids (Hunt, 1974). panic, i.e., the suture is indistinct. In all Fusion between the ecto- and entotympanic, individuals, the rostral entotympanic tapers however, occurred early in the ontogeny of toward the anterior end. Hesperocyon, and inferences about the in- A low, anteromedial septum inside the bul- dividual components must be based on subtle la has long been considered unique to canids external surface features of the bulla (a line (Flower, 1869). The presence of a septum in of minute pits along the ventral apex), and Hesperocyon was first mentioned by Tedford on a low septum on the internal surface. By (1976). In the present study, a low septum these criteria it can be recognized that the was found in all Hesperocyon specimens in inflation of the bulla is mostly caused by the which the internal surface of the bullae was expansion ofthe caudal entotympanic which prepared (F:AM 50345, 76163 and FMNH 20 AMERICAN MUSEUM NOVITATES NO. 3092

UC495) or in specimens preserved as endo- Hunt stressed the different topographic re- casts of the bullae so that the septum can be lationships of the septa: "A tympanic [ecto- seen as embedded within the endocasts (e.g., tympanic] contribution can be definitely ex- F:AM 63942, a Chadronian individual with cluded in the Canidae in which the septum uninflated entotympanic). The septum lies develops from the edge of the caudal ento- along the ecto-/entotympanic suture, and may tympanic nearest the rostral entotympanic, taper offtoward the anterior end (fig. 1 C). In that is, from the dorsal edge. In felids and FMNH UC495, the septum is rather low and viverrids, the septum bullae develops from frequently interrupted by tiny notches, that part of the caudal entotympanicfarthest whereas in F:AM 50345, it is higher and forms from the rostral entotympanic, that is from a continuous thin blade. the ventral or opposite edge which is in con- Within hesperocyonines, the most primi- tact with the tympanic" (Hunt, 1974: 38-39, tive subfamily ofCanidae, there is a tendency original italic, brackets added). toward increasing the length and depth ofthe Although not easily verified in H. gregar- septum. Thus, in KUVP 85067, a larger ius, the morphology in some fossil canids ap- Hesperocyon, the septum runs the entire pears to agree with part ofHunt's description, length of the ecto-/entotympanic suture in- i.e., the entotympanic forms the septum. Al- stead of only part ofthe length as in some H. though composition ofthe septum cannot be gregarius specimens. Furthermore, in Me- ascertained because of the advanced stage of socyon coryphaeus (AMNH 6859, holotype) fusion, the development of a dorsal segment and "M." josephi (MCZ 2102, holotype of ofthe septum, posterior to the presumed con- M. hortulirosae, fig. 5A), the septum becomes tact with the rostral entotympanic (e.g., in much deeper, and may also develop a sep- MCZ 2102, "Mesocyon" josephi, fig. SA) tumlike low ridge along the in-bent dorsal clearly involves an in-bent margin of caudal edge of the caudal entotympanic (fully ex- entotympanic, i.e., the free edge of caudal posed in MCZ 2102). These two septa, one entotympanic grows toward the empty space formed along the ventral edge of entotym- ofthe middle ear cavity. We thus assume that panic (ecto-/entotympanic suture) and the the ventral septum seen in all early canids is other along its dorsal edge (rostral/caudal en- similarly formed by the same ontogenetic totympanic suture, or a free edge in the pos- process in the expansion ofthe in-bent caudal terior half of caudal entotympanic), join at entotympanic. their anterior tips and form a complete ring Hunt's (1974) claim that the canid septum inside the bulla. In some derived hespero- is formed by the dorsal edge of caudal en- cyonine canids (e.g., Enhydrocyon, CMNH totympanic only, however, is in contradic- 13588, fig. SB); in borophagines: (Tomarctus, tion to the present observations on early can- F:AM 49088, 62300; Aelurodon, F:AM ids. In Hesperocyon, a dorsal septum is absent 61755, 67025; Epicyon, F:AM 61475); and along the rostral and caudal entotympanic in canines (personal observations on repre- suture; instead, a ventral septum is invariably sentatives of all living genera), however, the present between the ectotympanic and the septum is mostly confined to the anterior end caudal entotympanic. In more derived hes- of the bulla, and is at an angle with the par- perocyonines such as "Mesocyon "josephi and asagittal plane. This limited septum sits on Enhydrocyon pahinsintewakpa, a dorsal sep- the anteriormost part of the internal carotid tum begins to develop due to the expansion canal, in parallel or at a small angle with the of the caudal entotympanic, but the ventral canal, and stretches ventrolaterally toward the septum is even more pronounced (fig. 5). It tympanic membrane. appears that the anteriorly limited occur- Hunt (1974) emphasized the nonhomolo- rence ofthe septum in most canids (all extant gous nature of the unilaminated canid sep- canines, some borophagines, and derived tum to the bilaminar, complete septum of members of hesperocyonines) is a derived most living feliform carnivorans. The canid condition from a more ubiquitous presence septum is composed of the entotympanic of the septum along all edges of the caudal alone rather than by both ento- and ectotym- entotympanic in early canids. In fact, the sep- panics as in feliforms. More importantly, tum may not be limited to the area ofcontact 1 994 WANG AND TEDFORD: CANIDS AND MIACIDS 21 between the rostral and caudal entotympanic promontory foramen and the anteromedial in living Canis either. The septum in adult rim of the fossa for the tensor tympani (fig. Canis can be seen to reach down as far as the 6). The occupant of this groove is unknown. ventral apex of the bulla, far exceeding the The hyomandibular ramus ofthe facial nerve range of the rostral entotympanic. is still partially exposed near the posterior PETROSAL AND SURROUNDING BONES: The half of the tensor tympani fossa in some in- basioccipital is essentially the same as in "H. " dividuals (AMNH 39442; F:AM 76160, wilsoni; it has a thickened lateral border sur- 76163) and is fully covered by a thin layer of rounding a small inferior petrosal sinus. The bone in others (FMNH UC495). In the single medial border of petrosal forms a corre- Chadronian specimen (F:AM 63942), the fa- sponding groove to surround the lateral half cial nerve is more exposed than in the Orellan ofthe sinus. The inferior petrosal sinus is thus individuals, following the general tendency completely embedded within a bony tube toward a better ossified tegmen tympani in formed by the basioccipital and petrosal, in later taxa. The course of the nerve is stable contrast to the condition in ursids and am- among carnivorans -passing dorsolaterally phicyonids where the sinus is housed within around the fenestra vestibuli and emerging a large embayment formed in the medial side from the middle ear region via the stylo- of the basioccipital. mastoid foramen. The petrosal is in full contact with the sur- Mitchell and Tedford (1973) noted the rounding bones. Its anterior edge is in smooth presence ofa small suprameatal fossa in Hes- transition with the tympanic processes of al- perocyon. This fossa is only at an incipient isphenoid and squamosal; there is no sudden stage of development, and is far from reach- drop of the tegmen tympani floor from the ing the size or varied locations it does in mus- posterior edge of the alisphenoid as in Vul- teloids (Schmidt-Kittler, 1981). As in M. par- pavus and to a lesser extent in Miacis. The vivorus, the tip of the mastoid tubercle is Eustachian tube and the tendon for the tensor composed of the petrosal rather than the vela palatine muscle enter through the Eusta- squamosal as in Vulpavus. chian canal (musculotubal canal of Evans INTERNAL CAROTID ARTERY (fig. 2D): There and Christensen, 1979). Contrary to the find- is little doubt as to the course of the internal ing in Hough (1948: 94), there is a bony par- carotid artery in Hesperocyon, not only be- tition between the Eustachian tube and the cause of the clear presence of its canal pre- anterior "carotid foramen" (= foramen lac- served by the surrounding entotympanic, erum medium of Hough, 1948), i.e., the as- which is consistently observed in all individ- cending pharyngeal artery is fully enclosed by uals, but also because of the well-known an- the entotympanic (see description below). atomical details in living Canis of the gross The promontorium is flattened rather than anatomy of adults (Evans and Christensen, globular in shape as it is in Vulpavus and 1979), the osteology of neonatal individuals Miacis. In all specimens, the ventral surface (Hunt, 1974), and the ontogenetic transfor- ofthe promontorium is very smooth and de- mations of early embryonic stages (Wible, void of any indication ofartery or nerve im- 1984). pressions. The piriform fenestra is complete- From the internal view of the bulla, the ly closed, leaving only a small promontory position ofthe internal carotid artery is clear- foramen, from which the palatine ramus of ly marked by a prominent bony welt formed facial nerve (major petrosal nerve) emerges by the entotympanic. On specimens (F:AM from within the petrosal (Evans and Chris- 76163, 129176) that are fortuitously broken, tensen, 1979: fig. 15-16). This ramus of the this crest can be seen as a column of matrix facial nerve immediately turns forward and surrounded by the petrosal dorsomedially and leaves the middle ear cavity through the Eu- the entotympanic ventrolaterally. This col- stachian canal. In most specimens, a very thin, umn is the endocast of the internal carotid but distinct groove is left by the nerve on the artery, and can be traced in its entire course alisphenoid floor between the promontory along the medial wall of the bulla. foramen and the Eustachian canal (fig. 6, In most specimens, the posterior entrance smp). There is also a groove between the of the internal carotid canal is at the same 22 AMERICAN MUSEUM NOVITATES NO. 3092 level as, or slightly ventral to, the caudal en- enter the brain cavity through the middle lac- totympanic/basioccipital suture. A small, erate foramen. In contrast, this loop occurs short groove on the external bullar surface is within the contact zone between the entotym- usually seen ventral to the posterior lacerate panic and basisphenoid in Hesperocyon. This foramen. The posterior opening of the inter- is also reflected in the relatively small, an- nal carotid canal is thus visible from outside teriorly restricted "carotid foramen" of Hes- ofthe bulla. This is slightly different from the perocyon, which is much smaller in diameter condition in most canids in which the inter- than the adjacent Eustachian canal. Not only nal carotid artery first enters into a common is it impossible for this small opening to ac- recess for the posterior lacerate foramen and commodate the volume of a doubled artery, the internal carotid canal, then turns anteri- the "bottleneck" will not even permit the orly into the internal carotid canal; therefore, passage ofa single internal carotid artery with no posterior carotid foramen can be seen on its original volume. In comparison, the open- an unbroken specimen. The more ventral ening of the anterior carotid foramen in Canis trance of the internal carotid artery in Hes- is enlarged relative to the thickness ofinternal perocyon may represent the morphotypic carotid canal and is larger than the nearby condition for Caniformia. From this primi- Eustachian canal. Consequently, the thick- tive condition, two directions ofchange may ened lateral wall ofthe basisphenoid in Canis occur. In more derived canids, the posterior is free ofany marks left by the internal carotid entrance of the internal carotid canal moves artery because the artery is not even in con- dorsally into the recess for posterior lacerate tact with the basisphenoid in this region. A foramen. In arctoids, however, the entrance sheet of entotympanic bone separates the ar- moves ventrally and anteriorly and becomes tery from such contact. an opening on the bullar wall anterior to the In Hesperocyon, the short segment of the posterior lacerate foramen. canal formed by the basisphenoid and the After entering the carotid canal, the inter- entotympanic just anterior to the loop of the nal carotid artery follows the contour of the internal carotid artery is the site of union of ventral surface of the petrosal, i.e., turning the internal carotid and the slender ascending dorsally slightly and then maintaining a rath- pharyngeal arteries (fig. 2D). In living Canis er straight course forward. Within the con- the latter branches off the internal carotid finement of the internal carotid canal, the artery outside (anterior to) the bulla (Evans artery turns medially toward the lateral mar- and Christensen, 1979: figs. 1 1-16). This in- gin ofbasisphenoid at a level slightly anterior tracanal condition of the ascending pharyn- to the middle lacerate foramen. It then im- geal artery in Hesperocyon suggests that the mediately turns posteriorly when contacting ascending pharyngeal canal in Hesperocyon the basisphenoid, and finally enters the mid- (as dictated by the enclosed vessel) is con- dle lacerate foramen. This tight U-turn is verted to house the looped internal carotid clearly imprinted on the surrounding basi- artery in Canis. sphenoid, whith exhibits a tightly coiled sul- Wible (1984: 124) documented the disap- cus (fig. 6). Such a landmark is a bony sig- pearance of the stapedial artery early in the nature that is often correlated with a medially ontogeny of Canis familiaris. The proximal positioned internal carotid artery. stapedial artery gives rise to the inferior and The peculiar looped-back internal carotid superior rami after penetrating through the artery occurs in living Canis (e.g., Evans and stapes at the 14.5-mm stage. This vessel is Christensen, 1979: figs. 1 1-16). This loop can reduced in calibre by the 17-mm stage, and now be similarly traced in Hesperocyon. There completely degenerated by the 25-mm stage. is, however, an important distinction be- At this stage, the internal carotid artery between living canids and their archaic rela- comes extrabullar because of the newly tives. The actual looping occurs in an extra- formed ossification center of the caudal en- bullar fashion in Canis, i.e., the internal totympanic lateral to the artery. The adult carotid artery emerges from the anterior tip Hesperocyon has apparently reached this same of the bulla through the anterior carotid fo- stage of middle-ear arterial circulation as in ramen and then turns back immediately to living Canis. 1 994 WANG AND TEDFORD: CANIDS AND MIACIDS 23

Fig. 7. Upper P4-M2 (or M3) ofMiacis and Hesperocyon. A, "M. "gracilis, CMNH 12063; B, "M." cognitus, TMM 40209-200; C, "H." wilsoni, TMM 40504-126; D, H. gregarius, F:AM 63930. Scale = 5 mm.

CHARACTER ANALYSIS trigonids oflower molars clearly indicate spe- cialized tendencies toward hypocamivory Basicranial characters as well as dental (i.e., "cercoleptoid" omnivores of Matthew, characters were incorporated in the following In analysis (character numbers correspond to 1909). most instances, the presumed ple- those in the character matrix in table Po- siomorphic conditions in Vulpavus are con- 1). sistent with such primitive viverravids as larities were determined by outgroup comparison. The primitive feliform viverravids, Protictis. the ofmia- 1. REDUCTION OF PARASTYLE ON P4: Can- commonly considered sistergroup iforms have a reduced parastyle on the P4 in cids, are here used as the primary outgroup. Recent study by Hunt and Tedford (1993), TABLE 1 however, suggests that the caniform-feliform Character Matrix for Selected Miacids and (i.e., miacid-viverravid) dichotomy may be Primitive Canids traced as far back as the Cretaceous Cimo- Numbers of characters correspond to those listed lestes, and considerable independent evolu- under Character Analysis. Question marks rep- tion may thus have occurred within the two resent missing data. See text for discussions of separate lineages by the time of their first polarities. appearances in the fossil record. More prob- 11111111 lematic is the supposed ancestral status of characters: 12345678901234567 creodont to the camivorans, despite its pop- taxa: ularity as a hypothetical ancestor among car- Outgroup 00000000000000000 nivore systematists (Flynn et al., 1988; Woz- Vulpavus 110000000000?1000 encraft, 1989; Wyss and Flynn, 1993). In cases Miacis parvivorus 111111000000?0000 ofpolarity conflicts, we use the Creodonta as "Miacis " sylvestris 112111111000?0000 the next outgroup outside the Camivora. "Miacis" cognitus 112111100010?1??0 Based on these polarity assessments, the bas- "Miacis" gracilis 11?11?1111?0?0001 icranium of Vulpavus is quite close to a prim- "Hesperocyon " wilsoni 12211111111100001 itive morphotype of the Caniformia, even Hesperocyon gregarius 12211111111111100 112111100010?1010 though its quadrate upper molars and the low Daphoenus 24 AMERICAN MUSEUM NOVITATES NO. 3092 contrast to the primitive retention of such a that a globular (Character 9) and isolated parastyle in feliforms (Flynn et al., 1988). (Character 4) promontorium is primitive for Polarity: 1(0), presence ofa well-developed caniform carnivorans, this is not so evident parastyle on P4; 1(1), reduction or absence if feliform miacoids are taken into consid- of the parastyle. eration. For example, the Tiffanian Protictis 2. LINGUAL CINGULAE ON UPPER MOLARS: and Bridgerian Viverravus (both members of Presence of a complete lingual cingulum to the Viverravidae) have flat promontoria, be continuous with the pre- and postcingul- which are medially expanded to be in full um of the upper molars is apomorphous for contact with the basioccipital (Gingerich and the Caniformia (Flynn et al., 1988). Reduc- Winkler, 1985; Matthew, 1909). The primi- tion of anterior segment of the internal cin- tive condition for Carnivora remains to be gulum ofMl and enlargement ofits posterior settled. Among taxa examined here, how- segment, resulting in an asymmetrical ar- ever, an isolated, globular promontorium is rangement of internal cingulum, are charac- primitive. teristic of early hesperocyonine canids. In Polarity: 4(0), promontorium isolated; 4(1), most individuals of Hesperocyon, the ante- promontorium in contact with basioccipital. rior segment is completely lost, leaving a pe- 5. REDUCTION OR Loss OF M3: The nearly culiar, posteriorly swung appearance of the universal absence ofM3 in living carnivorans internal cingulum (fig. 7C, D); the protocone (except in isolated cases of secondary acqui- is consequently exposed anteriorly. sition of three molars, e.g., Otocyon) was a Polarity: 2(0), absence of a complete lin- convenient synapomorphy for Carnivora in gual cingulum; 2(1), presence of a complete Wozencraft's phylogeny (1989). This char- lingual cingulum on upper molars; 2(2), re- acter is similarly used by Wyss and Flynn duction or loss ofanterior segment ofinternal (1993) in support of their hypothesis that cingulum. Miacidae (the contents ofwhich were not ex- 3. CLOSURE OF PIRIFORM FENESTRA AND plicitly stated) lies outside of a Viverravidae OSSIFICATION OF TEGMEN TYMPANI: Primi- + Carnivora clade. Loss ofthe posterior mo- tively the piriform fenestra opens widely and lars is a repeated trend throughout the history the tegmen tympani is poorly ossified, i.e., of the Carnivora, especially among various lacks an ossified floor for the facial nerve. In hypercarnivorous lineages which minimize canids, the piriform fenestra is closed except grinding (molars) and maximize shearing for a small promontory foramen, and the teg- (carnassials) dentitions. Data in fossil sam- men tympani is fully ossified, i.e., the facial ples are often sufficiently dense to show grad- nerve becomes fully enclosed in the tensor ual reduction and loss of the last molars. In tympani region. the case ofcanids, it is likely that the M3 was Polarity: 3(0), wide open space for a piri- lost within the canid clade, as suggested by form fenestra between petrosal and alisphe- the occasional presence ofM3 in Chadronian noid/squamosal, facial nerve exposed ven- H. gregarius (Wang, MS). We did not further trally; 3(1), narrowing ofthe piriform fenestra, divide the reduction of M3 into multiple facial nerve partially embedded within teg- character states as would be desirable to aid men tympani and floored in anteromedial in the resolution of different lineages of mia- segment; (2), complete closure ofthe piriform cids; this will have to wait for more detailed fenestra leaving only a small promontory fo- analysis of the group. ramen, facial nerve beneath a bony sheath Polarity: 5(0), presence ofa relatively large which defines the fossa for tensor tympani M3; 5(1), reduction and loss of M3. muscle. 6. COMPOSITION OF MASTOID TUBERCLE: 4. CONTACT OF PROMONTORIUM AND BAS- Flynn and Galiano (1982: 19) suggested that IOCCIPITAL: Primitively, the promontorium is a mastoid tubercle (= process hyoideus, at- separated from the basioccipital on the me- tachment for the cartilaginous tympano- dial side. The free space between the pro- hyoid) composed of the squamosal may be a montorium and basioccipital is partially synapomorphy for Carnivora. The mastoid closed in Miacis parvivorus and further fused tubercle is composed of the squamosal in in all canids. Although it may be apparent Vulpavus but is made up of the petrosal in 1 994 WANG AND TEDFORD: CANIDS AND MIACIDS 25 the remaining taxa examined in this study sory cusps on P3 or p3; 10(1), presence of (fig. 2). posterior accessory cusp on these premolars. Polarity: 6(0), mastoid tubercle formed by 11. POSITION OF INTERNAL CAROTID squamosal; 6(1), mastoid tubercle formed by ARTERY: Presence ofa promontory artery ly- petrosal. ing along a groove on the ventral surface of 7. ANTERIOR LOOP OF INTERNAL CAROTID the promontorium and a stapedial artery ARTERY: In derived miacids ("Miacis" syl- penetrating the stapes is a widespread con- vestris and "M." cognitus), canids, and am- dition among mammals (e.g., MacPhee, 1981; phicyonids, the internal carotid artery makes Novacek, 1986). Ontogenetic evidence sug- an anterior loop before entering the brain gests that this may be the derived state among cavity through the middle lacerate foramen, eutherians (Wible, 1983). The taxa in the in contrast to a direct entry without this loop present study suggest that the transpromon- as in Vulpavus and Miacis parvivorus. The torial position of the internal carotid artery derived condition is readily recognized in (promontory and stapedial arteries) is prim- fossils by a sigmoid or oval fossa on the tym- itive for caniform carnivorans (e.g., Vulpavus panic wing ofthe basisphenoid. In living Ca- profectus, Miacisparvivorus, "M. " sylvestris). nis, the loop becomes extrabullar in position. This view is further strengthened by similar Polarity: 7(0), lack of an anterior loop of arterial configurations in some living feliform the internal carotid artery; 7(1), presence of families (Hunt, 1977), in the archaic feliform the loop. Protictis schafi (Gingerich and Winkler, 1985) 8. SUPRAMEATAL FoSSA: A small supra- and Viverravus (Matthew, 1909), and possi- meatal fossa is found in H. gregarius and bly in nimravids (Neff, 1983; Bryant, 1991; "H. " wilsoni. The only canid that has a well- but see Hunt, 1987, and Flynn et al., 1988, developed suprameatal fossa is "Cynodes- for an alternative view). This primitive in- mus" cooki Macdonald, which is an extreme- trabullar condition contrasts with the extra- ly aberrant borophagine. A small fossa is bullar state of canids, in which the internal present in the primitive ursid Cephalogale, carotid artery travels within a bony canal be- in addition to its well-known occurrences in tween the caudal entotympanic ventrally and musteloids (Schmidt-Kittler, 1981), but it is the petrosaVrostral entotympanic dorsally lost in the amphicyonids. (Hough, 1948; Hunt, 1974; Tedford, 1976). Polarity: 8(0), absence of a suprameatal Flynn et al. (1988: 77) stated that "all living fossa; 8(1), presence of a small suprameatal carnivorans are derived in the lack of a sta- fossa. pedial artery," because ofthe absence or weak 9. SHAPE OF PROMONTORIUM: In Vulpavus development of a groove leading toward the and Miacis parvivorus, the promontorium is fenestra vestibuli. In caniform carnivorans, globular. From this primitive condition, the the primitive condition is likely the presence canid promontorium becomes flattened; of a proximal stapedial artery (Wang and whereas the amphicyonids appear to have re- Tedford, 1991). In Miacis parvivorus and tained the primitive condition. "M. "sylvestris, not only is there a clear groove Polarity: 9(0), promontorium globular; 9(1), on the promontorium for a stapedial artery, promontorium flattened. there is also a corresponding foramen near 10. POSTERIOR ACCESSORY CUSPS ON PRE- the posterior part of the epitympanic recess MOLARS: In primitive canids (most hespero- for the exit of the superior ramus of the sta- cyonines and borophagines), a posterior ac- pedial artery (the presence of a wide Glas- cessory cusp is present on the P3 and p3 in erian fissure may also indicate the presence addition to the anterior and posterior cin- of a inferior ramus of the stapedial artery). gular cusps. This is in contrast to many mia- These latter structures are found in Vulpavus cids that have a posterior accessory cusp only profectus except for the lack ofa clear groove on the p4 and no accessory cusps present on leading to the fenestra vestibuli. the upper premolars (e.g., M. parvivorus, "M. " Wible (1984: 119) described an ontoge- cognitus). "M." gracilis possessed the ad- netic sequence for the transformation of the vanced condition as in Hesperocyon. cranial artery systems in Canisfamiliaris that Polarity: 10(0), absence ofposterior acces- is similar to the phylogenetic series in the 26 AMERICAN MUSEUM NOVITATES NO. 3092 present study. The internal carotid artery dis- still leave the attachment of an ossified bulla plays an "indifferent position" early in on- to the basicranium (strong enough to be pre- togeny. During late-fetal to neonatal stages, served in fossils) as a derived condition, and the artery shifts to an extrabullar position the known distribution ofthis condition con- because of the appearance of the caudal en- tinues to be a useful character to diagnose totympanic forming the ventral lateral border advanced caniforms. ofthe internal carotid artery. Finally, in post- Polarity: 12(0), absence of an ossified en- natal stages, the artery becomes trapped with- totympanic bulla; 12(1), presence of an os- in the canal formed by the expanding caudal sified entotympanic bulla. and rostral entotympanic. 13. Low ENToTYMPANIC SEPTUM IN BULLA: Polarity: 1 1(0), internal carotid artery lat- Generally considered to be nonhomologous erally positioned, transpromontorial, pres- with the bilaminar, complete septum of fe- ence of a promontory artery and stapedial lids, hyaenids, and viverrids, in which both artery, as well as superior and inferior rami the ectotympanic and caudal entotympanic ofthe stapedial artery; 11(1), internal carotid contribute to the septum (Hunt, 1974); the artery medially positioned, extrabullar, in- low, incomplete septum of canids is here inside a bony canal formed by the caudal en- ferred to derive from the caudal entotym- totympanic and the petrosal/rostral ento- panic only. The low septum tends to be tympanic, stapedial artery lost. formed around all edges of the caudal ento- 12. OSSIFICATION OF ENroTYMPANIC BULLA: tympanic in Hesperocyon and primitive hes- An ossified entotympanic bulla is unknown perocyonines (sensu Wang, MS), but is re- in available miacid skulls. Beginning with stricted to the anteromedial corner ofthe bulla "Hesperocyon" wilsoni, a fully ossified bulla in derived hesperocyonines and in living canis present. Its degree of inflation, which is ids (fig. 5). comparable to that of H. gregarius, suggests Polarity: 13(0), absence of a low septum; significant contribution from the entotym- 13(1), presence of a low septum. panic. In H. gregarius, the ossified bulla 14. SIZE OF PARASTYLE OF M1-M2: Pres- clearly consists ofecto- and entotympanic el- ence ofa long, transversely oriented parastyle ements as confirmed by the presence ofa low on Ml -M2 is primitive for all miacoids (fig. septum inside the bulla. 7A). Accompanying this large parastyle is the Flynn et al. (1988: 100) noted the negative more medially placed paracone so that the aspect of the character "bulla ossification," metacone has a more lateral position relative i.e., miacoids may have had an ossified bulla to the paracone. In Hesperocyon (fig. 7D) and although none are preserved. Recent cladistic more so in later canids, the parastyles of the analysis by Wozencraft (1989) on living fam- upper molars are reduced to a labial cingu- ilies postulated a unique synapomorphy of lum, sometimes swelling slightly and the la- an ossified bulla for all extant carnivorans bial cusps lie in tandem. The condition in except Nandinia. A similar conclusion was "H. " wilsoni is only slightly advanced toward reached by Wyss and Flynn (1993), who rea- the canid type (fig. 7C). On the other hand, soned that the possibility still exists that an "Miacis" cognitus (fig. 7B), Daphoenus, and ossified entotympanic may eventually be other amphicyonids have a very reduced found in miacoids. In their data matrix, how- parastyle and side-by-side paracone and me- ever, Viverravidae and Miacidae are still tacone much like that of Hesperocyon. coded as lacking an ossified bulla, thus lend- Polarity: 14(0), large, ridgelike parastyle of ing support to their phylogeny which ex- MI -M2; 14(1), reduced parastyle. cludes these fossil families from the Carniv- 15. SIZE OF METACONID ON M2: In Hesper- ora. Perhaps more radical is their suggestion ocyon and most early canids the metaconid that the cartilaginous caudal entotympanic in of m2 is nearly equal to or larger than the Nandinia, and the lack of an ossified ento- protoconid, in contrast to the more dominant tympanic in primitive amphicyonids, may be protoconids in all Miacis and in "H. " wilsoni. secondarily derived from a primitively fully Although shown as an autapomorphy under ossified condition. In any case, this would the Phylogeny (below), this character is syn- 1 994 WANG AND TEDFORD: CANIDS AND MIACIDS 27 apomorphous for the Canidae (excluding (Flynn et al., 1988), and some cursorial ad- "H." wilsoni). aptations of the postcranial skeleton in Hes- Polarity: 15(0), protoconid of m2 larger perocyon (Tedford, 1978; Wang, 1990) and than metaconid; 15(1), protoconid subequal in "M. " gracilis (Clark, 1939). to metaconid. 16. INFERIOR PETROSAL SINUs: As dem- onstrated by Hunt (1977), amphicyonids are PHYLOGENY characterized by an enlarged embayment of the basioccipital, presumably to house a One shortest tree (fig. 8) was found by the looped internal carotid artery within the in- HennigG86 program (version 1.5, written by ferior petrosal sinus as a counter current heat J. S. Farris), with a total length of 20 steps, exchange mechanism as seen in living ursids. a consistency index of 80, and a retention Such a large sinus is also found in primitive index of 82 (all calculated after the autapo- ursoids such as hemicyonines and amphi- morphies are removed). The character ma- cynodontines (Tedford et al., in press). trix is listed in table 1 and all characters are Polarity: 16(0), inferior petrosal sinus small; nonadditive. 16(1), inferior petrosal sinus greatly enlarged. That "H. " wilsoni is closest to Hesperocyon 17. RELATIVE SIZE OF METACONE TO is supported by the present analysis (Char- PARACONE OF M2: The M2 metacone is about acters 12 and 2), even though its teeth are the same size as the paracone in Hesperocyon still largely primitive. In addition to the ex- (fig. 7D), "Miacis " cognitus (fig. 7B), and later ternal bullar morphology noted by Gustafson canids, whereas it is very reduced or nearly (1986), the anatomy inside the bulla further absent in "M." gracilis (fig. 7A) and "H." supports this relationship. The lack of a low wilsoni (fig. 7C). septum in "H." wilsoni, however, is inter- Polarity: 17(0), metacone of M2 subequal esting, with its two alternative implications: to paracone; 17(1), metacone much smaller nonpreservation (lack of ossification) of a than paracone. septum in "H." wilsoni, or acquisition of a septum beginning in H. gregarius. The first Some of the basicranial characters de- possibility cannot be ruled out until more and scribed above may be correlated with one better preserved bullae are found. The sec- another. For example, the expansion and in- ond, a canid septum independently derived creased ossification of the petrosal may be from those of feliforms, is a tentative con- linked to the closure ofpiriform fenestra (an- clusion reached in this analysis. terior expansion of the petrosal), the articu- The tree in figure 8 is in general agreement lation of the petrosal with the basioccipital with Clark's (1939) contention of a close af- (lateral expansion of petrosal), and the ossi- finity between "M." gracilis and Hespero- fication of the floor of facial nerve. Until the cyon; support for such a relationship is based basicrania of a larger series of miacids are on the presence ofa posterior accessory cusps known, all of these are treated as individual on the third premolars (Character 10). As characters in the present analysis. Merging of pointed out by Clark (1939), the postcranial these characters does not affect the topology skeleton of "M. " gracilis offers the best evi- of the phylogeny. dence for affinity with Hesperocyon. In- In addition to above characters, some den- creased cursoriality (elongated limbs and oth- tal autapomorphies of H. gregarius and other associated structures) appears to be one of ers useful for discussion within Canidae are the key evolutionary innovations by canids not listed above, because they do not con- (Wang, 1993), that contrast with the rela- tribute to the resolution of phylogeny at this tively stout, plantigrade limbs in most arc- level of analysis. Other potentially useful toids. Within the limits ofcurrent knowledge, characters are excluded from the present "M. " gracilis appears to be close to the base analysis primarily because of the paucity of of the canid clade. materials. These include some soft anatomy "Miacis " sylvestris is most interesting in its or postcranial characters for the Caniformia transitional characteristics of the basicran- 28 AMERICAN MUSEUM NOVITATES NO. 3092

Cynoidea Arctoidea a character most often associated with a me- 1 I I dial position of the artery in advanced can- Canidae iforms. "M." sylvestris is here linked to the 1 cynoid clade by its flattened promontorium (Character 9) and presence of a suprameatal .tb Id) .Da z fossa (Character 8). We included the amphicyonids in the anal- ,,'.a ysis to indicate a possible position ofthe arc- I.'.. n 0 toids within the present phylogenetic frame- r- V) work. We assumed that amphicyonids (here Daphoenus) represent a close approximation to the primitive arctoid morphotype as sug- 0. . I',1N 0 gested in recent studies by Flynn et al. (1988) and Tedford et al. (in press). Basicranially, the amphicyonids retain the primitive conditions of a rather globular promontorium and the lack of a bulla. The position of the internal carotid artery is not certain; the lack of a transpromontorial groove on the promontorium suggests a medial position of the artery (in advanced amphicyonids which have _ N ossified entotympanics, the artery is trapped in a bony tube within the entotympanic). In LO0*CD a addition, the amphicyonids possess an enlarged basioccipital embayment (Character 16) for a presumed "ursid loop" of the internal carotid artery within the inferior pe- NX trosal sinus, a unique arterial cooling system _-0 _ found in living ursids (Hunt, 1977). Whether or not this character is present in "Miacis" cognitus will be a further test of the present Fig. 8. A shortest tree found by the Hennig86 phylogeny which links it to the amphicyon- program (version 1.5) based on a 9 taxa by 18 ids. Dentally, primitive amphicyonids (e.g., character matrix in table 1. Tree statiistics are (after Daphoenus) are little different from primitive removing the autapomorphies): treie length = 20 canids (e.g., Hesperocyon); the fact that mem- steps, consistency index = 80, retei idex = bers ofthe Amphicyonidae were consistently 82. The program settings are: all ( htion included in the Family Canidae (e.g., Mat- nonadditive to avoid preconceived haracterss aes of character transformations with nr sultiple states thew, 1930; Simpson, 1945) further testifies calculations of trees are in the ie (implicit enu- to this primitive resemblance. However, one meration) option, a search procediure certain to dental character presently recognized as a find all trees of minimal length. So]lid bars repre- synapomorphy of the amphicyonid/arctoid sent unique synapomorphies; open bars represent clade (i.e., reduced parastyles of upper mo- homoplasies. Character numbers above the bars lars, Character 14) was independently ac- correspond to those listed in the Clharacter Anal- quired by canids. ysis section and table 1; numbers b)elow the bars With regard to the broader question ofcan- represent character states. iform phylogeny, the presence of a sulcus on the basisphenoid that is indicative of an anterior loop of the internal carotid artery ium and its rather primitive teetth. Although (Character 7) seems to be a synapomorphy still retaining a lateral (intrabulllar) position for the common ancestors of the Cynoidea of the internal carotid artery, it has a fossa and Arctoidea. The ossification ofthe tegmen to accommodate an anterior loc)p of the ar- tympani is another character [3(2)] that unites tery in front of the middle lacersate foramen, these two clades. 1994 WANG AND TEDFORD: CANIDS AND MIACIDS 29

Given the current confusion surrounding 1909) and in the lower Adobe Town Member the systematics of miacids, the present char- of the Washakie Formation equivalent to acterization of the genotypic species Miacis Bridger C-D (Tumbull, 1978) as recorded in parvivorus serves by default as a morphotype this paper. "Miacis" gracilis has been iden- for the paraphyletic family Miacidae. Among tified only in the late Uintan (Uinta C, Clark, the miacids in the present study, M. parvi- 1939). "Hesperocyon" wilsoni is known only vorus is too plesiomorphic to be included in from the Chadronian strata in the Sierra Vie- any derived caniform family. Basicranially, ja approximately 35-36 Ma (Emry et al. it is more derived than Vulpavus in that it 1987). Hesperocyon gregarius (with its syn- has a better ossified tegmen tympani [Char- onym H. paterculus, Wang, MS) occurs in acter 3(1)] and there is a contact of the pro- early Chadronian strata (Emry, 1992) and the montorium and basioccipital (Character 4). genus extends well into the Duchesnean (Bry- A mastoid tubercle (processus hyoideus) made ant, 1992). "Miacis" cognitus is known only up of the petrosal (Character 6) first appears from early Chadronian rocks in the Sierra in M. parvivorus and may be a synapomorphy Vieja, approximately 36-37 Ma (Prothero and at some level within the caniform clade [as- Swisher, 1992). Daphoenus spp. has been re- suming the primitive condition is a squa- corded within the Duchesnean (Bryant, 1992) mosal process which occurs in Viverravus, and the generic range is here extended to the Vulpavus and Didymictis (Flynn and Galia- beginning of that age. no, 1982)], although its distribution in vari- These data provide minimum ages for the ous clades of carnivorans has yet to be clar- initiation ofthe taxa considered in this paper, ified. which in turn can be used to postulate min- A partial classification that is consistent imum divergence dates for the clades rec- with the present phylogeny is indicated in ognized. The time of divergence of the Can- figure 8. Most species of "Miacis" are obvi- iformia from other Carnivora (fig. 9A) is ously in need of generic reassessment, a task constrained by the range of Vulpavus whose beyond the scope of this study. first appearance in Holarctica is at the beginning ofthe Eocene (Savage and Russell, 1983). The divergence of the Cynoidea and Arcto- PHYLOGENETIC HISTORY idea from their sister taxon Miacis parvivorus Ifwe regard the cladogram as a hypothesis (fig. 9B) occurred during the later Wasatchian ofphylogeny, it bears clear implication ofthe as did the appearance of those superfamilies historical relationships between the forms (fig. 9C), both events constrained by the early studied. It can be translated into a phyloge- Bridgerian presence of the Miacis species in- netic diagram by setting the temporal ranges volved. The late Bridgerian presence ofMia- ofthe taxa involved into a time scale (fig. 9). cis sylvestris indicates initiation of the clade The cladistic relationships of the taxa can that includes the Cynoidea (fig. 9D) in the now be calibrated. The resulting hypothesis early Bridgerian close to 50 Ma. This is in about the history of the group will embody agreement with DNA hybridization data predictions about the timing of cladogenetic compiled by Wayne et al. (1989) and cali- events that can be tested as knowledge ofthe brated by an estimate of 40 Ma for the di- temporal ranges of the taxa involved im- vergence of modern caniform families. Our proves and other sister taxa are recognized. revised divergence estimate of the Canidae In this example the temporal ranges of the from "Miacis gracilis" (fig. 9E) in the Uintan taxa are determined as follows: Vulpavus spp. at approximately 45 Ma is younger than the has a generic range extending from its first 50 Ma estimate for the origin of the family appearance at the beginning of the Wasatch- obtained by Wayne et al. (1989) but close to ian (Kristalka et al., 1987) to its documented the 40 Ma estimate they adopted for the ap- last record in Bridger B, early Bridgerian pearance of living carnivore families from (Matthew, 1909). Miacis parvivorus is re- previously available data. Although the rec- corded only from Bridger B, early Bridgerian ord of the Arctoidea does not extend beyond (Matthew, 1909). "Miacis"sylvestris occurs in the Duchesnean, its sister relationship with Bridger C in the Bridger Basin (Matthew, the Cynoidea and the temporal constraints 30 AMERICAN MUSEUM NOVITATES NO. 3092

CANIFORMIA CYNOIDEA ARCTOIDEA CANIDAE I :i*Yij I I E,.1. C: QZ c) I"~~I-co) I 02 0)~~~~~C) -CQ t1 36l- 'I -I' (-u- 0 1 C 'C_) co 42,. . C ...... 0U, 4C-~~~~~~~~~~~~~-O D (D 0.)2X 42 F -J

*C 0 C 46 _a 0 eDco

CA 5 - I D. co 52_ w)0 0 C DC CO 540

co :E

MA --- a m. a. .C Y.

Fig. 9. Phylogenetic history ofthe Caniformia and related taxa obtained by calibrating the cladogram (fig. 8) using the known temporal ranges of the taxa studied. Lettered branch-points are explained in the text. placed on the history of that group, predicts CONCLUSIONS a Bridgerian and Uintan record during which This study was initiated to investigate the time other arctoid families may have arisen large gap in our knowledge between the known (fig. 9G). This revised estimate places initial basicranium morphology ofcanids and those cynoid and arctoid differentiation in the late of the little known miacids. An attempt was Wasatchian or early Bridgerian at about 50 made to arrange a series of miacids into a Ma. The Canidae do not appear in this anal- morphological sequence leading to the emer- ysis to be markedly older than other living gence of the well-known early canid, Hes- caniforms as claimed by Wayne et al. (1989). perocyon. A number of trends may be sum- 1994 WANG AND TEDFORD: CANIDS AND MIACIDS 31 marized along the transformation series from men of "Hesperocyon " wilsoni. Fruitful dis- miacids to primitive canids: cussions with John J. Flynn, Robert M. Hunt, 1. Ossification of an entotympanic bulla. Jr., Ross D. E. MacPhee, and Andre R. Wyss The nearly simultaneous (geologically speak- are gratefully acknowledged. We appreciate ing) occurrences ofa completely ossified bulla the opportunity of examining a manuscript in various families ofcarnivorans suggest the by A. R. Wyss and J. J. Flynn before it is functional importance of a rigid middle ear published. R. M. Hunt supplied stereopho- space rather than the homology ofthis feature tographs of "Miacis" gracilis for our exam- across the order Carnivora as recently pro- ination. Edward Pedersen skillfully prepared posed by Wyss and Flynn (1993), i.e., fossil some key specimens. Mary R. Dawson, John evidence indicates independent appearance J. Flynn, Ernest L. Lundelius, Guy Musser, of a bony bulla in nimravids and amphi- and Charles R. Schaff are thanked for per- cyonids, in addition to canids. The canid bul- mitting us to study specimens under their la, however, has a unique development of a care. Kevin Seymour of Royal Ontario Mu- low septum through the in-bent edge of cau- seum loaned us the cast of Protictis schaffi. dal entotympanic, that is distinguished from We thank Mr. Joel H. Curran of Jackson, the bilaminar septum in the bulla of the fel- Wyoming, for his generous donation of iforms. AMNH 129284 to the American Museum; 2. Anterior and medial expansion of the John P. Alexander of AMNH ably prepared petrosal. The petrosal promontorium is this specimen. transformed from a globular structure isolat- We are indebted to the two reviewers, Drs. ed from the surrounding bones (except the Annalisa Berta and Harold N. Bryant, whose lateral side), to a flattened, pear-shaped struc- incisive criticisms and detailed comments ture in full contact with the adjacent basi- have greatly improved this paper. occipital, basisphenoid, alisphenoid, and squamosal. Perhaps correlated with the pe- REFERENCES trosal expansion is the ossification oftegmen Bryant, H. N. tympani, the flooring of the canal for facial 1991. Phylogenetic relationships and system- nerve, and the closure of piriform fenestra. atics of the Nimravidae (Carnivora). J. 3. Medial migration ofthe internal carotid Mammal. 72: 56-78. artery and loss of the proximal stapedial ar- 1992. The Carnivora ofthe Lac Pelletier lower tery. The promontory branch of the internal fauna (Eocene: Duchesnean), Cypress carotid artery in the primitive miacids shifts Hills Formation, Saskatchewan J. Pa- toward the medial side to become extrabullar leontol. 66: 847-855. in position and the primary carrier ofarterial Cartmill, M., and R. D. E. MacPhee blood to the brain. The stapedial artery, on 1980. Tupaiid affinities: the evidence of the the other hand, is completely lost. This loss carotid arteries and cranial skeleton. In W. P. Luckett (ed.), Comparative biol- of the stapedial artery is often seen in me- ogy and evolutionary relationships of dium to large-size mammals, presumably to tree shrews, pp. 95-132. New York: Ple- avoid the physicdal constraint of a small ob- num. turator foramen of the stapes (Fleischer, Clark, J. 1978). 1939. Miacis gracilis, a new carnivore from In addition to the above trends in the bas- the Unita Eocene (Utah). Ann. Carnegie icranium, canids can be distinguished from Mus. 27: 349-370. miacids by the better developed accessory Cope, E. D. cusps on the premolars, the reduction of the 1872. Third account of new Vertebrata from Ml parastyle, and the loss of the third upper the Bridger Eocene of Wyoming Terri- tory. Paleontol. Bull. 3: 469-472. molar. 1877. Report upon the extinct Vertebrata obtained in New Mexico by parties of the ACKNOWLEDGMENTS expedition of 1874. Report upon Unit- ed States Geological Surveys west ofthe We thank Ernest L. Lundelius who per- one hundredth meridian, Part II, 4: 1- mitted us to further prepare the type speci- 324. 32 AMERICAN MUSEUM NOVITATES NO. 3092

1880. On the genera of the Creodonta. Proc. Hough, J. R. Am. Philos. Soc. Philadelphia 19: 76- 1948. The auditory region in some members 82. of the Procyonidae, Canidae, and Ur- 1883. On the extinct dogs of North America. sidae, its significance in the phylogeny Am. Nat. 17: 235-249. of the carnivora. Bull. Am. Mus. Nat. Emry, R. J. Hist. 92: 71-118. 1992. Mammalian range zones in the Chad- Hunt, R. M., Jr. ronian White River Formation at Flag- 1974. The auditory bulla in Carnivora: an an- staffRim, Wyoming. In D. R. Prothero atomical basis for reappraisal of carni- and W. A. Berggren (eds.), Eocene-Oli- voran evolution. J. Morphol. 143: 21- gocene climatic and biotic evolution, pp. 76. 106-115. Princeton: Princeton Univ. 1977. Basicranial anatomy of Cynelos Jour- Press. dan (Mammalia: Carnivora), an Aqui- Emry, R. J., P. R. Bjork, and L. S. Russell tanian Amphicyonid from the Allier 1987. The Chadronian, Orellan, and Whit- Basin, France. J. Paleontol. 51: 826-843. neyan North American land mammal 1987. Evolution of the aeluroid Carnivora: ages. In M. 0. Woodburne (ed.), Ce- significance of auditory structure in the nozoic mammals ofNorth America, ge- nimravid cat Dinictis. Am. Mus. Nov- ochronology and biostratigraphy, pp. itates 2886: 74 pp. 118-152. Berkeley: Univ. California 1989. Evolution ofthe aeluroid carnivora: sig- Press. nificance of the ventral promontorial Evans, H. E., and G. C. Christensen process of the petrosal, and the origin 1979. Miller's anatomy of the dog. Philadel- ofbasicranial patterns in the living fam- phia: W. B. Saunders. ilies. Am. Mus. Novitates 2930: 32 pp. Fleischer, G. Hunt, R. M., and R. H. Tedford 1978. Evolutionary principles of the mam- 1993. Phylogenetic relationships within the malian middle ear. Adv. Anat. Em- aeluroid carnivora and implications of bryol. Cell Biol. 55: 1-70. their temporal and geographic distri- Flower, W. H. bution. In F. S. Szalay, M. J. Novacek, 1869. On the value of the characters of the and M. C. McKenna (eds.), Mammal base ofthe cranium in the classification phylogeny: Placentals, pp. 52-73. New of the order Carnivora, and on the sys- York: Springer-Verlag. tematic position of Bassaris and other Krishtalka, L., R. K. Stucky, R. M. West, M. C. disputed forms. Proc. Zool. Soc. Lon- McKenna, C. C. Black, T. M. Bown, M. R. Daw- don 1869: pp. 4-37. son, D. J. Golz, J. J. Flynn, J. A. Lillegraven, Flynn, J. J., and H. Galiano and W. D. Turnbull 1982. Phylogeny of early Tertiary Carnivora, 1987. Eocene (Wasatchian through Duches- with a description of a new species of nean) biochronology ofNorth America. Protictis from the middle Eocene of In M. 0. Woodburne (ed.), Cenozoic northwestern Wyoming. Am. Mus. mammals of North America, geochro- Novitates 2725: 64 pp. nology and biostratigraphy, pp. 77-117. Flynn, J. J., N. A. Neff, and R. H. Tedford Berkeley: Univ. California Press. 1988. Phylogeny of the Carnivora. In M. J. Benton (ed.), The phylogeny and clas- MacPhee, R. D. E. sification of the tetrapods, Volume 2: 1981. Auditory regions of primates and eu- Mammals. Systematics Association therian insectivores: morphology, on- Special Volume 35B: 73-116. Oxford: togeny, and character analysis. Contrib. Clarendon Press. Primatol. 18: 1-282. Gingerich, P. D., and D. A. Winkler MacPhee, R. D. E., M. J. Novacek, and G. Storch 1985. Systematics of Paleocene Viverravidae 1988. Basicranial morphology of early Terti- (Mammalia, Carnivora) in the Bighorn ary erinaceomorphs and the origin of Basin and Clark's Fork Basin, Wyo- Primates. Am. Mus. Novitates 2921: 42 ming. Contrib. Mus. Paleontol., Univ. pp- Michigan, 27: 87-128. Marsh, 0. C. Gustafson, E. P. 1872. Notice of some new fossil mammals 1986. Carnivorous mammals of the Late Eo- from the Tertiary formations of the cene and Early Oligocene of Trans-Pe- West, Part II-IV. Am. J. Sci. 4: 202- cos Texas. Texas Mem. Mus. Bull. 33: 224. 1-66. Matthew, W. D. 1994 WANG AND TEDFORD: CANIDS AND MIACIDS 33

1901. Fossil mammals of the Tertiary of River Oligocene, Part I. Insectivora and Northeastern Colorado. Mem. Am. Carnivora. Ibid., 28: 1-980. Mus. Nat. Hist. 1: 355-448. Simpson, G. G. 1909. The Carnivora and Insectivora of the 1945. The principles of classification and a Bridger Basin, Middle Eocene. Ibid., 9: classification of mammals. Bull. Am. 291-567. Mus. Nat. Hist. 85: 1-350. 1930. The phylogeny ofdogs. J. Mammal. 11: Tedford, R. H. 117-138. 1976. Relationship of Pinnipeds to other car- Mitchell, E., and R. H. Tedford nivorans (Mammalia). Syst. Zool. 25: 1973. The Enaliarctinae a new group ofextinct 363-374. aquatic Carnivora and a consideration 1978. History of dogs and cats: a view from ofthe origin ofthe Otariidae. Bull. Am. the fossil record. In Nutrition and Man- Mus. Nat. Hist. 151: 203-284. agement of dogs and cats, pp. 1-10. St. Neff, N. A. Louis: Ralston Purina Company. 1983. The basicranial anatomy of the Nim- Tedford, R. H., L. G. Barnes, and C. E. Ray ravidae (Mammalia: Carnivora): char- In press. The early Miocene littoral ursoid car- acter analyses and phylogenetic infer- nivoran, Kolponomos; systematics and ences. Ph.D. diss., City Univ. of New mode of life. In A. Berta and T. A. De- York. mere (eds.), Marine Mammal Evolution Novacek, M. J. Symposium, 51st Annual Meeting of 1980. Cranioskeletal features in tupaiids and Society of Vertebrate Paleontology, selected Eutheria as phylogenetic evi- Proceedings of the San Diego Natural dence. In W. P. Luckett (ed.), Compar- History Museum. ative biology and evolutionary relation- Tumbull, W. D. ships of tree shrews, pp. 35-93. New 1978. The mammalian faunas ofthe Washak- York: Plenum. ie Formation, Eocene age, of southern 1986. The skull of leptictid insectivorans and Wyoming. Part I: Introduction: The ge- the higher-level classification ofeuther- ology, history and setting. Fieldiana: ian mammals. Bull. Am. Mus. Nat. Hist. Geol. 30: 569-601. 183: 1-111. Wang, X. Presley, R. 1990. Systematics, functional morphology, 1979. The primitive course of the internal ca- and evolution of primitive Canidae rotid artery in mammals. Acta Anat. (Mammalia: Carnivora). Ph.D. diss., 103: 238-244. Univ. Kansas, Lawrence. Prothero, D. R, and C. C. Swisher III 1993. Transformation from plantigrady to 1992. Magnetostratigraphy and geochronolo- digitigrady: functional morphology of gy of the terrestrial Eocene-Oligocene locomotion in Hesperocyon (Canidae: transition in North America. In D. R. Carnivora). Am. Mus. Novitates 3069: Prothero and W. A. Berggren (eds.), Eo- 23 pp. cene-Oligocene climatic and biotic evo- MS. Phylogenetic systematics of Hespero- lution, pp. 46-73. Princeton: Princeton cyoninae (Carnivora: Canidae). Univ. Press. Wang, X., and R. H. Tedford Savage, D. E., and D. E. Russell 1991. Primitive course of internal carotid ar- 1983. Mammalian paleofaunas of the World. tery in caniform "miacoid" carnivor- Reading: Addison-Wesley. ans: implications for phylogenetic re- Schmidt-Kittler, N. lationships. J. Vert. Paleontol. 11 (suppl. 1981. Zur Stammesgeschichte der marden- 3): 60A. verwandten Raubtiergruppen (Muste- Wayne, R. K., R. E. Benveniste, D. N. Janczewski, loidea, Carnivora). Eclo. Geol. Helv. 74: and S. J. O'Brien 753-801. 1989. Molecularand biochemical evolution of Scott, W. B. the Carnivora. In J. L. Gittleman (ed.), 1895. The mammalia ofthe Deep River Beds. Carnivore behavior, ecology, and evo- Trans. Am. Philos. Soc. Philadelphia 18: lution, pp. 465-494. Ithaca: Cornell 55-185. Univ. Press. 1898. Notes on the Canidae ofthe White Riv- Wible, J. R. er Oligocene. Ibid., 19: 325-415. 1983. The internal carotid artery in early euth- Scott, W. B., and G. L. Jepsen erians. Acta Palaeontol. Polonica 28: 1936. The mammalian fauna of the White 281-293. 34 AMERICAN MUSEUM NOVITATES NO. 3092

1984. The ontogeny and phylogeny of the Wozencraft, W. C. mammalian cranial arterial pattern. 1989. The phylogeny ofthe Recent Carnivora. Ph.D. diss., Duke Univ., Durham. In J. L. Gittleman (ed.), Carnivore be- Wilson, J. A., P. C. Twiss, R. K. DeFord, and S. havior, ecology, and evolution, pp. 495- E. Clabauch 535. Ithaca: Cornell Univ. Press. 1968. Stratigraphic succession, potassium-ar- Wyss, A. R., and J. J. Flynn gon dates, and vertebrate faunas, Vieja 1993. A phylogenetic analysis and definition Group, Rim Rock country, Trans-Pecos of the Carnivora, pp. 32-52. In F. S. Texas. Am. J. Sci. 266: 590-604. Szalay, M. J. Novacek, and M. C. Mc- Wortman, J. L. Kenna (eds.), Mammal phylogeny: Pla- 1901. Studies of Eocene Mammalia in the centals. New York: Springer-Verlag. Marsh Collection, Peabody Museum. Ibid., 11: 1-145.

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.

0) This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).