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

NovitatesAMERICAN MUSEUM PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 3023, 34 pp., 16 figures September 11, 1991

Evolution of the Aeluroid Camivora: Viverrid Affinities of the Miocene Camivoran Herpestides

ROBERT M. HUNT, JR.'

CONTENTS Abstract ...... 2 Introduction ...... 2 Acknowledgments ...... 2 Abbreviations ...... 4 Geologic and Geographic Setting ...... 4 Basicrania of Herpestides ...... 5 Auditory Region ofHerpestides antiquus ...... 6 Comparative Morphology of the Aeluroid Middle Ear ...... 12 Auditory Ossicles in Aeluroids ...... 21 Phylogenetic Affinities ofHerpestides ...... 26 The Problem of Hyaenid-Herpestid Relationship ...... 30 Antiquity of the Viverrid Auditory Region ...... 32 Conclusions ...... 32 References ...... 33

XResearch Associate, Department of Vertebrate Paleontology, American Museum of Natural History; Curator, Division of Vertebrate Paleontology, University of Nebraska, Lincoln, NE 68588-0514.

ABSTRACT Although the time of origin of viverrid and viverrid grade of development in the early Mio- hyaenid carnivorans has not been clearly docu- cene (European Neogene mammal zone MN2a), mented in the fossil record, their theater of evo- and suggests that diversification ofthe Viverridae lution has long been established by a mid-Ceno- was in progress by this time. In ongoing work to zoic fossil distribution entirely confined to the Old be published elsewhere, the mid-Miocene Asian World. Recent examination of the basicranial camivoran Tungurictis spocki, long regarded as a morphology of important early aeluroid crania viverrid, is identified as an early hyaenid, follow- from Europe and Asia significantly alters earlier ing preparation and restudy ofthe auditory region views of viverrid and hyaenid origins. The early of the genoholotype cranium from Tung Gur, Miocene carnivoran Herpestides antiquus, consid- Mongolia. These discoveries indicate that sepa- ered a potential ancestral hyaenid or herpestid in ration of the modem aeluroid families as discrete earlier studies, is identified as a true viverrid on lineages had been accomplished by the beginning the basis of a large sample of skulls of both ju- of the Neogene in the Old World, and that diver- veniles and adults from Aquitanian sediments of sification within these families must have been the Allier basin, France. The basicranial mor- initiated in the early to mid-Miocene. phology of Herpestides has attained the modem

INTRODUCTION Among the superb carnivoran fossils re- complete and partial crania ofHerpestides at covered from the Aquitanian deposits ofthe the Musee Guimet d'Histoire Naturelle in Allier basin, France, are abundant cranial and Lyon, and at the Naturhistorisches Museum, postcranial remains of a small civetlike ae- Basel, representing about 12 individuals, luroid, originally described by Blainville which forms the basis for this report. (1842) under the name "Viverra" antiqua. I did not examine the question ofthe num- As additional remains were discovered, they ber ofspecies attributable to Herpestides from were placed by later European students (Po- the St.-Gerand region. It is sufficient to note mel, 1853; Filhol, 1879; Schlosser, 1890; Vi- that Beaumont (1967), who has studied the ret, 1929) in several species allocated to the nature ofdental variation in this carnivoran, extant genus Herpestes. However, in a thor- determined that dental and cranial materials ough, recent review of this material, Beau- suggest a range ofvariation somewhat greater mont (1967) affirmed its distinctness from than that found in most living aeluroid spe- Herpestes and created the genus Herpestides cies (fig. 2). The amount ofvariation in dental for this Aquitanian camivoran lineage, rec- and cranial dimensions, however, is in keep- ognizing only a single species highly variable ing with geographically variable populations in size and dental morphology, Herpestides of a lineage sampled over a short interval of antiquus (Blainville). time, as pointed out by Beaumont. Signifi- Herpestides is the oldest Eurasian or Af- cantly, the degree of variation observed in rican aeluroid carnivoran represented by nu- the dentition, both in terms of morphology merous skulls with intact basicrania, includ- and size, is not evident in the basicranial ing auditory bullae (fig. 1). A number ofthese morphology, which displays a high degree of fossil skulls retain osseous basicranial struc- uniformity. Thus, regardless of the number ture as well preserved as in the living animal, of species of Herpestides finally determined and both juveniles and adults are represent- from the Aquitanian sediments of the Allier ed. My intent was to examine an adequate basin, the fossils indicate a morphologically sample of the crania of Herpestides in order uniform, closely related assemblage referable to determine its basicranial and bullar mor- to a single genus. phology; to discuss the ontogenetic pattern of bulla formation; and to attempt to estab- lish the higher-level systematic relationships ACKNOWLEDGMENTS at the family level. In September 1989, I was I wish to express my appreciation to Dr. given the opportunity to study the sample of Michel Philippe for the numerous courtesies 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES 3

Fig. 1. Cranium and associated lower jaw ofthe viverrid Herpestides antiquus (Blainville), MGL St.- G. 3066, from the Aquitanian, Allier basin, France. Natural size.

Fig. 2. Rostra of a large (left, NMB 6373) and a small (right, NMB 12379) individual ofHerpestides from Montaigu-le-Blin, Allier basin, France, demonstrating the range in size attributed to H. antiquus by earlier workers. On dental traits, both are adults. Scale bar in this and all subsequent figures is 1 cm in length. 4 AMERICAN MUSEUM NOVITATES NO. 3023

extended me during my research at the Musee m depression in mastoid produced by cau- Guimet d'Histoire Naturelle in Lyon, and to dal entotympanic Drs. Pierre Mein and Marguerite Hugueney OS orbitosphenoid for their assistance in the collections of P petrosal the p epitympanic wing of petrosal Faculte des Sciences, Universite de Lyon. I PC groove in entotympanic for internal ca- also wish to thank M. Hugueney for recent rotid artery, leading to posterior carotid publications and discussion of the geologic foramen setting of the St.-Gerand deposits. My work plf posterior lacerate foramen at the Naturhistorisches Museum, Basel, Pp, pp paroccipital process of exoccipital benefited from conversations with Dr. Burk- PT pterygoid art Engesser and Prof. Dr. Johannes Hiirzeler R rostral entotympanic that improved my comprehension of issues rc basioccipital attachment for rectus capi- relevant to this study. M. Francois Escuillie, tis ventralis Dept. des Sciences de la Terre, S suprameatal fossa Lyon, gen- sb septum bullae erously allowed me to examine two recently SQ squamosal collected skulls of Herpestides from the lo- T ectotympanic calities ofMontaigu-le-Blin and Gepiac which t fossa for the tensor tympani he will describe in his work on these carni- th tympanohyal vorans. I am particularly grateful to Dr. Guy V ventral process ofthe petrosal promon- Musser and Dr. Patricia Freeman who pro- torium vided the opportunity to study and dissect vf Vidian foramen the basicrania ofrepresentative aeluroid car- x apex of the caudal entotympanic fused nivorans in collections in their care. Refer- to the posterior edge ofrostral entotym- ence casts of basicrania used in this study panic were produced by Head Preparator Gregory Brown, University of Nebraska State Muse- AMNH American Museum of Natural History, um. My thanks to R. M. Joeckel, Harold Bry- Mammalogy ant, and CliffLemen for review ofthe manu- FSL Faculte des Sciences, Universite de script. Lyon, France MGL Musee Guimet d'Histoire Naturelle, ABBREVIATIONS Lyon, France NMB Naturhistorisches Museum, Basel, A alisphenoid Switzerland a epitympanic wing of alisphenoid UNSM University of Nebraska State Museum, ac alisphenoid canal Zoology b epitympanic wing of basisphenoid BO basioccipital bof basioccipital flange GEOLOGIC AND GEOGRAPHIC BS basisphenoid SETTING c canal for the facial (VII) nerve Ca pit in squamosal for anterior crus ofec- The remains ofHerpestides discussed here- totympanic in were found in nonmarine early Miocene d depression in alisphenoid for anterior (Aquitanian) sediments ofthe St.-Gerand re- limb of ectotympanic gion, Limagne d'Allier, central France. The EO exoccipital Limagnes are Cenozoic fault-bounded basins -E caudal entotympanic ("fosses d'effondrement") within the Massif e epitympanic recess Central of France that developed contem- ee epitympanic wing of exoccipital poraneously with Alpine tectonism. f flange of septum bullae contacting pro- The montorium anterior to round window Limagne d'Allier that contains the fossilif- fo foramen ovale erous Aquitanian sediments of the St.-Ge- g postglenoid foramen rand sub-basin is a complex graben fill oflate h hypoglossal (condyloid) foramen Eocene to early Miocene age, dissected by the ica tube for internal carotid artery modern Allier River flowing northward from L middle lacerate foramen the Massif Central. Aquitanian sediments in M mastoid process the vicinity of St.-Gerand-le-Puy have been 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES 5 commercially exploited for over a century for trial reptiles and birds; there are abundant limestone, resulting in a network of quarries remains of diverse terrestrial mammals (di- that produce the vertebrate fauna. The dis- delphid marsupials; insectivores; bats, lago- tribution of the principal localities in the re- morphs and rodents; viverrid, felid, and gion has recently been illustrated by Bucher, mustelid carnivorans; small artiodactyls). Ginsburg, and Cheneval (1985), and a map Of particular interest is the occurrence of of the quarries in the St.-Gerand sub-basin vertebrate bones in brown marls with hy- near the villages ofMontaigu-le-Blin, St.-Ge- drobids. These brown marls are similar to rand, and Langy appears in Cheneval and the green marls in that they also infiltrate Hugueney (1985). between stromalitic mounds, but in the brown Fossils of Herpestides that I have studied marls the associated vertebrates are primarily in the Museum at Lyon are simply attributed aquatic (freshwater fishes and aquatic birds); to "St.-Gerand," and lack a more specific lo- mammals are represented only by abundant cality designation. As is well known, fossils remains of viverrid carnivorans, a situation designated by the term "St.-Gerand" could not understood at the present time. However, have been found in any ofthe limestone quar- similar lack of diversity appears in contem- ries ofthe region (Viret, 1929; Cheneval and poraneous fluvio-deltaic biocalcarenite in Hugueney, 1985). However, the crania from which only partial skeletons ofa single family the Naturhistorisches Museum, Basel, are all of aquatic birds have been found. Such sit- attributed to Montaigu-le-Blin, with the year uations reflect local ecologic and taphonomic ofcollection appended. Today several active biases. as well as inactive quarries occur near the Vertebrate bones have been found within village of Montaigu-le-Blin (Cheneval and the stromatolites themselves, but are rare. In Hugueney, 1985), but it is uncertain at which "Les Perards" quarry near Montaigu, only of these sites the crania were found. Recent aquatic vertebrates have been found within studies of the Aquitanian sediments near these structures. Montaigu-le-Blin allow an improved under- Bentonites composed of montmorillonitic standing of their depositional environment clay associated with these deposits indicate (Bucher et al., 1985; Hugueney, 1984; Don- contemporaneous volcanism, but volcano- simoni and Giot, 1977). genic crystals for radiometric dating have not The depositional setting of the vertebrate been recovered from the bentonites (Bucher remains are described in a recent study of etal., 1985). "Les Perards" quarry near Montaigu-le-Blin Was Herpestides an aquatic carnivoran? by Bucher, Ginsburg, and Cheneval (1985), How can the abundance of its remains be and my subsequent remarks are based on their explained in these deposits? Its skeletal struc- illuminating publication and an excursion ture does not possess any evident specializa- guide by Cheneval and Hugueney (1985). De- tion that might indicate restriction to an positional environments are primarily flu- aquatic setting; the animal may have been a vio-lacustrine, and record the waxing and capable swimmer but this assumption is waning of local carbonate-rich lakes within probably not necessary to account for its good the Allier basin. Terrestrial vertebrates (in- representation in lacustrine sediments. Rath- cluding viverrid carnivorans) occur in green er, the abundance ofHerpestides suggests that marls with abundant shells ofthe pulmonate it frequented lake margins, and its dental gastropod Cepaea moroguesi deposited be- morphology and postcranial skeleton indi- tween columnar lacustrine stromatolites cate that it was an active predator, probably (phryganid algal bioherms). These green marls taking small terrestrial and aquatic verte- lack internal stratification and have been in- brates living in these perilacustrine environ- terpreted as mudflows that swept littoral de- ments. bris (including vertebrate bones) downslope onto the floor ofthese lakes, surrounding and influencing the growth pattern of the stro- BASICRANIA OF HERPESTIDES matolite community. Vertebrates within the Complete or partial crania of 12 individ- green marls include both aquatic and terres- uals were examined in the collections at Lyon 6 AMERICAN MUSEUM NOVITATES NO. 3023 and Basel. Of these, four are represented by has been attached to the posterior cranium complete skulls (two with associated lower with plaster, there is no certain contact be- jaws), seven are posterior crania (only one tween the two parts. The rostrum includes with associated upper teeth), and one is the the left P3-4. skull of a small individual lacking only the (6) FSL unnumbered. - Complete skull occiput. These fossils comprise the hypodigm with basicranium and both auditory bullae under study and are listed and briefly de- intact, currently undergoing preparation and scribed below: study by M. Escuillie, Universite de Lyon. (1) MGL St.-G. 3065. - Complete skull (7) NMB S.G. 11583. - Posterior cranium, (basilar length, 89.6 mm) without lowerjaws, lacking rostrum and dentition. Basicranium but retains in the maxillae both right and left with right and left auditory regions very well P3-4 and Ml. This is the only skull in the preserved. Auditory bullae are represented MGL collection that preserves a complete only by the right anterior ectotympanic and and intact auditory bulla. The complete left rostral entotympanic, and the left rostral en- bulla includes both ectotympanic and caudal totympanic; caudal entotympanics on both entotympanic; the rostral entotympanic is not sides are missing. Montaigu-le-Blin, 1921. visible. The right bulla lacks only the ventral (8) NMB S.G. 12377. - Partial posterior part of caudal entotympanic. Length of the cranium, with left auditory region only, but complete bulla from the anteriormost part of with most of bulla missing. Only a dorsal ectotympanic to posterior limit ofthe caudal remnant of caudal entotympanic remains in entotympanic is 19.4 mm (exclusive of the place. Associated with left and right maxillae paroccipital process). with P2-4 present on each side. Montaigu- (2) MGL St.-G. 3066. - Complete skull le-Blin, 1921. (basilar length, 102.8 mm) associated with (9) NMB S.G. 6890. - Posterior cranium, both lower jaws. Upper dentition includes lacking rostrum and dentition, but with both right C, P2-4, M1-2; left C, P3-4, M1-2 auditory regions preserved. Bullae are absent (rarely does a maxilla retain M2). Lower den- except for a small caudal entotympanic frag- tition includes right c, p2-4, m 1-2; left c, p3- ment on the right, and a detached ectotym- 4, m 1. A well-preserved skull with intact ba- panic. Montaigu-le-Blin, 1913. sicranium, in which the auditory bullae lack (10) NMB S.G.2935.-Posteriorcranium, only the ventral parts ofthe caudal entotym- lacking rostrum and dentition, but with both panics. auditory regions well preserved. Bullae are (3) MGL St.-G. 3067. - Posterior crani- entirely lacking. A juvenile with open basi- um, including part of the frontal region but sphenoid-basioccipital suture. Montaigu-le- lacking rostrum and dentition. The left au- Blin, 1913. ditory region (including the petrosal) and the (11) NMB S.G.7244.-Posteriorcranium, occiput are missing; the petrosal and ecto- lacking rostrum and dentition, and without tympanic chamber of the right auditory re- the occiput. Both petrosals are present, but gion are preserved. Ofinterest because it rep- bullae are absent. A juvenile with open basi- resents the largest individual in the MGL sphenoid-basioccipital suture. Montaigu-le- sample (estimated basilar length, 110 mm). Blin, 1932. (4) MGL St.-G. 3068.-Juvenile posterior (12) NMB S.G. 11407. - Complete skull cranium (indicated by an open basisphenoid- and associated lower jaws. Both auditory basioccipital suture), without rostrum or regions are present, but only the left auditory dentition. Both auditory bullae are absent, bulla is exposed. Skull somewhat crushed. suggesting that in juveniles ecto- and ento- Montaigu-le-Blin, 1920. tympanic elements are loosely attached to the basicranium. Both petrosals are present and well preserved. AUDITORY REGION OF (5) MGL St.-G. 2009. - Nearly complete HERPESTIDES skull (estimated basilar length, 80 mm), The sample ofHerpestides crania from the lacking occiput and basicranium, represent- St.-Gerand region represents an age spectrum ing a very small individual. Caution should from juveniles to mature adults (neonatal in- be exercised in designating this specimen as dividuals are not present). The morphology one individual because, although the rostrum of the auditory region can be determined at 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES 7

Fig. 3. Basicrania ofHerpestidesjuvenile (A, NMB 2935) and adult (B, NMB 11583) from Montaigu- le-Blin in ventral view. In B, small triangles indicate the perimeter of attachment for the caudal entotympanic in the right posterior auditory region; both right and left caudal entotympanics have been removed from the basicranium. Note that in the adult (B) the ectotympanic and rostral entotympanic are not fused to surrounding bones, and the sutures between epitympanic elements forming the roof of the posterior auditory region remain open. For abbreviations in this and all subsequent figures, see p. 4. Stereopairs. several ontogenetic stages by examining a se- observed in this sample ofHerpestides crania ries ofthese basicrania in which the auditory demonstrate a morphology typical of extant bullae range from complete, to broken open, Viverridae. to fully detached from the skull. As a result, the internal structure and geometry of the middle ear and the composition of the bulla PETROSAL-MASTOID can be studied in detail. Several skulls also In both juveniles (figs. 3A, 4A) and adults possess intact auditory regions for which ex- (figs. 3B, 4B) the petrosal displays a charac- ternal basicranial morphology can be deter- teristic form most similar to that of living mined. In all respects, the auditory regions viverrid carnivorans such as Civettictis civet- 8 AMERICAN MUSEUM NOVITATES NO. 3023

Fig. 4. Comparison of juvenile (A, NMB 2935) and adult (B, NMB 11583) auditory regions of Herpestides in ventrolateral view, same individuals as in figure 3, both from Montaigu-le-Blin. Note the basioccipital flange (bof) and the pocketing of the lateral part of basioccipital (asterisks) that develop in adults during ontogeny. Stereopairs. ta. The structural geometry ofthe middle ear bulla's medial wall. The posteromedial face region, including the configuration ofthe bul- of the promontorium is deeply impressed by la elements, is largely determined by a robust an inflated caudal entotympanic element in promontorium that forms a transverse ridge juvenile and adult. Lateral to this pro- dividing the middle ear into anterior and pos- nounced indentation the promontorium terior chambers. A prominent ventral pro- bulges outward behind the round window as cess of the promontorium forms a medial a knoblike rugose process. The posterior edge extension ofthis ridge, buttressing the lateral ofthe petrosal is prolonged as one or two thin face of the basioccipital immediately poste- sheetlike epitympanic processes ofbone that rior to the basisphenoid-basioccipital suture contribute to the roof ofthe posterior cham- (fig. 3A). This process is as well developed in ber; the thin dorsal surface of the caudal en- Herpestides as in living aeluroid carnivorans, totympanic element is closely applied to these in which it is a key synapomorphy of the epitympanic processes. group (Hunt, 1989). The tegmen tympani is identified as a Both the anterior and posterior faces ofthe prominent bony shelf ofthe petrosal that ex- promontorium slope steeply dorsad away tends anterior and lateral to the promonto- from the transverse ridge that forms the pro- rium; its ventral surface bears a sharply de- montorial apex. On the anteromedial face of marcated tensor tympani fossa, epitympanic the promontorium rests an ossified rostral recess, and facial canal, described by Beau- entotympanic element that contributes to the mont (1967). Whether the tegmen is in fact 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES 9 a composite structure formed by both a true dividuals, is always quite small, indicating a tegmen and by additional epitympanic pro- vestigial postglenoid vein. Posterodorsal to cesses of the petrosal, as described in some this foramen is a deep pit for insertion ofthe eutherians by MacPhee (1981), is unknown, anterior crus of ectotympanic. Medial to the but for descriptive purposes the entire shelf pit the squamosal is grooved by the Glaserian anterior as well as lateral to the promonto- fissure for the chorda tympani branch of the rium is herein termed the tegmen. In juve- facial nerve. The fissure is situated imme- niles, an unfused suture between the petrosal, diately lateral and parallel to the anteropos- sphenoid, and squamosal clearly defines the teriorly aligned alisphenoid-squamosal su- limits of the tegmen (fig. 3A). The anterior ture. portion of the tegmen is deeply pocketed by The squamosal forms the roof ofthe bony a medially placed tensor tympani fossa and external auditory meatus directly posterior to a laterally placed epitympanic recess. The pe- the pit for the anterior crus of ectotympanic. trosal-squamosal suture runs anteroposteri- Injuveniles and adults the roofofthe meatus orly through the epitympanic recess, hence is impressed by a deep suprameatal fossa, as its lateral part lies in the squamosal and its well developed as those observed in living medial portion in the tegmen. Posteromedial procyonids, presumably incorporating a to the epitympanic recess the facial canal prominent pars flaccida of the tympanum. emerges dorsal to the oval window in the Dorsal to the suprameatal fossa the medial lateral portion of the tegmen; the nerve fol- face ofthe squamosal is pocketed by the epi- lows a groove in the petromastoid, turning tympanic recess (there were no auditory os- laterad to exit the cranium via the stylomas- sicles preserved in any of the crania under toid foramen. A prominent fossa lies postero- study). The suprameatal fossa is excavated medial to the facial canal and dorsolateral entirely in the squamosal; its posterior wall to the round window in the posterior part of is formed by the posttympanic process ofthe the tegmen. squamosal applied against the anterior face In several juveniles a suture between the ofthe mastoid, the two bones uniting to form mastoid and tegmen may be present (fig. 3A), the mastoid process. The mastoid-squamosal as in certain living viverrids. In Viverricula suture must fuse early in ontogeny because, and Civettictis there is a persistent suture be- even in juveniles, a distinct sutural contact tween the lateral face of petrosal and what between the mastoid bone and the posttym- appears to be a fused squamosal-mastoid plate panic process is difficult to identify, despite lateral to it. In herpestids, felids, and hyae- an open sutural contact between squamosal nids a distinct mastoid is present that be- and tegmen. comes attached to the petrosal in early development, the mastoid later fusing to squamosal. In viverrids, it seems probable ALISPHENOID-BASISPHENOID that the squamosal-mastoid fusion occurs The epitympanic wings of alisphenoid and first, forming a squamous lamina that covers basisphenoid contribute the roof of the au- the petrosal without fusing to it. ditory region anterior to the petrosal. In juveniles a visible basisphenoid-alisphenoid SQUAMOSAL suture demonstrates that the basisphenoid wing completes the roof medial to the tensor In the basicranium the sutural boundaries tympani fossa and a somewhat larger alisphe- of the squamosal are well defined, particu- noid wing is situated anterior to the fossa. larly in juveniles (fig. 3A). The basicranial The basisphenoid wing closes the antero- portion of the squamosal posterior to the medial corner ofthe auditory region; the ros- postglenoid process contributes the antero- tral entotympanic is applied against its lateral lateral part of the auditory region to which face, and it is perforated in its anterior part the ectotympanic crura are attached. On the by the middle lacerate foramen for entrance posterior slope of this process near its base of the internal carotid artery into the cranial is a reduced postglenoid foramen which, al- cavity. The pterygoid (Vidian) canal for the though variable in diameter in different in- Vidian nerve (made up of the greater super- 10 AMERICAN MUSEUM NOVITATES NO. 3023

ficial and deep petrosal nerves) can be seen inent elliptical depression for a medial ex- as a narrow channel on the surface of the pansion of caudal entotympanic into basisphenoid wing (fig. 4B). basioccipital; and the edge of the basioccip- The ectotympanic's anterior part indents ital is ventrally extended to buttress the en- the epitympanic wing ofalisphenoid, forming larging caudal entotympanic chamber. There a prominent elliptical depression in adults. is no epitympanic wing of the basioccipital: The size, depth, and orientation of impres- instead the basioccipital is entirely involved sions in alisphenoid and squamosal made by in supporting the medial wall of the bulla, the anterior face of ectotympanic and the tip and in consequence its lateral margin be- ofits anterior crus, respectively, provide im- comes modified by contact with the growing portant information on the configuration of caudal entotympanic element. the anterior bulla chamber in a variety of Similarly, the exoccipital is also in contact fossil carnivorans in which the ectotympanic with the expanding caudal entotympanic and has been lost from the skull. becomes modified in shape during the pro- The basisphenoid-basioccipital suture is cess of ontogenetic growth. The exoccipital open in juveniles and young adults, and is forms the posterior wall of the auditory re- fused in mature animals (compare fig. 3A, B). gion, extending ventrad as the paroccipital process (fig. 3). This posterior wall becomes BASIOCCIPITAL-EXOCCIPITAL thinned and tightly appressed against the rap- idly inflating caudal entotympanic chamber The basioccipital buttresses the greater part of the bulla. The expansion of caudal ento- of the medial wall of the auditory bulla, tympanic is so pervasive that the adjacent whereas the exoccipital supports its posterior mastoid bone becomes impressed and slight- wall. During ontogeny in Herpestides, these ly displaced by caudal entotympanic growth. two bones are significantly modified to a Furthermore, the exoccipital develops an epi- greater degree than any other basicranial tympanic wing that extends dorsal to the cau- bones bordering the auditory region. This dal entotympanic and fills in the roof of the modification results from progressive onto- posterior chamber. This epitympanic wing genetic growth and encroachment ofthe cau- retains an open sutural contact with the mas- dal entotympanic on the basi- and exoccip- toid and with the posterior epitympanic wing ital. of the petrosal, even in adults (fig. 3B). Figures 3 and 4 indicate the extent of this modification during bulla growth. In the juvenile (figs. 3A, 4A) the lateral edge of ba- RoSTRAL ENToTYMpAMc sioccipital is not produced, but in the adult The rostral entotympanic in adults (figs. (figs. 3B, 4B) this edge is greatly extended 3B, 4B) is situated in the anterointernal cor- ventrad, and the rugosities on the basioccip- ner of the auditory region, where it forms a ital's ventral surface for attachment of the small wedge ofbone between the anterior face rectus capitis ventralis muscles are much more of the petrosal promontorium and the epi- developed. When seen in lateral view (fig. 4), tympanic wing ofthe basisphenoid. Approx- the edge ofbasioccipital that borders the mid- imately triangular in lateral view, its apex is dle ear cavity can be divided into an anterior directed dorsad toward the tensor tympani part in direct association with the ventral fossa. Its anterior edge fits against the basi- promontorial process of the petrosal, and a sphenoid where a small ridge separates it from posterior part behind this process. In the ju- the pterygoid canal, and its posterior edge is venile (fig. 4A), the anterior and posterior applied to the petrosal. The ventral margin parts are not deeply indented by the caudal ofrostral entotympanic is in contact and fused entotympanic, nor is the edge ofbasioccipital to the ectotympanic and caudal entotympan- strongly extended ventrad. However, in the ic (fig. 4B), but even in adults it does not fuse adult (fig. 4B), the anterior part is indented to either petrosal or basisphenoid; these su- by the forwardly growing apex of caudal en- tures remain open. The medial edge of the totympanic; the posterior part shows a prom- ectotympanic contacts and fuses to the ven- 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES I1I

tral edge ofrostral entotympanic (thictic state, ectotympanic is shown in relationship to the Hunt, 1987). The anterior apex of the for- petrosal, the entire caudal entotympanic hav- ward-growing caudal entotympanic element ing been removed. In figure 5B the ectotym- contacts and fuses to the posteroventral cor- panic appears in relation to the dorsal part ner of rostral entotympanic (fig. 4B). The of the caudal entotympanic (and the petro- ventral margin ofrostral entotympanic is ex- sal). posed on the bulla surface and can be rec- Note in figure 5A the relatively inflated or ognized even after it has fused with ectotym- expanded ectotympanic with robust anterior panic (fig. 3B). crus seated in the squamosal, and a more The anteroventral corner of rostral ento- gracile posterior crus resting on the posttym- tympanic borders on the middle lacerate fo- panic process of the squamosal. Internal to ramen. The foramen is entirely within the the tip ofthe posterior crus the triangular end basisphenoid and transmits the internal ca- of the tympanohyal can be seen entering the rotid artery to the cranial cavity. Directly in auditory region where it fuses with the crista front ofthe middle lacerate foramen is a tiny parotica of the petrosal. Closure of the pos- Vidian foramen for the nerves of the ptery- terolateral wall ofthe ectotympanic chamber goid (Vidian) canal. is accomplished by a prominent septum bullae. The bilaminar nature of the septum EcroTyMPANIc formed by both ectotympanic and caudal entotympanic contributions is evident in figure Relations of the chambered ectotympanic 5A where a thin entotympanic was applied are demonstrated in figures 3B, 5A, B. In to the ectotympanic surface. figure 3B the posterior half of the right ec- The dorsal edge of the septum bullae is totympanic in an adult is missing, thus it is produced into a flange that rests on the ven- possible to observe the relation of the ante- tral surface of the petrosal promontorium. rior crus and limb ofectotympanic where they The thickened flange makes a strong pro- contact squamosal and alisphenoid bones. montorial contact (fig. SB) that produces a Note the inflated volume of the anterior facet on the surface of the petrosal immedi- chamber of the auditory bulla formed by ec- ately anterior to the round window. The lin- totympanic, in contact with the small rostral ear contact of the septum with the promon- entotympanic and petrosal, and the location torium is an aeluroid trait (Hunt, 1989), and ofthe suprameatal fossa outside the tympan- is the result of the spatial relationship of the ic cavity proper. The crista tympanica is ectotympanic's posterior limb to the pro- slightly inclined inward from the parasagittal montorium in early ontogeny (see subsequent plane, indicating the near-vertical orienta- discussion). tion of the tympanum. The crista is in direct The close apposition of ectotympanic and linear continuity with the medial edge of the petrosal, together with the fusion ofectotym- suprameatal fossa, thus the fossa is properly panic to rostral entotympanic, segregate an situated to receive the pars flaccida of the anterior chamber of the auditory bulla from tympanum. a posterior chamber formed by caudal en- The medial rim ofectotympanic is in con- totympanic. The relative volume and dimen- tact with and fused to the rostral entotym- sions of the two chambers are illustrated in panic (thictic condition, Hunt, 1987). The the adult offigure SB. By analogy with living external auditory meatus ofthe ectotympanic viverrids of nearly identical bulla configu- is not laterally prolonged as a bony tube, thus ration, the posterior chamber of the Herpes- it conforms to the pattern found in living tides bulla grew forward into the anteroin- viverrids, and differs from that in most her- ternal corner of the auditory region (fig. SB), pestids and hyaenids in which a bony meatal migrating along the medial wall of the ecto- tube develops. tympanic, and thereby extended the length The complete ectotympanic bone (fig. 5) is ofthe septum bullae craniad. This pattern of known in two large individuals with basilar ontogenetic growth, in which caudal ento- skull lengths of 10-11 cm. In figure 5A the tympanic penetrates into the anterointernal 12 AMERICAN MUSEUM NOVITATES NO. 3023

corner of the auditory region, is typical of tube formed by the caudal entotympanic el- viverrids and felids, not hyaenids or herpes- ement. It travels anterolaterad within the en- tids. totympanic tube, which opens on the ventral apex of the promontorium (figs. 5B, 6C). CAUDAL ENTOTYMPANIc From this point the artery ascends the an- The caudal entotympanic gradually en- terior face of the promontorium (the pro- larges during ontogeny to produce a volu- montorium is slightly grooved), travels for- minous posterior chamber of the auditory ward along the lateral face of rostral bulla (figs. 5B, 6). This process ofgrowth and entotympanic, and then descends to enter the remodeling causes the caudal entotympanic middle lacerate foramen in the basisphenoid to impinge upon and indent the adjacent sur- (figs. 4B, 7B). of the This U-shaped arterial course and its re- faces basioccipital, exoccipital, mas- lationship to surrounding bullar and basicra- toid, and petrosal. The floor and sidewalls of nial bones are typical of viverrids caudal entotympanic expand to form a large (fig. 7). In hypotympanic cavity. As growth progresses, figure 7A the path of the internal carotid the ventral process of the petrosal (V) be- through the auditory region ofthe bush civet comes enclosed and hidden from view be- Civettictis civetta exemplifies the viverrid tween the caudal entotympanic and basioc- condition (Hunt, 1987, 1989). In Herpestides cipital (fig. 5B), and hence is not visible in antiquus the artery follows the same course figure 6C (left auditory region). (fig. 7B), and manifests the same relation- The external form of the caudal entotym- ships to surrounding basicranial structures. panic is typical ofviverrids (fig. 6). The prin- This arterial pattern distinguishes Herpes- cipal axis of inflation runs from the antero- tides from living and fossil herpestids in which medial to posterolateral surface in ventral the internal carotid follows a straight antero- view. The direction of ontogenetic growth is posterior course within the medial wall ofthe forward into the anterointernal corner of the auditory bulla (perbullar course, Hunt, 1989: auditory region (fig. 6A, C). It is particularly fig. 4). The herpestid character state is de- noteworthy that overgrowth of the ectotym- rived, a synapomorphy uniting the genera of panic by the caudal entotympanic takes place that family. The construction ofthe petrosal, in the same manner as observed in living auditory bulla, and surrounding basicranial viverrids (fig. 6A, B). This degree ofanterior architecture in Herpestides thus differs from penetration by caudal entotympanic, and its the basicranial morphology ofliving and fos- overgrowth ofectotympanic, is not observed sil herpestids, but conforms closely to that of in any herpestid or hyaenid. In hyaenids and the Viverridae. herpestids the caudal entotympanic is restricted to the posterior part of the auditory COMPARATIVE MORPHOLOGY OF region. THE AELUROID MIDDLE EAR INTERNAL CAROTID ARTERY In order to adequately evaluate basicranial morphology in Herpestides, it is necessary to The path of the internal carotid artery as describe and analyze two particularly diag- it travels through the auditory region is for- nostic attributes ofthe auditory region in the tuitously preserved in several Herpestides living aeluroid families: (a) the ontogeny of crania in which the bulla has been broken to the auditory bulla, especially the relationship reveal the internal anatomy ofthe middle ear between ectotympanic and petrosal in early cavity. In the adult (fig. 6A, C) the artery ontogeny, and (b) the stage of evolution of enters the auditory region behind the ventral the ossicular chain, as demonstrated by the process of the petrosal, immediately lateral form and orientation ofthe auditory ossicles. to the prominent ridge on the basioccipital In the first case, the relationship ofectotym- for attachment of the rectus capitis ventralis panic to petrosal promontorium determines muscle. The artery passes between the lateral the geometry of the anterior bulla chamber, edge ofbasioccipital and the caudal entotym- and also appears to influence the final struc- panic, and is enclosed in a conspicuous bony tural relationship between caudal entotym- 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES 13

Fig. 5. Relationship of ectotympanic to petrosal in Herpestides: A, posterolateral view of right auditory region (MGL St.-G. 3067) of large individual; caudal entotympanic, basi- and exoccipital removed; white triangles indicate line of attachment of caudal entotympanic to petrosal and mastoid; B, posterolateral view ofleft auditory region (MGL St.-G. 3066, see also fig. 1) ofanother large individual in which the dorsal part of caudal entotympanic is in place. Note contact between the ectotympanic flange and the promontoriumjust anterior to the round window in both individuals; this contact produces a small characteristic facet on the promontorium. Stereopairs. panic and ectotympanic. In the second, the herpestid auditory patterns are sharply de- morphology ofthe auditory ossicles ofviver- fined, and Herpestides, in its adult morphol- rids and felids reflects a more primitive stage ogy and in what can be determined of its of evolution than the ossicles of herpestids auditory ontogeny, is identifiable as a true and hyaenids. As a result, the viverrid and viverrid. B

Fig. 6. The auditory bulla of Herpestides (MGL St.-G. 3065): A, high lateral oblique view of basicranium, showing intact left bulla (black triangles indicate the line of fusion between ectotympanic and caudal entotympanic), and broken right bulla opened to show internal structure ofthe posterior chamber, including course of internal carotid artery (dashed line) as it enters the middle ear; B, lateral view of auditory bulla showing inflation of caudal entotympanic relative to ectotympanic as in viverrids; C, posteroventral view of the basicranium, demonstrating penetration of caudal entotympanic into the anterointernal corner of the auditory region. Stereopairs. 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES 15

Fig. 7. Comparison of the auditory regions of (A) the viverrid Civettictis civetta (UNSM 14114), Kibwezi, Kenya, and (B) Herpestides antiquus (NMB 11583), Montaigu-le-Blin, France. Dashed lines in A and B indicate the path of the internal carotid artery from its point of entrance into the auditory region until it enters the cranial cavity at the middle lacerate foramen. Stereopairs.

The course of ontogenetic development in early ontogenetic stages from a representative the auditory region of aeluroids is perhaps viverrid and herpestid in order to explore the best understood in terms of the different on- developmental basis for these differences. togenetic pathways adopted by viverrids and Because the felid pattern of bulla devel- herpestids: I intend to describe a sequence of opment is nearly identical to that of viver- 16 AMERICAN MUSEUM NOVITATES NO. 3023 rids, and has been previously described at to the edge of the squamosal, directly lateral length (Hunt, 1974, 1987, 1989), I will not to the gonial process ofthe malleus, whereas review it here. The pattern of felid bullar on- the posterior crus simply overlaps its lateral togeny differs from that of viverrids only in surface at the level of the posttympanic pro- the more elaborate inflation ofthe caudal en- cess of the squamosal. totympanic and its diagnostic encroachment Because the ectotympanic is essentially on surrounding basicranial bones. As part of horizontal, its anterior limb is pressed against this process, the felid caudal entotympanic the roof of the tympanic cavity, which is penetrates into the anterointernal corner of formed by the epitympanic wings ofbasi- and the auditory region, inserting itself between alisphenoid. Its posterior limb is applied to ectotympanic and rostral entotympanic (bra- the petrosal promontorium slightly posterior dynothictic state, Hunt, 1987); this insertion to the round window. At this stage, the pos- does not occur in viverrids. In the Aquitanian terior part of the tympanic cavity is closed sediments of the Allier basin, both viverrid by a sheet of connective tissue attached to and felid patterns can be recognized. Com- the posterior margin of ectotympanic, and parison of these contemporaneous forms in- extending from that margin to the edges of dicates that Herpestides is not an early felid. the basioccipital, exoccipital, and mastoid Insight into the development of the hyae- bones. nid bulla and ossicular chain is limited by The malleus and incus within the epitym- lack ofknowledge ofearly ontogenetic stages. panic recess are relatively large and already However, following discussion ofviverrid and ossified, and the posterior process ofthe incus herpestid morphologies, I review what is and anterior process ofthe malleus thatjoint- known of the hyaenid auditory region, and ly determine the axis of rotation are devel- argue that there is sufficient knowledge of oped (the rotational axis of the carnivoran hyaenid bulla and ossicular structure to con- malleus-incus complex is a straight line drawn fidently exclude Herpestides from the Hy- through the posterior process ofthe incus and aenidae. the tip ofthe anterior process ofthe malleus). In this and all subsequent developmental BuLLA ONTOGENY IN HERPESTIDAE stages, even in adults, a line drawn parallel to the manubrium of the malleus intersects In an ontogenetic series of crania of Her- the rotational axis of the malleus-incus at a pestes (Xenogale) naso from various localities high angle. Furthermore, the axis ofossicular in Zaire, a progression from an incipient stage rotation is nearly horizontal (here defined as of bulla formation to the fully formed adult a line parallel to the basicranial axis), and is bulla defines the herpestid pattern of devel- not markedly inclined. The tensor tympani opment. fossa at this stage remains open and has no The earliest available ontogenetic stage is bony floor. a skull (AMNH 51092, Niangara, Zaire) with In the next available stage ofdevelopment, basilar length of 30.9 mm in which the tips a skull (AMNH 51093, basilar length, 42.7 ofthe principal cusps ofDP3-4 arejust form- mm) from Faradje, Zaire, the DP3-4 crowns ing in the dental crypts, and the only ossified are fully formed and partially erupted, and bulla element is the ectotympanic (fig. 8C). the three separate bulla elements are present The ectotympanic crescent lies in the hori- as unfused ossifications. The ectotympanic zontal or frontal plane parallel to the basi- crescent has widened considerably, especially cranial axis. Its anterior and posterior crura in its medial part, tilting the crescent (and are attached to the squamosal. The ventral eardrum) slightly laterad. Although the ec- edge ofthe squamosal between the two crura totympanic's posterior limb rests on the pos- is bent inward as a lamina forming a floor terior slope of the promontorium, the pro- for a large epitympanic recess. The recess montorium remains largely within the contains the robust heads of malleus and in- circumference of the ectotympanic (hence cus (a deeply pocketed epitympanic recess within the bulla's anterior chamber). The ro- within the squamosal is typical ofherpestids). tational axis ofthe auditory ossicles does not The anterior ectotympanic crus is anchored alter its alignment, and remains nearly per- 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES 17 pendicular to the manubrium ofthe malleus. to the anterior chamber, however the direc- At this stage the tensor tympani fossa now is tion ofinflation is primarily ventrad, and the largely encapsulated by bone, and the exter- caudal entotympanic does not penetrate the nal auditory meatus between the ectotym- anterior part ofthe auditory region as it does panic crura has begun to close by progressive in viverrids and felids. In a number of her- ossification of the rim of the aperture. Ap- pestid species the ectotympanic crescent and pressed against the posterior margin of ec- eardrum remain nearly horizontal; in other totympanic and the posterior surface of the species such as Herpestes auropunctatus, the promontorium is a small caudal entotym- ectotympanic plane rotates outward, but the panic chamber, only weakly ossified and pronounced anterolateral tilt found in viver- slightly inflated, enclosing a volume that is rids (and felids) never occurs. perhaps half that of the anterior chamber. At this early stage of development the entrance of the internal carotid artery into the BULLA ONTOGENY IN VIVERRIDAE middle ear cavity is already blocked by os- Using an ontogenetic series of skulls of sification ofthe bulla wall. Subsequently, the Viverricula indica, it is possible to examine artery becomes isolated within a bony tube equivalent stages ofbasicranial development in the medial wall ofthe bulla (perbullar state, in a typical viverrid (fig. 8A), and identify Hunt, 1987), a condition found in all living differences relative to herpestid ontogeny. At herpestids but in no other living aeluroids. the earliest available stage, the skull (AMNH A slightly more advanced stage ofauditory 59935, Hainan, China, basilar length 38.2 development (fig. 8B) is present in two skulls mm) shows the tips of DP3-4 cusps just (AMNH 51609, basilar length, 46.1 mm; forming in the alveolar crypts. The only os- AMNH 51089, basilar length, 45.2 mm) from sified bulla element is the ectotympanic. The Medje, Zaire, in which DP3-4 are erupted to ectotympanic crescent does not lie in the hor- a somewhat greater degree than these same izontal (or frontal) plane parallel to the ba- teeth in the 42.7 mm stage. The bulla shows sicranial surface as in herpestids. The plane the initiation of caudal entotympanic infla- ofthe ectotympanic is tilted forward and out- tion, enlarging in ventral and medial direc- ward (anterolaterad) relative to its orienta- tions, whereas the ectotympanic is about the tion in herpestids, the result ofdisplacement same size. The amount of inclination of the ofthe ectotympanic's posterior limb by ven- ectotympanic crescent and eardrum remain tral protrusion of an enlarged petrosal pro- the same, but closure ofthe external auditory montorium. Because the ectotympanic cres- meatus by bone has progressed, and this pro- cent is unable to fully surround the enlarged cess continues during ontogeny until the mea- petrosal, its posterior limb is captured by the tal aperture is completely filled (or nearly so) ventrally extended promontorium. This dis- by bone in the adult. This new bone forms a placement causes the posterior limb to lie ventral floor below the eardrum. The rostral across the apex ofthe promontorium, slightly entotympanic has increased in size, following anterior to the round window. As a result, a its initial appearance between the 30.9 and substantial part ofthe promontorium lies be- 42.7 mm stages discussed above. hind the posterior limb, where it is covered Dissection of the bulla reveals a bilaminar by connective tissue forming the posterior septum bullae that will fuse into a single par- wall of the tympanic cavity (in herpestids, tition in the adult. The plane formed by the more ofthe promontorium is enclosed by the septum bullae at the juncture ofectotympan- ectotympanic). This connective tissue sheet ic and caudal entotympanic is somewhat pos- is attached at the posterior margin of ecto- teriorly inclined (see Hunt, 1989: fig. 5). The tympanic and extends to the mastoid and oc- caudal entotympanic chamber ofthe bulla is cipital bones just as in herpestids. confined to the posterior part ofthe auditory In a later developmental stage (AMNH region behind the promontorium and it re- 60065, Hainan, China, basilar length 51.8 mains so in the adult. mm) in which DP3-4 are formed but un- In adult herpestids the caudal entotym- erupted, the ectotympanic has been enlarged panic chamber continues to inflate relative by the addition ofbone to its medial margin, Fig. 8. Petrosal-ectotympanic relationship of a viverrid and herpestid in early ontogeny. A, Basi- cranium in ventral view of viverrid Paradoxurus hermaphroditus (AMNH 59933, female neonate, Hainan, China). The caudal entotympanic has been removed from the right auditory region to demonstrate the petrosal-ectotympanic relationship: note large amount of petrosal exposure posterior to ectotympanic (compare fig. 8B); B, Basicranium in ventral view of herpestid Herpestes (Xenogale) naso (AMNH 51609, male neonate, Medje, Zaire). The caudal entotympanic has been opened on the left side to show petrosal nearly completely overlapped by ectotympanic: there is almost no petrosal exposure posterior to ectotympanic (compare A); C, Basicranium in ventral view ofherpestid Herpestes (Xenogale) naso (AMNH 51092, male neonate, Niangara, Zaire). The ectotympanic nearly encompasses the petrosal promontorium at this very early developmental stage in herpestids. 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES 19

creating an anterior bulla chamber. A small dition in which the long axis of the manu- but well-defined rostral entotympanic ossi- brium maintains a high angle (>600) relative fication has appeared on the anterior slope of to the rotational axis. This difference between the promontorium, and has made contact the neonatal mallei of herpestids and viver- with the medial rim of ectotympanic by a rids persists in the adults, and its origin and narrow isthmus ofbone. Ossification of cau- explanation are of considerable interest. dal entotympanic has progressed about three- We can determine that the "bent" viverrid quarters of the distance from the rear of the manubrium does not result from the change bulla element toward the anterointernal cor- in orientation ofthe ectotympanic during on- ner, migrating along the tympanic membrane togeny because, in prenatal stages in which medial to ectotympanic. The ossification front the ectotympanic lies nearly in the horizontal is a smooth convex margin that has reached plane, the manubrium has already adopted a point just anterior to the ventral process of its alignment relative to the rotational axis. the promontorium. In front of this margin In fact, the "bent" manubrium is plesiomor- (and posterior to the ectotympanic-rostral phic for aeluroids, and probably for Carniv- entotympanic complex), a small portion of ora (see below); it is an attribute ofthe prim- the tympanic membrane remains unossified. itive therian malleus. Two skulls (AMNH 45504, Fukien, China; Furthermore, in viverrids the rotational AMNH 107605, Bali; basilar lengths about axis of the malleus-incus complex is tilted 52-54 mm), with DP3-4 nearly fully erupted, slightly upward (i.e., the attachment of the indicate a slightly later stage in which the incus to the posterior wall ofthe epitympanic three ossified bulla elements have joined, but recess is dorsal to the point of attachment of fusion is only incipient, so that the bounda- the anterior process ofthe malleus), deviating ries of each element remain evident. There from the horizontal to a greater degree than is no significant change in orientation or in- the herpestid axis of rotation. flation ofthe ectotympanic or anterior cham- Felids display the same ectotympanic-cau- ber, and the posterior limb of ectotympanic dal entotympanic growth pattern as viverrids remains in contact with the surface of the during ontogeny, including displacement of promontorium anterior to the round win- the posterior limb of ectotympanic by an en- dow. As a consequence, the posterior part of larged promontorium and, as a result, appli- the enlarged promontorium containing the cation of the posterior ectotympanic limb to round window continues to protrude into the the promontorium anterior to the round win- posterior chamber of the bulla (fig. 8A). dow (fig. 9). Felids and viverrids also share The caudal entotympanic has completed the "bent" orientation ofthe manubrium, and its anterior growth and is entirely ossified. Its the slightly tilted ossicular rotational axis. In rounded anterior margin has made contact these aspects of bulla ontogeny, and ossicle and fused with rostral entotympanic and the form and orientation, the two families are medial edge of ectotympanic, and its degree extremely similar. These features were con- of inflation is visibly greater than in the pre- firmed in an ontogenetic series of the do- vious developmental stage. The application mestic cat, and also in a comparable series ofcaudal entotympanic to the posteromedial ofthe African wild cat, Felis silvestris, and in rim of ectotympanic has produced a bilam- juveniles and adults ofalmost all living felids. inar septum bullae about 1.5 mm in height. In all developmental stages of Viverricula indica, the manubrium ofthe malleus is "bent BULLA ONTOGENY IN HYAENIDAE forward" (strongly inclined anteromesad), The hyaenid auditory region is distin- placing the long axis of the manubrium at a guished by a unique ontogenetic pattern of very low angle relative to the rotational axis bulla development in which the ectotympan- of the auditory ossicles (I refer to the incli- ic contributes an enlarged anterior bulla nation of the manubrium in the plane of the chamber that extends backward, ventral to ossicular rotational axis in lateral view, not the more confined posterior chamber formed to medial deviation ofthe manubrium which by caudal entotympanic (Hunt, 1974, 1987, can occur independently in mammals). This 1989). The earliest known hyaenid basicrania arrangement occurs in other viverrids and in from the mid-Miocene of Asia already have felids, but contrasts with the herpestid con- developed this bulla configuration (Qiu et al., 20 AMERICAN MUSEUM NOVITATES NO. 3023

Fig. 9. Petrosal-ectotympanic relationship of a felid in early ontogeny. A, Basicranium in ventral view of newborn domestic cat Felis, demonstrating the large portion of the petrosal promontorium posterior to the ectotympanic, as in viverrids (compare fig. 8A); B, same specimen as A, medial view ofauditory bulla, showing the considerable exposure ofthe promontorium posterior to the ectotympanic. Black triangles indicate ectotympanic-caudal entotympanic suture.

1988; Hunt, 1989). I know of no represen- onstrate that the fundamental hyaenid bulla tative ontogenetic series of the hyaenid basi- pattern was already developed in the earliest cranium; however, the few juvenile stages of ontogenetic stages presently identified (e.g., living hyaenids that have been studied dem- Hunt, 1974: fig. 35). It seems likely that the 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES 21 A B C

incus

transverse

Iectotympanic Fig. 10. Evolution of the malleus-incus complex in therian mammals: A, primitive therian; B, transitional stage; C, derived. Diagrams depict right ossicular complex in lateral view, dorsal to top, anterior to right. Key trends include change in inclination of rotational axis, reduction of malleal attachment to ectotympanic, reorientation ofthe manubrium, and relative size increase ofthe incus. Living aeluroid carnivorans belong to stages B and C. hyaenid bulla ossifies in the neonatal animal tary bones into the auditory region has been in much the same configuration seen in young much better understood in recent years (Al- juveniles. Even in the aberrant hyaenid Pro- lin, 1975; Kermack and Mussett, 1983). The teles, the basic morphology of the hyaenid orientation and probable functional relations bulla can still be recognized, despite the fail- ofthe angular (ectotympanic), articular (mal- ure of the anterior chamber to extend back- leus), quadrate (incus), and stapes in primi- ward under the posterior chamber (Hunt, tive therians have been determined both from 1974). fossils and from anatomical studies of living Hyaenids share with viverrids and felids a eutherians and metatherians. Fleischer (1973, "bent" manubrium and an inclined ossicular 1978) has proposed a series of stages in the rotational axis. However, hyaenids and her- evolution of the auditory ossicles and ecto- pestids lack the prominent pocketed anterior tympanic that progresses from a primitive lamina ofthe malleus found in viverrids and therian morphotype (Ausgangstyp) via a felids that connects the neck of the malleus transitional arrangement (Ubergangstyp) to a with the anterior process. The anterior lam- final derived condition in which the ossicles ina ofherpestids and hyaenids is strongly re- are freely mobile within the middle ear space duced. (fig. 10). All these stages can be found among living therian mammals, and their functional AUDITORY OSSICLES IN and anatomical attributes have been the ob- AELUROIDS ject of extensive investigation (Fleischer, 1978; Hunt and Korth, 1980). OSSICULAR ORIENTATION IN MAMMALS From my observations ofeutheres and me- Ossicular patterns contribute an additional tatheres, and those of Segall (1969, 1970, dimension to an understanding of auditory 1971), it is possible to identify three prop- evolution in Carnivora. In living aeluroid erties of the ossicular chain that change carnivorans, morphology and orientation of through time. Each ofthese properties seems the auditory ossicles are closely correlated to be able to evolve independently ofthe oth- with bulla morphology, particularly with ec- ers: totympanic placement. This information al- (1) Altered orientation ofthe rotational axis lows a reasonable prediction of the mor- of malleus-incus. - The plesiomorphic ori- phology and orientation of the auditory entation of the ossicular rotational axis in ossicles in Herpestides antiquus, and also sup- early mammals is slightly tilted or inclined plies evidence for a hypothesis describing the (fig. 10A), having an approximately antero- evolution ofthe ossicular system in aeluroids. posterior alignment (Allin, 1975: pl. 3, fig. Incorporation ofthe mammalian postden- 10; Segall, 1969: fig. 1; Fleischer, 1973: fig. 22 AMERICAN MUSEUM NOVITATES NO. 3023

58B). The axis passes forward and downward to ectotympanic, and a well-developed trans- from the point at which the incus (quadrate) verse lamina (fig. lOA); both the process and contacts the periotic bone, thence through the the lamina become much reduced in deriv- gonial process ofthe malleus (articular) along ative eutherian mallei (fig. lOB, C). its line of attachment to the ectotympanic (3) Realignment ofthe manubrium relative (angular). This primitive orientation of the to the axis ofrotation. - Fleischer (1978) in- ear ossicles is seen in marsupials such as Di- dicated that the long axis of the manubrium delphis and Caenolestes, where the axis of can alter its alignment relative to the rota- rotation makes an angle of about 20 to 230 tional axis of the malleus-incus complex. In with the horizontal (Segall, 1969), and is also primitive therians the manubrium is gener- known in placentals with primitive ossicular ally aligned parallel to the rotational axis (fig. configurations, where one finds similar lOA). In many modem lineages with freely amounts of axial inclination, ranging from mobile ossicular chains, the long axis of the about 20 to 420 (Segall, 1971). In living mar- manubrium is nearly perpendicular to the ro- supials and placentals, the rotational axis ei- tational axis (fig. lOC), a derived character ther is maintained in an inclined orientation, state in therian mammals. or has transformed by rotation into a more Thus we can identify plesiomorphic and horizontal alignment (fig. lOB, C). Thus an derived morphological states of the auditory inclined axis of rotation reflects a more ple- ossicles involving orientation ofthe rotation- siomorphic ossicular alignment relative to al axis, nature of the attachment of malleus derived states that approach or attain the hor- to ectotympanic, and orientation of the ma- izontal. nubrium. We can apply this knowledge to the (2) Loss ofmalleal attachment to ectotym- ossicular morphology and orientation of the panic. - The plesiomorphic mammalian os- living aeluroids. As a result, we find that ossicular chain (fig. 1 OA) incorporates a large sicular patterns are apparent at the family U-shaped malleus (one leg formed by the ma- level, and that both plesiomorphic and de- nubrium, the other by the gonial process) rived conditions exist among aeluroids that firmly attached to the ectotympanic, and a can be used to constrain evolutionary hy- relatively small incus (Allin, 1975: pl. 3, figs. potheses. 10, 11; Fleischer, 1978: figs. 1, 8). This type, or a close approximation, is known in living marsupials, monotremes, and placentals, and OSSICULAR ORIENTATION AND its ontogenetic development (for example, in MORPHOLOGY IN AELUROIDS certain marsupials) shows a clear correspon- Viverrids and felids have an ossicular ori- dence to the postdentary bones of cynodont entation and geometry that reflect a transi- reptiles and early mammals (Kermack and tional stage of ossicular development (fig. Mussett, 1983). An ossicular chain of this lOB). The axis of rotation is tilted forward kind exhibits considerable torsional stiffhess and downward (fig. 1 IC, D), inclined 25 to due to the plesiomorphic bony fusion ofmal- 300 from the horizontal (determined as a line leus to ectotympanic. As the ossicular ap- drawn parallel to the basicranial axis). Thus paratus evolves from primitive (fig. 1 OA) to the viverrid-felid alignment approximates the more advanced states (fig. lOB, C), the mal- axial orientation found in a living therian such leus often becomes more loosely attached to as Didelphis. The manubrium is oriented at ectotympanic, and as a result the degree of an angle of about 15 to 250 relative to the stiffhess is proportionately reduced. This ossicular rotational axis (fig. 12C, D), hence looser attachment is achieved by reduction similar to primitive manubrial alignments ofthe bony process anchoring the malleus to that parallel this axis, and unlike derived ma- ectotympanic to either a ligamentous con- nubria that lie nearly perpendicular to the nection, or if osseous, one that is very thin axis. and flexible (living aeluroid carnivorans all The malleus ofviverrids and felids is firmly maintain a thin, flexible, bony attachment). yet delicately attached by thin, flexible bone Thus a primitive eutherian malleus will have to the ectotympanic in both neonates and a prominent gonial process firmly attached adults. The lamina of the malleus is well de- 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES 23 veloped, and its lateral face is characteristi- adults), the lamina is strongly reduced as in cally pocketed (fig. 12C, D); viverrids and herpestids, and only a vestige ofthe pocketed felids preserve this plesiomorphic configu- lamina remains (figs. 1 2A, 1 3B). It is evident ration of the lamina, whereas the lamina is that hyaenids have not passed through a her- strongly reduced or lost in herpestids and pestid grade of ossicular orientation. The in- hyaenids (figs. 1 2A, B). The pocketed lamina clined rotational axis ofhyaenids seems more is a vestige of the well-developed transverse plesiomorphic than the nearly horizontal axis lamina of the malleus found in primitive of herpestids, but both groups have under- therians (fig. IOA). gone significant reduction of the pocketed Alignment and morphology ofthe ossicles lamina of the malleus. in viverrids and felids thus appear to differ The fact that viverrids and felids share near- from a primitive eutherian ossicular pattern identity in their ossicular structure, whereas (fig. 13) only in (a) reduction of the gonial herpestids are much different and arguably process of the malleus to produce a more more derived, does not necessarily substan- flexible bony attachment (the anterior pro- tiate the sister-group relationship that I have cess, fig. 12C, D) ofmalleus to ectotympanic; proposed for viverrids and felids on the basis (b) a slightly greater deviation ofthe long axis of bulla ontogeny (Hunt, 1987). The orien- of the manubrium from the axis of ossicular tation ofthe ossicular rotational axis and the rotation; (c) size increase of incus relative to form of the malleus in viverrids and felids malleus. There also may have been a small are plesiomorphic (fig. 13), and could simply (but uncertain) amount of declination of the reflect the absence of significant change in rotational axis toward the horizontal, based these features over time. However, the fact on the supposition that the primitive therian that viverrids and felids share the enlarged axis was more steeply inclined, but there is petrosal promontorium and the concomitant no reason why the present inclination in vi- tilting of the ectotympanic, correlated with verrids and felids might not closely approx- the forward-growing caudal entotympanic el- imate the primitive state for carnivorans. ement, constitutes more reliable grounds for Herpestids on the other hand have an os- the proposed sister relationship between these sicular arrangement similar to the freely mo- two families. Herpestids stand apart from this bile type of Fleischer (fig. IOC). The axis of viverrid-felid dichotomy because oftheir de- rotation is anteroposteriorally aligned with- rived bulla ontogeny and their ossicular ori- out significant deviation from the horizontal entation and morphology: they cannot be an- (figs. l11B, 13E). The manubrium is more cestral in their present form to any other group nearly perpendicular to the axis of rotation of living aeluroids. Living hyaenids are sim- (600 to 800, fig. 1 2B). The malleus is delicately ilarly derived in the form of their malleus, attached to the ectotympanic by a fragile, and so cannot be ancestral to viverrid and much reduced anterior process (a bony con- felid carnivorans in which the condition of nection, however, still persists in neonates the malleus is more plesiomorphic. However, and adults). The lamina is strongly reduced, the similarity ofhyaenid and herpestid mallei and only a minute vestige ofa pocket remains is not necessarily evidence of close relation- (fig. 1 2B). Herpestids appear to have evolved ship, because it could represent parallel evo- the most derived ossicular geometry and ori- lution, a result of the independent develop- entation among the living aeluroid carnivo- ment ofa more mobile ossicular chain in the rans (fig. 13). two groups. Hyaenids share aspects of both the viver- The close correlation between bulla mor- rid-felid and herpestid end-members in terms phology and ossicular geometry and orien- of ossicular form and axial orientation. tation in living aeluroids suggests a plausible Hyaenids retain an alignment of the rota- reconstruction of the ossicular chain in tional axis similar to that of viverrids and Herpestides antiquus. One can predict that felids, inclined approximately 33-370 from this Aquitanian viverrid possessed a slightly the horizontal (fig. 11A). However, the at- inclined rotational axis; the incus will be small tachment ofmalleus to ectotympanic is quite relative to the size of the malleus; the ma- delicate (it is a thin osseous connection in nubrium ofthe malleus will lie at a low angle 24 AMERICAN MUSEUM NOVITATES NO. 3023

BASICRANIAL AXIS A

330 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES 25 B

A______\ BASICRANIAL AXIS

270O

Fig. 11. Inclination of the ossicular rotational axis within the middle ear of aeluroid camivorans, measured as the angle ofintersection with the basicranial axis (an estimate ofthe horizontal). A, Hyaenid Hyaena brunnea (UNSM 16442); B, herpestid Herpestes auropunctatus (AMNH 232811); C, viverrid Genetta sp. (UNSM 14275); D, felid Lynx rufus (UNSM 14674). Lateral view, ventral to top, anterior to left (hyaenid in ventrolateral view). 26 AMERICAN MUSEUM NOVITATES NO. 3023

,2mm A B

manubrium Af 2mm D XMALLEUS

) , t ~~~short crus anterior process vA 19 AXIS OF ROTATION pocketed laminaS

Fig. 12. Orientation ofthe manubrium relative to the ossicular rotational axis in aeluroid carnivorans (measured as the angle between the long axis of the manubrium [m] and the rotational axis [ar]). A, Hyaenid Hyaena brunnea (UNSM 16442); B, herpestid Mungos mungo (AMNH 118859); C, viverrid Genetta sp. (UNSM 14275); D, felid Panthera tigris (UNSM 15484). Lateral view, ventral to top, anterior to left. Viverrids and felids retain a plesiomorphic form of the malleus in which the pocketed lamina is retained as a vestige of the prominent transverse lamina of early therians. The lamina is reduced in hyaenids and nearly absent in herpestids.

to the axis of ossicular rotation; the malleus is readily comparable to basicranial patterns will retain a well-developed lamina, probably found in living viverrids. Among the features pocketed as in living viverrids; the connec- common to Herpestides and viverrids are: (1) tion between the malleus and ectotympanic petrosal form and orientation; (2) configu- will be osseous, not ligamentous. Viverrids, ration and ontogenetic growth pattern of the among living aeluroids, retain the most ple- caudal entotympanic relative to other bulla siomorphic ossicular chain, one which is elements and surrounding basicranial bones; markedly different from the derived ossicular (3) form, size, and structural relations of the morphology found in herpestids. three bulla elements (rostral and caudal entotympanics, ectotympanic); (4) path of the PHYLOGENETIC AFFINITIES OF internal carotid artery in the auditory region. HERPESTIDES Herpestides has been considered a herpestid (Viret, 1929) or ancestral hyaenid (Beau- Basicrania attributed to Herpestides in the mont, 1967; Beaumont and Mein, 1972; Basel and Lyon collections, including several Hunt, 1989) in earlier studies. Petter (1974) belonging to complete skulls with associated believed that the auditory bulla was of the dentitions, all invariably possess a common herpestid type. My opinion, initially based morphology of the auditory region. Both in- on illustrations in the European literature, tact and crushed skulls document the asso- changed when I was able to examine the fos- ciation ofthese basicrania (and auditory bul- sils. The preceding review of herpestid and lae) with dentitions undoubtedly referable to hyaenid auditory regions points to the ab- Herpestides antiquus (figs. 14, 15). sence of any distinguishing features in the The basicranial morphology ofHerpestides Aquitanian aeluroid that would ally it with 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES 27

HORIZONTAL B

HORIZONTAL D

Fig. 13. Evolution of the ossicular complex in aeluroid Carnivora (B through E each presumed to be independently evolved from A). A, Early therian with marked axial inclination and plesiomorphic ossicle morphology (developed transverse lamina ofmalleus, small incus relative to malleus); B, hyaenid with moderate inclination and derived ossicle morphology (reduced lamina of malleus, incus enlarged relative to malleus); C, D, viverrid and felid with moderate inclination and plesiomorphic ossicle morphology (pocketed lamina ofmalleus, incus small); E, herpestid with low inclination and derived ossicle morphology (strongly reduced lamina, incus large relative to malleus). Lateral view of right middle ear, dorsal to top, anterior to right. either of these families. Here I briefly sum- pani fossa is unknown in living viverrids, fe- marize the important distinctions between lids, and hyaenids, and is considered a de- Herpestides and herpestids/hyaenids. rived trait of herpestids on the basis of Herpestids differ in their auditory structure outgroup comparison with other carnivo- from viverrids in having a differently shaped rans. An encapsulated tensor tympani fossa petrosal, characterized by a particularly is not found in Herpestides. prominent bony capsule anterior to the pars The caudal entotympanic forms the pos- cochlearis that houses the tensor tympani terior chamber of the auditory bulla in ae- muscle. The muscle is nearly encapsulated by luroid carnivorans. In herpestids, the caudal bone in adults (a trait unknown in other ae- entotympanic remains restricted to the pos- luroids), and emerges through a restricted, terior part of the auditory region during on- laterally directed aperture about the same di- togenetic growth. In living herpestid species, ameter as the round window (see Hunt, 1989: inflation of the posterior (caudal entotym- fig. 4B). This construction of the tensor tympanic) chamber takes place during ontogeny 28 AMERICAN MUSEUM NOVITATES NO. 3023

Fig. 14. Association of Herpestides antiquus dentitions with typical viverrid basicrania occurs in a number of skulls from the Aquitanian of the Allier basin, France (NMB 11407, Montaigu-le-Blin). but is primarily directed ventrad and to the result the posterior limb of ectotympanic sides, never forward. However, in contrast to comes in contact with the apex of the en- the orientation in herpestids, the caudal enlarged promontorium (just anterior to the totympanic chamber grows forward in viver- round window), causing the ectotympanic to rids and felids, penetrating into the anteroin- tilt forward and outward so that it initiates ternal corner of the auditory region, often its growth from a position much different than overgrowing the anterior chamber ofthe bul- that found in herpestids. The orientation of la. Herpestides shows a pattern of caudal en- the ectotympanic in both juvenile and adult totympanic growth comparable to that of Herpestides demonstrates a viverrid-like, viverrids. No living or fossil herpestid has a rather than herpestid-like, ectotympanic-pe- bulla pattern even remotely similar to that trosal relationship. In Herpestides the ecto- seen in Herpestides. tympanic has been tilted forward and out- The placement of the ectotympanic, and ward by an enlarged promontorium, and the its relation to the petrosal, also distinguishes posterior limb ofectotympanic, unable to en- herpestids. Because the petrosal promonto- close the promontorium, is applied to the rium is relatively small and almost fully en- petrosal surface anterior to the round win- closed (more so in some species than in oth- dow. ers) within the perimeter ofthe ectotympanic A hallmark ofthe herpestid auditory region crescent in early developmental stages, the is an ectotympanic oriented nearly in a fron- herpestid ectotympanic lies flat against the tal plane in juveniles and even in adults (ec- basicranium in the horizontal plane. How- totympanic inflation causes slight outward ever, in viverrids and felids the ectotympanic rotation in some species). As a consequence crescent fails to enclose the promontorium the herpestid tympanum faces nearly directly as completely as in herpestids, apparently be- ventrad, and were it to remain so in the adult, cause the promontorium is quite large and the anterior bulla chamber would be floored protrudes below the basicranial surface. As a by the eardrum. To avoid this, herpestids 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES 29

Fig. 15. The dentition of Herpestides antiquus has been attributed by Beaumont (1967) to a single, highly variable species. The upper teeth, although more plesiomorphic, show similarities in occlusal pattern to those of the African bush civet, Civettictis civetta, which may be a lineal descendant. close the auditory meatus of the ectotym- gion. Herpestides demonstrates the viverrid panic with bone, thereby creating an osseous caudal entotympanic configuration, not the floor beneath the eardrum. In viverrids and herpestid. felids the more parasagittal orientation ofthe Enclosure of the internal carotid artery in ectotympanic, both in early developmental the medial bulla wall (perbullar course) in all stages and as a further result ofectotympanic herpestids is a unique synapomorphy of the inflation, does not result in closure of the family not found in other aeluroids. The ar- auditory meatus by bone; rather, the meatus tery in Herpestides does not follow such a remains open as a prominent aperture into course but instead takes a transpr6montorial adult life (Hunt, 1989: fig. 2). In bothjuvenile path very similar to that of viverrids such as and adult Herpestides, the bony external au- Civettictis (fig. 7). This transpromontorial ditory meatus is open and shows the viverrid course is probably plesiomorphic for Aelu- orientation; there is no evidence of the roidea (Hunt, 1989). uniquely derived herpestid condition of the The hyaenid auditory region differs fun- meatus. damentally from that of Herpestides in the It seems probable that the different ori- configuration of anterior and posterior bulla entation ofthe ectotympanic crescent relative chambers. In living and fossil hyaenids (ex- to the promontorium in herpestids and viver- cept Proteles) the bulla's anterior chamber, rids/felids is the basis for the different con- formed primarily by ectotympanic, extends figurations ofthe caudal entotympanic cham- backward to cover the posterior chamber. The ber in these two groups. The penetration of posterior chamber formed by caudal ento- the caudal entotympanic into the anterointer- tympanic remains small in volume relative nal corner ofthe auditory region in viverrids to the enormous anterior chamber. In fossil and felids is correlated with outward and for- hyaenids there is no evidence that the pos- ward displacement of ectotympanic in early terior chamber ofthe hyaenid bulla was ever ontogeny. Because the ectotympanic is not any larger than in living species, and the an- displaced in herpestids, the caudal entotym- tiquity of this hyaenid bulla pattern extends panic cannot penetrate into the anterointer- to very primitive forms (Tungurictis, Percro- nal auditory region and, as a result, it remains cuta) in the mid-Miocene. This hyaenid pat- within the posterior part of the auditory re- tern is the antithesis of the Herpestides pat- 30 AMERICAN MUSEUM NOVITATES NO. 3023 tern, in which the caudal entotympanic grows common ancestry, or is this a case ofparallel forward and becomes the dominant chamber. evolution? Hyaenids plausibly originate from an aelu- If one accepts the morphocline polarity roid morphotype in which the caudal ento- presented in figure 10 for the evolution ofthe tympanic was never enlarged and primarily auditory ossicular chain in eutherian mam- confined to the posterior auditory region. mals, then a malleus with a reduced lamina Because the dominant chamber of the represents a derived character state. The hy- hyaenid bulla corresponds to the subordinate pothetical morphocline of figure 10 is sup- chamber of the viverrid-felid bulla, and vice ported by the taxonomic distribution of versa, it is evident that these ontogenetic primitive and derived ossicular states among growth patterns represent divergent solutions living mammals: the presence of a malleus to the problem of middle ear enclosure. The with well-developed transverse lamina and hyaenid pattern and the pattern identified in gonial process occurs in all monotremes, juvenile and adult Herpestides lie on inde- nearly all metatherians, and in eutherians such pendent developmental trajectories. Thus a as tenrecs, erinaceids, soricids, solenodon- relationship of the Aquitanian aeluroid to tids, and bats; the configuration of the au- hyaenids can be rejected on the basis of their ditory ossicles (fig. 1OA) with which such a distinguishing and contradictory bulla malleus is associated is generally regarded as morphs. the plesiomorphic therian state (Fleischer, 1978). Mallei with reduced laminae (fig. 1OB, THE PROBLEM OF C) are widely distributed across many orders HYAENID-HERPESTID of mammals, including both caniform and RELATIONSHIP feliform Carnivora. Such a distribution is likely to result from repeated instances ofpar- The placement of the three auditory bulla allel evolution, in each case directed toward elements in living hyaenids and herpestids the development offreely mobile ear ossicles suggested that the adult hyaenid bulla pattern having low mass, little stiffhess, and reduced might be derived from the configuration frictional resistance (Hunt and Korth, 1980). adopted by these elements in the neonatal Consequently, independent evolution ofthe herpestid bulla (Hunt, 1987, 1989: fig. 5). mallei seen in living hyaenids and herpestids This remains a plausible hypothesis open to must be entertained as a hypothesis equally future testing, however, few other features of as plausible as derivation ofthese mallei from herpestid and hyaenid morphology lend sup- a common ancestor. Such mallei can be pre- port to this view. In my initial analysis of dicted as the frequent product of selective aeluroid carnivorans, some derived traits ap- pressures directed toward a more mobile os- peared to link hyaenids and herpestids (Hunt, sicular mechanism in various eutherian lin- 1987: fig. 19, node 4) but the nature of the eages. Hence, hyaenids and herpestids cannot evidence for a sister-group relationship was be linked solely on the basis of the form of not compelling: (a) restriction of caudal en- the malleus. Are there any other features of totympanic to the posterior auditory region; the auditory region that suggest affinity be- (b) blunt nonretractile claws; (c) an invagi- tween the two groups? nated anal pouch with glands or sacs; (d) an The few species ofhyaenids and herpestids absent or vestigial aural bursa; (e) external that have been available for dissection ofthe auditory meatus prolonged as a bony tube. auditory region show similar petrosal morphs. As an additional complication, herpestids The promontorium appears to possess a share a large number ofautapomorphies that characteristic shape, surmounted by a tall, tend to obscure their relationships with other bladelike ventral promontorial process. aeluroids. In this context, the morphological However, it has not been possible to survey similarity of herpestid and hyaenid mallei petrosal shape in a large enough sample to merits comment. Both families have mallei determine the taxonomic distribution of this with a reduced lamina and a short anterior morph, and in the interim there remains a process, contrary to felids and viverrids in paucity of compelling synapomorphies to which a well-developed pocketed lamina is unite the two families. Thus, figure 16 depicts present. Is this similarity inherited from a hyaenids and herpestids as independent ae- E F G H J

g...... -,.,I

......

...... ~~~~~~...... 9 - * P......

...... , ...... ~~~~~~~~~...... ~~~~~~~~~...... _

A f sssww@***+@@8*

...... _ ,_~~~~_......

@ @.

.. :::::::::::::::::::::::.N~~~~~~~~~~::::

......

...... :::::::::::::: 'STEM AELUROIDS...... z g ...... \ ...... UR ID

:~~~~ ~~~~~~~~..... ::......

MODERN . . . MODERN ARCHAIC ARCHAIC BASICRANIAL BASICRANIAL BASICRANIAL BASICRANIAL PATTERNS 1. . PATTERNS PATTERNS PATTERNS INFERRED INFERRED Fig.::::::::::::::::::::::::::::16. Evolution::::::::.ofbasicranial patterns within the aeluroid Carnivora (modified from Hunt, 1989). The modern basicranial patterns characteristic of the living families emerge during the late Oligocene to early Miocene:::::::::::::::::::::...... interval in the Old World, and persist relatively unchanged for the remainder of the Neogene within:::::each::::::::family.:::::::::::::::.This revised chart incorporates new information on Eurasian aeluroids discovered since the publication of my initial hypothesis (Hunt, 1989: fig. 13): (a) transfer of early Miocene Herpestides to Viverridae, and (b) indication of a more ancient and independent derivation of hyaenids and herpestids from the stem aeluroid group. Early or mid-Miocene basicrania of modern grade are now known for all aeluroid families except Herpestidae. A, Stenogalejulieni; B, Palaeoprion- odon lamandini; C, Herpestides antiquus; D, Tungurictis spocki; E, Panthera onca; F, Civettictis civetta; G, Nandinia binotata; H, Crocuta crocuta; J, Galidia elegans. 32 AMERICAN MUSEUM NOVITATES NO. 3023 luroid lineages that may in time prove to be form, but that the auditory regions may have derived from a common ancestry exclusive been nearly identical. An important aspect of of that for felids/viverrids. the auditory region ofPalaeoprionodon is that the mastoid and exoccipital bones surround- ANTIQUITY OF THE VIVERRID ing the caudal entotympanic are quite thin AUDITORY REGION and narrow in ventral view (Teilhard, 1915: pl. 9, fig. 10), and do not form a wide bony Prior to examination of the Aquitanian shelf as in the primitive aeluroid Nandinia. samples of Herpestides, my earlier research This configuration ofmastoid and exoccipital indicated that the oldest confirmed viverrid reveals a posterior auditory region much like auditory regions belonged to Plio-Pleistocene the modern viverrids Prionodon and Poiana. fossils from the Siwaliks ofsouthern Asia and In addition to Palaeoprionodon, there is the Langebaanweg locality in South Africa another Old World Oligocene aeluroid car- (Hunt, 1989). Other fossils that I was able to nivoran that often has been regarded as a study in the British Museum (N.H.) from viverrid: the Quercy Stenoplesictis cayluxi. Oeningen in Europe and from the Miocene The auditory region of Stenoplesictis, al- Siwaliks suggested that viverrid auditory though undoubtedly aeluroid, is not certainly regions of modern grade possibly extended viverrid if an assessment is based on the su- to the mid-Miocene, but there was no con- perb skull described by Piveteau (1943) from firming evidence. Now, however, because the the collection of the Faculte des Sciences, basicranium ofHerpestides is essentially that Marseilles. Here the ectotympanic is much of a modern viverrid, the Lyon and Basel more inflated than in Palaeoprionodon, and samples extend the record of viverrid basi- apparently encompasses the petrosal pro- crania to the early Miocene (European Neo- montorium. This ectotympanic configura- gene mammal zone MN2a). We may now tion would seem to preclude assignment to anticipate that true viverrids were not only either viverrids or felids, and it appears to be present but probably diverse in the Miocene too strongly inflated to belong to a herpestid. of the Old World, and that the viverrid au- Moreover, the caudal entotympanic chamber ditory pattern probably developed in Oligo- must have been relatively small. This sug- cene ancestors. gests that the Marseilles cranium referred to The oldest viverrid auditory region, how- Stenoplesictis is either an early aeluroid lin- ever, appears to belong not to Herpestides, eage that terminates without living descen- but to the small Quercy aeluroid Palaeo- dants, or possibly represents an ancestral prionodon lamandini, whose skull was illus- hyaenid in which the bulla chambers presage trated by Teilhard (1915: pl. 9, fig. 10). I have the modern configuration. not examined the skulls figured by Teilhard in the collections of the Paris Museum, but his photograph indicates an auditory region CONCLUSIONS with large petrosal promontorium, and an 1. The auditory region of the Aquitanian ectotympanic applied to the promontorium (early Miocene) aeluroid Herpestides antiqu- in the manner ofviverrids and felids, where- us is of the viverrid type, and cannot be re- by the posterior limb rests on the promon- ferred to Herpestidae or Hyaenidae as pre- torium just anterior to the round window. viously believed. Furthermore, although there is no caudal en- 2. Basicranial and auditory bulla patterns totympanic preserved in Palaeoprionodon, it of viverrids and felids had evolved by the is clear from Teilhard's figure 10 that a small early Miocene in Europe. The auditory bullae caudal entotympanic was probably present, of the viverrid Herpestides antiquus and the occupying the posterior auditory region, and felid Proailurus lemanensis, both from the very likely extending forward a short distance Aquitanian of the Allier basin, France, are along the medial side of the ectotympanic essentially ofmodern grade, and demonstrate crescent. Comparison of Teilhard's photo- that the two families existed as independent graph with the skull of the living Prionodon lineages by that time. suggests that the Quercy aeluroid had not only 3. Viverrids and felids exhibit parallels in a caudal entotympanic of similar size and the pattern of ontogenetic development of 1991 HUNT: VIVERRID AFFINITIES OF HERPESTIDES 33 their auditory structure that suggest they are Beaumont, G. de sister groups, e.g., in the relationship of ec- 1967. Observations sur les Herpestinae (Vi- totympanic to petrosal promontorium, the verridae, Carnivora) de l'Oligocene su- geometry of the petrosal, and the pattern of peieur avec quelques remarques sur des Hyaenidae du Neogene. Arch. Sci. Ge- bulla development. neve 20(1): 79-107. 4. Orientation and morphology ofthe au- Beaumont, G. de, and P. Mein ditory ossicles in aeluroids indicate that these 1972. Recherches sur le genre Plioviverrops carnivorans display transitional to derived Kretzoi (Carnivora, ?Hyaenidae). C.R. states of the ossicular mechanism relative to des Seances, SPHN Geneve 25(3): 383- a eutherian morphotype. Among aeluroids, 394. viverrids and felids retain the most plesio- Blainville, H. M. D. de morphic patterns, whereas hyaenids and her- 1842. Des Viverras. Osteographie ou descrip- pestids exhibit more derived states. Viverrids tion iconographique comparee du sque- and herpestids are extremely different in os- lette et du systeme dentaire des mam- sicular morphology, one more item in a miferes recent et fossiles pour servir de base 'a la zoologie et 'a la geologie. Vol. steadily accumulating body of evidence in- II, Carnivora. dicating the marked distinction between the Bucher, H., L. Ginsburg, and J. Cheneval two groups. 1985. Nouvelles donnees et interpretations sur 5. The hyaenid bulla pattern appears in la formation des gisements de vertebres the mid-Miocene fully developed in at least aquitaniens de Saint-Gerand-le-Puy two distinct lineages of widely divergent (Allier, France). Geobios 18(6): 823- adaptive type (Percrocuta, Tungurictis), in- 832. dicating that the origin ofthe pattern predates Cheneval, J., and M. Hugueney the divergence of these lineages, and there- 1985. Excursion a la carriere "Les Perards" fore is of or (Saint-Gerand-le-Puy; Aquitanien de probably early Miocene age older. l'Allier), 21 Septembre 1985. Excursion 6. Because aeluroids do not appear in the guide: Table ronde internationale sur fossil record until mid- to late Oligocene, fol- l'evolution des oiseaux d'apres le te- lowing the "Grande Coupure" event in Eur- moignage des fossiles, 13 pp., 6 figs. asia, and because basicranial patterns of Donsimoni, M., and D. Giot modern grade first appear in viverrids, felids, 1977. Les calcaires concretionnes lacustres de and probably hyaenids in the early Miocene, l'Oligocene superieur et de l'Aquitanien the development of these patterns may take de Limagne (Massif central). Bull. Bur. place over a relatively brief interval within Rech. geol. min., Paris 1(2): 131-169. the later part of the Oligocene. By inference, Filhol, H. the herpestid pattern probably also originates 1879. Etude des mammiferes fossiles de Saint- Gerand-le-Puy (Allier). Bibliotheque de during the late Oligocene to early Miocene l'Ecole des Haute Etudes (Sect. Sci. Nat.) interval. 19(1): 1-252. 7. Aeluroid basicranial types can be iden- Fleischer, G. tified primarily from the structural relation- 1973. Studien am Skelett des Gehororganes ships ofauditory bulla, petrosal, and auditory der Saugetiere, einschliesslich des ossicles. These patterns on present evidence Menschen. Saugetierkd. Mitteilungen must develop within the mid-Cenozoic Oli- (Miinchen) 21: 131-239. gocene interval from about 34 to 24 Ma. Once 1978. Evolutionary principles of the mam- established, they persist as stable morphs for malian middle ear. Adv. Anat. Em- the remainder ofthe Cenozoic history ofthese bryol. Cell Biol. 55(5): 1-70. Hugueney, M. carnivoran groups. Stasis in patterns of au- 1984. Evolution du paleoenvironment dans le ditory morphology characterizes the aeluroid Tertiare de Limagne (Massif central, Carnivora during the Neogene. France) a partir des faunes de mammiferes. Geobios. Mem. special 8: 385- REFERENCES 391. Hunt, R. M., Jr. Allin, E. F. 1974. The auditory bulla in Carnivora: an an- 1975. Evolution ofthe mammalian middle ear. atomical basis for reappraisal of carni- J. Morphol. 147: 403-438. vore evolution. J. Morphol. 143: 21-76. 34 AMERICAN MUSEUM NOVITATES NO. 3023

1987. Evolution of the aeluroid Carnivora: sin hydrographique superieur de la Loire significance of auditory structure in the et de l'Allier. Paris. nimravid cat Dinictis. Am. Mus. Novi- Qiu, Zhanxiang, Ye Jie, and Cao Jingxuan tates 2886: 74 pp. 1988. A new species ofPercrocuta from Tong- 1989. Evolution of the aeluroid Carnivora: xin, Ningxia. Vertebr. Palasiatica 26(2): significance ofthe ventral promontorial 116-127. process of the petrosal, and the origin Schlosser, M. ofbasicranial patterns in the living fam- 1890. Die Affen, Lemuren, Chiropteren, In- ilies. Am. Mus. Novitates 2930: 32 pp. sectivoren, Marsupialier, Creodonten Hunt, R. M., Jr., and W. Korth und Carnivoren des europaiischen Ter- 1980. The auditory region of Dermoptera: tiiirs, Part III. Beitrage Palaontol. Os- morphology and function relative to terreich-Ungarns Orients 8: 401-407. other living mammals. J. Morphol. 164: Segall, W. 167-211. 1969. The auditory ossicles (mallus, incus) and Kermack, K. A., and F. Mussett their relationships to the tympanic in 1983. The ear in mammal-like reptiles and marsupials. Acta Anat. 73: 176-191. early mammals. Acta Palaeontol. Po- 1970. Morphological parallelisms ofthe bulla lonica 28(1-2): 147-158. and auditory ossicles in some insecti- MacPhee, R. D. E. vores and marsupials. Fieldiana Zool. 1981. Auditory regions of primates and eu- 51: 169-205. therian insectivores. Contrib. Primatol. 1971. The auditory region (ossicles, sinuses) 18: 1-282. in gliding mammals and selected rep- Petter, G. resentatives of non-gliding genera. 1974. Rapports phyletiques des viverrides Fieldiana Zool. 58: 27-59. (Carnivores fissipedes). Les formes de Teilhard de Chardin, P. Madagascar. Mammalia 38(4): 605-636. 1915. Les carnassiers des Phosphorites du Piveteau, J. Quercy. Ann. Paleontol. 9(3-4): 103- 1943. Etudes sur quelques mammiferes des 192. Phosphorites du Quercy. Ann. Paleon- Viret, J. tol. 30: 63-72. 1929. Les faunes des mammiferes de l'Oli- Pomel, A. gocene superieur de la Limagne Bour- 1853. Catalogue methodique et descriptif des bonnaise. Ann. l'Univ. Lyon, n. ser., 47: vertebres fossiles decoverts dans le bas- 1-328.

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

THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER.