A new species of Hathliacynidae (Metatheria, Sparassodonta) from the middle Miocene of Quebrada Honda, Bolivia

Journal of Vertebrate Paleontology 26(3):670–684, September 2006 © 2006 by the Society of Vertebrate Paleontology

A NEW SPECIES OF HATHLIACYNIDAE (METATHERIA, SPARASSODONTA) FROM THE MIDDLE MIOCENE OF QUEBRADA HONDA, BOLIVIA ANALÍA M. FORASIEPI1,2, MARCELO R. SÁNCHEZ-VILLAGRA3, FRANCISCO J. GOIN4, MASANARU TAKAI5, NOBUO SHIGEHARA5, and RICHARD F. KAY6 1 Anatomical Sciences and Neurobiology, School of Medicine, University of Louisville, 500 S. Preston, Louisville, Kentucky 40292, U.S.A., [email protected]; 2 Museo Argentino de Ciencias Naturales ‘Bernardino Rivadavia’ Avenida Angel Gallardo 470, Capital Federal, 1405, Buenos Aires, Argentina, [email protected]; 3 The Natural History Museum, Department of Palaeontology, Cromwell Road, London SW7 5BD, England, [email protected]; 4 Departamento Paleontología de Vertebrados, Museo de La Plata, Paseo del Bosque s/n, La Plata, 1900, Buenos Aires, Argentina, [email protected]; 5 Primate Research Institute, Kyoto University, Inuyama, Aichi 484, Japan, [email protected]; 6 Department of Biological Anthropology and Anatomy, Duke University, Medical Center, Durham, North Carolina 27710, U.S.A., [email protected]

ABSTRACT—A new species of Hathliacynidae (Sparassodonta, Metatheria), Acyon myctoderos, from the middle Miocene of Quebrada Honda, Bolivia, is described. This new species is the largest known hathliacynid. Compared to the type species of the genus, A. tricuspidatus, Acyon myctoderos differs in having: (1) longer diastemata among premolars; (2) p2 comparatively more robust, with a better developed posterior cusp and with a sharp anterior crest; (3) lower molars with a more poorly developed anterobasal cingulum; (4) m1–m3 with hypoconulids less salient posteriorly and more vertically oriented; and (5) larger hypoconids at least on the m2. A phylogenetic analysis including nine taxa of Sparassodonta, with Mayulestes as the outgroup, showed that Acyon is more closely related to Cladosictis than to any other hathliacynid.

INTRODUCTION The extinct metatherian fauna of South America was very rich and diverse and included forms with ecological counterparts occupied by eutherians on the northern continents (Marshall, 1977a; Simpson, 1980, among others). This extinct fauna included insectivorous to omnivorous didelphimorphians, frugivorous paucituberculatans and polydolopimorphians, and carnivorous sparassodonts. The ecological significance of the sparassodonts is that they were the only group of mammalian carnivores in South American during most of the Tertiary, when this continent was isolated from other landmasses (e.g., Marshall, 1977a; Simpson, 1980). The fossil record of the Sparassodonta extends from the early Paleocene to the late Pliocene (Marshall, 1978; Simpson, 1980; Muizon, 1998). Among them, the Hathliacynidae represents one of the most generalized groups (Marshall, 1981; Muizon, 1999). Hathliacynids were of small to medium size, with a slender body form compared with other sparassodonts, and probably having scansorial (i.e., partly arboreal and partly terrestrial) life habits (Argot, 2003, 2004). Hathliacynids have been reported from Argentina, Brazil, and Bolivia (e.g., Ameghino, 1887, 1891; Paula Couto, 1952; Marshall, 1981; Villarroel and Marshall, 1982, 1983; Hoffstetter and Petter, 1983; Muizon, 1999, among others). The most comprehensive review of this family (Marshall, 1981) recognized 21 species (three of them are probably nomina vana). Except for Cladosictis patagonica and Sipalocyon gracilis (both Miocene), which are represented by almost-complete skeletons (Sinclair, 1906), most of the remaining recognized species are known only by dental and fragmentary cranial remains. The fossil-bearing Quebrada Honda Formation is exposed in the vicinity of the settlement of Quebrada Honda, southern Bolivia, between 21°53⬘–22°05⬘ south latitude and 65°05⬘–65°10⬘ west longitude (Fig. 1). Quebrada Honda is located in the De-

partment of Tarija, about 65 kilometers southwest of the department capital Tarija, and 20 km north of the Argentine frontier. It is at an elevation of about 3500 meters above sea level. The fossil-bearing exposures in the valley of the Río Honda and its tributaries occur principally, but not exclusively, in two drainages, the Quebrada Honda and Quebrada El Rosario. The approximately 360-m-thick section of red clays and silts with intercalated tuffs referred to the Quebrada Honda Formation were deposited in a small intermontane sedimentary basin just west of the principal Andean thrust in the Bolivian Eastern Cordillera. The sediments rest on strongly folded and partially metamorphosed Paleozoic (Cambrian) basement (Troëng et al., 1993). MacFadden et al. (1990) reported two conventional K/Ar dates and a local magnetic polarity stratigraphy for the Quebrada Honda section. A preferred age of 12.83 +/− 0.07 Ma for the tuff at the top of the fossil-bearing portion permits correlation of the fossil-bearing N1 and R1 local polarity intervals to Chrons C5Ar.3r and C5Ar.2n, calibrated to 12.991–12.775 Ma using the GMPTS of Berggren et al. (1995). Thus, based on published K/Ar ages and local magnetic polarity stratigraphy, the fossil mammals from Quebrada Honda have an extrapolated age of between 13.0 and 12.8 Ma, corresponding to the Laventan South American land mammal age in Colombia (Madden et al., 1997). The relatively rich mammalian assemblage recovered in Quebrada Honda includes the following groups: Metatheria (Borhyaenidae, Caenolestidae, Argyrolagidae), Xenarthra (Dasypodidae, Glyptodontidae, Mylodontidae, Megalonychidae), Notoungulata (Hegetotheriidae, Interatheriidae, Mesotheriidae, Toxodontidae), Astrapotheria (Astrapotheriidae), Litopterna (Protherotheriidae, Macraucheniidae), and Rodentia (Octodontidae, Echymidae, Capromyidae, Chinchillidae, Caviidae) (Hoffstetter, 1977; MacFadden and Wolff, 1981; Marshall and Sempere, 1991). The occurrence of sparassodonts in Quebrada

670

FORASIEPI ET AL.—NEW HATHLIACYNID FROM BOLIVIA

FIGURE 1. Bolivia.

Geographic location of Quebrada Honda, in southern

Honda was first mentioned in a faunal list by MacFadden and Wolff (1981); this record was later repeated by Marshall and Sempere (1991), but no description of a specimen was provided. Here we describe a new specimen, different from that mentioned previously, consisting of a well-preserved skull and partial postcranial skeleton belonging to a new species of the genus Acyon (Hathliacynidae), and analyze its phylogenetic relationships. The nomenclature in the descriptions is largely after Muizon (1998, 1999; for dental, cranial, and postcranial terminology), and Wible (2003; for cranial terminology). The comparisons with other sparassodonts are based on the specimens cited in Appendix 1 and in the referred literature. The phylogenetic analysis, largely based on characters mentioned by Muizon (1999), is restricted to sparassodonts and focused on hathliacynids. Abbreviations CORD-PZ, Museo de Paleontología, Facultad de Ciencias Exactas, Físicas y Naturales de la Universidad Nacional de Córdoba, Córdoba, Argentina; FMNH, Field Museum of Natural History, Chicago, Illinois; MACN A, Ameghino collection, Museo Argentino de Ciencias Naturales ‘Bernardino Rivadavia,’ Buenos Aires, Argentina; MLP, Museo de La Plata, Buenos Aires, Argentina; MNHN-Bol, Museo Nacional de Historia Natural, La Paz, Bolivia; SALMA: South American land mammal age; UCMP, University of California Museum of Paleontology, Berkeley, California. Capital and lower case letters, I/i, incisor, C/c, canine, P/p, premolar, and M/m, molar, refer to upper and lower teeth, respectively. Incisor nomenclature followed here is I1, 2, 3, 4, and i1, 2, 3 from front to back, without considering the homologies proposed by Hershkovitz (1982). SYSTEMATIC PALEONTOLOGY Subclass METATHERIA Huxley, 1880 Order SPARASSODONTA Ameghino, 1894 Family HATHLIACYNIDAE Ameghino, 1894 ACYON Ameghino, 1887 Acyon Ameghino, 1887:8. Anatherium Ameghino, 1887:8 (partim). Hathliacynus Mercerat, 1891 (nec Ameghino, 1887). ?Ictioborus Ameghino, 1894 (partim). Cladosictis Cabrera, 1927 (nec Ameghino, 1887) (partim). Type Species—Acyon tricuspidatus Ameghino, 1887:8. Included Species—A. tricuspidatus Ameghino, 1887, and A. myctoderos sp. nov.; Acyon herrerae (Marshall, 1981) could represent a junior synonym of A. tricuspidatus (see below).

671

Distribution and Age—South America (Argentina and Bolivia). Early to middle Miocene (Colhuehuapian, Santacrucian, and Laventan SALMA). Diagnosis—(Modified from that of Anatherium in Marshall, 1981): largest known hathliacynid (length of skull up to 180 mm; for other hathliacynids see Marshall, 1981); dentary long, low, and gracile; large diastemata separating p1 from canine and p2; premolars strongly laterally compressed; talonids of m1–m3 without entoconid (autapomorphy, character 45). Comments—The early Miocene (Santacrucian SALMA) species, Acyon tricuspidatus, was described by Ameghino (1887) based on a right mandibular fragment (MLP 11-64) with the last premolar and almost complete molars. Mercerat (1891) referred MLP 11-64 to Hathliacynus tricuspidatus, whereas Cabrera (1927) allocated the same specimen to Cladosictis tricuspidatus, as listed in the old catalog of the Museo de La Plata. In his review of the hathliacynids, Marshall (1981) regarded this species as a junior synonym of Anatherium defossus Ameghino, 1887. He included the following specimens in the hypodigm of Anatherium defossus: MACN A 669 (type of Anatherium defossus), MLP 11-64 (type of Acyon tricuspidatus), MACN A 9 (type of Acyon ?bardus), MACN A 5988 (type of Ictioborus destructor), and MACN A 646. Marshall (1981:103) also recognized a new species, Anatherium herrerae, on the basis of only one specimen (FMNH P 13521) from Colhuehuapian beds (early Miocene) at Gran Barranca (Chubut Province, Argentina). Review of the specimens mentioned above leads us to conclude that: (1) specimens MACN A 9, MACN A 646, and MACN A 669 (type of Anatherium defossus) are referable to Cladosictis sp., and, consequently, the genus Anatherium represents a junior synonym of Cladosictis; (2) specimen MACN A 5988 is only tentatively referable to Acyon tricuspidatus; and (3) Anatherium herrerae Marshall, 1981, is referable to the genus Acyon (see the generic diagnosis of Acyon), and is very close or probably conspecific with Acyon tricuspidatus. Therefore, we recognize Acyon as a valid genus, including two species, A. tricuspidatus Ameghino, 1887 (Santacrucian SALMA), and the new species from the Laventan SALMA described below. ACYON MYCTODEROS, sp. nov. (Figs. 2–9) Type and Only Specimen—MNHN-Bol-V-003668, a skull with associated lower jaws and postcrania. The left side of the skull is almost complete, including a portion of the left petrosal; in contrast, the right side of the skull is partially broken. Both lower jaws are preserved: the left is mostly complete except for the incisor crowns and part of the coronoid process. The postcranial elements include the atlas, axis, four isolated vertebral bodies, cuboid, ectocuneiform, a probable entocuneiform, and five fragmentary metapodials. Locality and Age—Papachacra, Quebrada Honda, Bolivia. Laventan SALMA, middle Miocene (between 13 and 12 Ma; MacFadden et al., 1990; Madden et al., 1997). The specimen was collected in 1995 by a joint expedition of the Kyoto University and the Museo Nacional de Historia Natural of La Paz led by M. Takai and bears the field number ‘Takai 3668.’ Derivation of Name—From the Greek mykter, nose, and deros, long, in reference to the long snout. Diagnosis—Largest species of the genus (A. myctoderos approximately 5% larger than A. tricuspidatus, based on measurements of the dentition; Table 1; Marshall, 1981:Table 25) and largest know hathliacynid; differs from A. tricuspidatus (Cabrera, 1927:fig. 6) in having longer diastemata among premolars; p2 comparatively more robust and with a better-developed posterior cusp; less-developed anterobasal cingulum on lower molars; hypoconulid less salient posteriorly and more vertically oriented on m1–m3; hypoconids larger, at least on m2.

672

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 26, NO. 3, 2006

TABLE 1. Measurements (in mm) of the upper and lower postcanine dentition of Acyon myctoderos sp. nov. MNHN-Bol-V-003668. LP1

WP1

LP2

WP2

LP3

WP3

LM1

WM1

LM2

WM2

LM3

WM3

LM4

WM4

6.40 6.20

2.70 2.90

8.40 8.00

3.00 3.00

10.05 10.04

4.35 4.30

11.85* 10.04*

7.45* 5.07*

10.6* —

3.05 —

10.85 —

3.15 —

5.50 —

7.20 —

Lp1

Wp1

Lp2

Wp2

Lp3

Wp3

Lm1

Wm1

Lm2

Wm2

Lm3

Wm3

Lm4

Wm4

7.75* 7.70

2.85 3.00

7.95 —

3.40 3.55

9.95 9.85

4.05 3.85

8.80 9.60

3.40 3.60

10.05 9.70

4.45 4.45

10.70 11.50

4.85 5.15

12.80 12.10

5.65 5.65

*Indicates that the measurement is approximate. First row of measurements is the left dentition; the second row of measurements is the right dentition. Abbreviations: L, anteroposterior length; W, mediolateral width.

DESCRIPTION Skull—The skull is preserved in several separate portions (Figs. 2–7). The main fragment of the skull is formed by the left portion of the snout with the dentition and part of the associated right nasal bone, the almost complete posterior hard palate, left orbital region, part of the left zygomatic arch, left glenoid cavity, and the anterior portion of the basicranium. Most of the roof of the choanal region and the lateral side of the skull are fractured; hence, anatomical features and sutures between bones are difficult to distinguish in these areas. The other fragments of the skull preserve the right posterior portion of the premaxilla associated with the right maxilla and the canine-postcanine dentition, the endocranial portion of the left petrosal bone, and a portion of the occiput with the left occipital condyle and the surrounding basicranium. The skull is fox-like in shape, with the preorbital and postorbital regions particularly elongated, as is characteristic of the Hathliacynidae (Figs. 2–6). The snout is long and narrow; at the

level of the diastema between P1 and P2 there is a shallow constriction, clearly seen in dorsal view. Posterior to P2 the snout widens to the level of M3–M4, where the base of the zygomatic arch is located. The postorbital processes are prominent and placed almost at the middle of the anteroposterior length of the skull. The postorbital constriction is very pronounced and located on the posterior two-thirds of the skull. The braincase is narrow, being at its most posterior extent as wide as the skull at the level of the postorbital processes. Both nasals are preserved; the left is almost complete but the right is broken posteriorly. The nasal contacts laterally (from anterior to posterior) the premaxilla, maxilla, lacrimal, and frontal. There is a wide contact between the nasal and the lacrimal bones. The suture between the nasals and frontals has a wide ‘W’ shape, with an anterior mid-sagittal wedge of the frontals (Fig. 2). The anterior two-thirds of the nasals are narrow with parallel lateral borders. At a point approximately dorsal to P2, they flare laterally, reaching their maximum width at the level of the suture with the lacrimal. The anterior-most edges of the nasals are bro-

FIGURE 2. Acyon myctoderos, MNHN-Bol-V-003668. Stereopair photographs and line-drawing reconstruction of the skull in dorsal view. Abbreviations: ap, ascending process of the premaxilla; FR, frontal; JU, jugal; LA, lacrimal; lc, lambdoidal crest; MX, maxilla; NA, nasal; PA, parietal; pn, precanine notch; pop, postorbital process; PX, premaxilla; sc, sagittal crest; sf, subsquamosal foramen; SQ, squamosal; tc, temporal crest. Scale bar equals 50 mm.

FORASIEPI ET AL.—NEW HATHLIACYNID FROM BOLIVIA

673

FIGURE 3. Acyon myctoderos, MNHN-Bol-V-003668. Stereopair photographs and line-drawing reconstruction of the skull in ventral view. Abbreviations: AS, alisphenoid; atp, alisphenoid tympanic process; BO, basioccipital; BS, basisphenoid; eam, external acoustic meatus; EC, ectotympanic bone; fm, foramen magnum; fo, foramen ovale; gf, glenoid fossa; hf, hypoglossal foramen; if, incisive foramen; jf, jugular foramen; JU, jugal; mpf, minor palatine foramen; mpp, medial palatine process; MX, maxilla; oc, occipital condyle; pcp, paracondylar process; pgf, postglenoid foramen; PL, palatine; pp, protoconid pit; PX, premaxilla; SQ, squamosal. Scale bar equals 50 mm.

ken; based on the condition observed in most sparassodonts (e.g., Cladosictis, Borhyaena, and Arctodictis), they probably did not overhang the external nasal aperture (versus the condition in didelphids). The left premaxilla is almost complete, but the right one is known only by its most posterior portion. In lateral view the premaxilla is seen to contact the nasal dorsally and the maxilla laterally; in ventral view, it contacts the maxilla posteriorly and the opposite premaxilla in the midsagittal plane. In lateral view (Figs. 4, 6A), the premaxilla has a roughly triangular shape, wide in the alveolar border and acute dorsally; both anterior and posterior borders are slightly concave posteriorly. The ascending process of the premaxilla (sensu Muizon, 1998) is thin and interposed between the nasal and the maxilla. It projects posteriorly to the level of the posterior border of the canine. The paracanine fossa (Figs. 4, 6A) is formed exclusively by the premaxilla as in other Sparassodonta, but differs from that of Mayulestes and didelphids in which the anterolateral process of the maxilla contributes to the formation of this fossa (Muizon, 1998, 1999). A low ventrally concave crest borders the paracanine fossa dorsally. In dorsal view (Fig. 2), the precanine notch (i.e., the constriction between the posterior incisor and the canine; Muizon, 1999) is well developed, as in Cladosictis and unlike Sipalocyon among Hathliacynidae (Muizon, 1999). The anterior border of the premaxilla is flat and almost straight, differing at least from Mayulestes (Muizon, 1998) and Sipalocyon (see MACN A 691– 703 and 5439–5449), where it is parabolic. Only three complete alveoli for the incisors are preserved in the anterior border of the

premaxilla; however, there is a significant space between the medial-most preserved alveolus and the sagittal plane. In addition, on the medial-most border of the preserved part of the premaxilla there is a small concavity facing anteriorly that suggests the presence of the alveolus for a fourth tooth closer to the mid-sagittal plane. In ventral view (Fig. 3), the palatine portion of the premaxilla forms approximately the anterior 10% of the hard palate. The most posteromedial portion of the left premaxilla is broken, but the information provided by the right element helps in the reconstruction of this area (Fig. 3). The incisive foramen is oval and anteroposteriorly elongated. It is bordered by the premaxilla anteriorly, laterally, and medially, and by the maxilla posteriorly. The medial palatine process of the premaxilla is a narrow bridge of bone interposed between the incisive foramen and the midsagittal plane. The most posterior portion of this process is broken on both sides of the skull; therefore, its extension and the exact position of the suture with the maxilla can not be located. The maxilla is known by the left and right elements. In lateral view (Figs. 4, 6A), the maxilla contacts the premaxilla anteriorly, the nasal dorsally, and the lacrimal and jugal posteriorly. The maxilla does not contact the frontal, as occurs in primitive metatherians (Rougier et al., 1998). In ventral view (Fig. 3), the maxilla contacts the premaxilla anteriorly, the palatine posteriorly, and the maxilla of the opposite side in the mid-sagittal plane; in the orbital region the sutures are difficult to distinguish. In lateral view (Figs. 4, 6A), the maxilla is considerably longer than high, similar to other hathliacynids, but differing from

674

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 26, NO. 3, 2006

FIGURE 4. Acyon myctoderos, MNHN-Bol-V-003668. Stereopair photographs of the skull in lateral view. Scale bar equals 50 mm.

borhyaenids, which have an anteroposteriorly short but deep maxilla. The infraorbital foramen is large and faces anteriorly. The border of the infraorbital foramen is ‘U’-shaped and located above the posterior border of P3. This border protrudes conspicuously laterally; therefore, the roof of the most anterior portion of the infraorbital canal can be seen in dorsal view. The root of the canine defines a low prominence on the lateral side of the maxilla, which extends nearly to the dorsal border of this bone. The zygomatic process of the maxilla is short and restricted to the base of the zygomatic arch. In other sparassodonts, as well as other basal metatherians (Rougier et al, 1998), the maxilla

FIGURE 5. Acyon myctoderos, MNHN-Bol-V-003668. Photographs of the left dentary in lateral (A) and medial (B) views. Scale bar equals 50 mm.

FIGURE 6. Line-drawing reconstruction of the skull of Acyon myctoderos, MNHN-Bol-V-003668 in lateral view (A), left dentary in lateral (B), and medial (C) views. Abbreviations: (as for Figs. 2, 3) anp, angular process; cd, condyle; corp, coronoid process; DEN (ar), ascending ramus; DEN (hr), horizontal ramus; EO, exoccipital fr, foramen rotundum; iff, infraorbital foramen; lf, lacrimal foramen; lt, lacrimal tubercle; masf, masseteric fossa; mc, masseteric crest; menf, mental foramina; mf, mandibular foramen; pf, paracanine fossa; pgp, postglenoid process; pp, post-tympanic process; sy, symphysis. Scale bar equals 50 mm.

behind the infraorbital foramen (including the zygomatic process) flares laterally, forming ‘cheeks,’ a feature that is most clearly seen in ventral view. Acyon myctoderos lacks ‘cheeks’ (Fig. 3). The anterior half of the palatal process is perforated by numerous small foramina probably for the palatine arteries, veins, and branches of the maxillary division of the trigeminal nerve, structures that in didelphids and some other mammals pass through the major palatine foramen (Evans, 1993; Wible, 2003). The largest of these foramina is located between the canines, as in other Sparassodonta (e.g., Borhyaena, Arctodictis, Cladosictis, and Sipalocyon). There is no major palatine foramen, nor any palatal vacuities, a resemblance to other Sparassodonta. In Mayulestes, a major palatine foramen is probably present (Muizon, 1998:fig. 6, identified as medial palatine foramen), but the palatine vacuities are absent. The minor palatine foramen is a small oval aperture anterolaterally-posteromedially elongated, which opens between the maxilla and the palatine. Lateral to the minor palatine foramen and between M3 and M4 there is a deep, round pit, which houses the tall protoconid of m4 when the jaws are closed. The anterior-most contact of the maxilla and the palatine bone is at the level of M1, as in Cladosictis. In dorsal view (Fig. 2), the maxilla contributes to the floor of the orbit and probably the floor of the infraorbital canal. The floor of the orbit bears numerous fractures, but some natural grooves and small foramina can be distinguished. These foramina and the corresponding grooves probably transmitted branches of the maxillary artery, vein, and nerve to supply the posterior teeth (Evans, 1993). Both palatines are preserved in the specimen described here, but the left is better preserved than the right. In ventral view (Fig. 3), the horizontal process of the palatine forms approximately 35% of the most postero-medial length of the hard palate. Anterolaterally the palatine contacts the maxilla by a rounded

FORASIEPI ET AL.—NEW HATHLIACYNID FROM BOLIVIA

FIGURE 7. Acyon myctoderos, MNHN-Bol-V-003668. Stereopair photographs and line-drawing reconstruction of the left petrosal in cerebellar view. Abbreviations: amf, anteromedial flange; ca, cochlear canaliculus; cp, crista petrosa; iaf, inferior acoustic foramen; iam, internal acoustic meatus; ips, sulcus for the inferior petrosal sinus; saf, superior acoustic foramen; sf, subarcuate fossa; sps, sulcus of the prootic sinus; va, vestibular aqueduct. Scale bar equals 5 mm.

suture and the palatine of the opposite side in the midsagittal plane. As was mentioned above, the minor palatine foramen opens in the suture between the palatine and the maxilla. The palatine extends posteriorly to the posterior edge of M4 as in Cladosictis, Sipalocyon, and Mayulestes. It differs from other sparassodonts (e.g., Borhyaena) where it is anterior or only reaches the posterior edge of M4. The posterior rim of the choanae is thick and double-arched, but lacks the postpalatine torus found in Mayulestes and didelphids (Muizon, 1998, 1999). Posterior to the level of the rim of the choanae, the palatine extends as a thin laminar bone forming the choanal walls. Only the left lacrimal is preserved. The facial process contacts the frontal and nasal dorsally, and the maxilla anteriorly; the zygomatic process contacts the jugal laterally, posteriorly, and medially (Figs. 2, 4, 6A). Unfortunately, the orbital process of the lacrimal is mostly broken. The facial process of the lacrimal is semicircular, with an anterior round border that extends to the level of the M1–M2 embrasure. Among other sparassodonts, Arctodictis and Borhyaena express the same anterior extension of the lacrimal as does the specimen described here, but in Cladosictis and Sipalocyon the lacrimal reaches the level of M1. In lateral view (Figs. 4, 6A), the rim of the orbit extends to the level of the middle of M2. An elongate lacrimal tuberosity is devel-

675

oped on the lateral face of the lacrimal bone, delimiting the orbital margin anteriorly. There is only one lacrimal foramen, which opens within the orbit. The presence of a lacrimal tuberosity and a single lacrimal foramen opening inside the orbit is also characteristic of other basal metatherians (Rougier et al., 1998). The zygomatic process of the lacrimal is small and rectangular, and scarcely contributes to the anterior base of the zygomatic arch. It extends back to the level of the posterior root of M2. Only the left zygomatic arch is preserved, but this element is broken on its most posterior portion. The anterior part of the zygomatic arch is formed by the zygomatic processes of the maxilla and lacrimal, as described above; the jugal forms its main portion, and the zygomatic process of the squamosal forms its posterior part. The jugal is a relatively flat medially curved bone. In lateral view (Figs. 4, 6A), this bone contacts the lacrimal dorsally and the maxilla anteriorly. The jugal-maxilla suture is obliquely oriented; the most anterior point of this suture reaches the M2 level, while the posterior one is located just dorsal to M4. The anterodorsal margin of the jugal is concave and forms the ventral rim of the orbit. Behind this area there is a low triangular protuberance located just below the postorbital process of the frontal bone (see below), where the postorbital ligament would be attached delimiting the orbit posteriorly (Evans, 1993). The ventral border of the jugal is concave. In lateral view (Figs. 4, 6A), dorsal and almost parallel to the ventral border, there is a low crest that limits the ventral one-third of this bone, corresponding to the area of attachment of the masseter muscle (pars profunda) (based on Didelphis; Turnbull, 1970). The frontal is the major component of the skull roof (Fig. 2) medial and immediately posterior to the orbits. In living marsupials, the frontals are two independently ossified dermal bones (Clark and Smith, 1993), but these components are co-ossified in the Quebrada Honda specimen; therefore, it is described as a single unit. In dorsal view (Fig. 2), the frontal contacts the nasal anteriorly and the lacrimal anterolaterally. The suture with the right and left nasals define: a wide ‘W’-shape with the lateral moieties longer than the medial one. Posteriorly, the frontal contacts the parietal by mean of an acute ‘V’-shaped suture. In lateral view (Figs. 4, 6A), the skull is fractured making it difficult to distinguish the sutures. Anteriorly, the frontal forms the concave dorsal orbital rim. The postorbital process is weaker than that of Cladosictis and Sipalocyon, but similar to Lycopsis, Borhyaena, and Arctodicits. Temporal crests arise from the postorbital processes and extend posteromedially joining just posterior to the postorbital processes and forming a sagittal crest. The sagittal crest continues posteriorly on the frontal and parietal bones becoming higher posteriorly. There are numerous small grooves and muscle scars on both sides of the sagittal crest that indicate the attachment area of the temporal musculature (based on Didelphis; Turnbull, 1970). Behind the frontal, there is only a single triangular bone on the posterior cranial roof (Fig. 2), which extends from the level of the postorbital constriction to the lamboidal crests. This bone is referred here as the parietal, but it could also represent the fusion of the interparietal and the parietals, as was assumed in Mayulestes (Muizon, 1998, 1999). A differentiation between the interparietal and parietal bones is clearly shown in some hathliacynids (e.g., Cladosictis and Sipalocyon), but in the other Sparassodonta examined (Borhyaena, Arctodictis, Lycopsis) there exists only a single bone. The sutures between bones forming the lateral wall of the skull and mesocranium are difficult to distinguish owing to the extensive damage on this part of the skull. The only feature clearly identifiable in the lateral wall of the skull is the foramen rotundum (Figs. 4, 6A); this is a circular foramen opening anteriorly. In living marsupials (Clark and Smith, 1993; Wible, 2003), as well as other metatherians (e.g., Mayulestes, Muizon, 1998), the fora-

676

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 26, NO. 3, 2006

men rotundum opens entirely in the alisphenoid; unfortunately the sutures in our specimen are not clearly delineated. Only the left squamosal bone has been preserved in the specimen from Quebrada Honda. The zygomatic process of the squamosal is a relatively lateral flat bridge of bone that contributes to the posterior portion of the zygomatic arch. This process is broken anteriorly and its contact with the jugal is not preserved. In ventral view (Fig. 3), and anterior to the glenoid fossa, there is a triangular area on the zygomatic process with irregular texture; this feature suggests that, as in other metatherians, the jugal extends back below the squamosal forming the preglenoid process. The squamous portion of the squamosal bone is roughly circular and located on the postero-ventral side of the skull (Figs. 4, 6A). It contacts the parietal dorsally, probably the alisphenoid anteriorly (although this suture is not clearly observable), and the occipital posteriorly. Unfortunately, one feature that is not observable in this specimen, but is noteworthy among sparassodonts, is the contact between the frontal and squamosal bones. As noted by Muizon (1999), hathliacynids (e.g., Sallacyon, Notogale, Cladosictis, and Sipalocyon), and didelphids in general, have an alisphenoid-parietal contact, whereas in other sparassodonts (e.g., Borhyaena, Prothylacynus, and Arctodictis), the two bones are separate and the frontal contacts the squamosal. The squamous process is poorly developed; by comparison, in Borhyaena and Arctodictis it reaches nearly to the sagittal crest. Near the suture with the occipital bone the squamosal becomes flatter, contributing to the formation of the most ventral portion of the lambdoidal crest. The subsquamosal foramen (sensu Muizon, 1998, 1999; suprameatal foramen sensu Wible, 2003) is large, oval, and opens posteriorly. It is located directly above the dorsal edge of the external acoustic meatus. The external acoustic meatus is a wide ‘U’-shaped excavation, ventrally concave, and developed entirely in the squamosal bone. It is bordered posteriorly by the post-tympanic process of the squamosal and anteriorly by the postglenoid process of the same bone. The post-tympanic process of the squamosal is a thick process that extends ventrally to the level of the basicranium in lateral view (Figs. 4, 6A); its most ventral portion is curved anteriorly making a contribution to the tympanic bulla. A similar anteriorly curved post-tympanic process occurs in other hathliacynids (Muizon, 1999; as pars mastoidea of the petrosal bone in Patterson, 1965), but is absent in borhyaenids where this process is almost vertical. The glenoid fossa (Fig. 3) is oval, with its transverse width approximately twice the anteroposterior length. It is bordered anteriorly by a weak preglenoid squamosal process and posteriorly by a prominent postglenoid process. As mentioned above, the preglenoid process of the jugal is not preserved in the specimen from Quebrada Honda, but it was probably present, as in other Sparassodonta. The postglenoid process is relatively narrow and high as in Cladosictis. The postglenoid process of Borhyaena and Arctodictis is considerable wider and lower than that of A. myctoderos, whereas that of didelphids (e.g., Didelphis, Monodelphis) is narrower and higher. The posterior wall of the postglenoid process forms a slightly convex and anteriorly inclined anterior wall of the external acoustic meatus (Fig. 3). A postglenoid foramen opens on the medial side of the postglenoid process, at the level of the medial-most edge of the glenoid fossa; this foramen is oval and faces posteroventrally. The glenoid process of the alisphenoid (sensu Wible, 2003, the triangular lateral expansion of the alisphenoid located at the anteromedial angle of the glenoid fossa, called entoglenoid process by Muizon, 1998, 1999, among others) is absent, as in other sparassodonts, and differing from Mayulestes and didelphids where it is clearly present. The basicranium is preserved in separate portions, one as part of the main fragment of the skull and the other associated with the occipital area (reconstructed in Figs. 3, 6A). This portion of

the skull is formed by the squamosal, alisphenoid, occipital, ectotympanic, and petrosal bones. The most conspicuous feature of the alisphenoid in the basicranium is its contribution to the formation of the tympanic bulla (Fig. 3). The tympanic process of the alisphenoid is a concave projection placed in the antero-ventral corner of the middle ear cavity. In some hathliacynids (e.g., Cladosictis) the tympanic process of the alisphenoid and the post-tympanic process of the squamosal contact to form the auditory bulla, and the ectotympanic is situated within the bulla (Patterson, 1965). In the specimen described here, a small fragment of the distal portion of the alisphenoid tympanic process is broken, leaving a small space ventrally where the ectotympanic is visible. The foramen ovale opens anterolateral to the alisphenoid tympanic process. This foramen is oval, enclosed entirely by the alisphenoid, and faces anteroventrally. There is no transverse canal (sensu Muizon, 1999; transverse canal foramen sensu Wible, 2003), whereas among sparassodonts a transverse canal is observed in Notogale, Sipalocyon, Lycopsis, Prothylacynus, and some specimens of Cladosictis. Anterior to the foramen ovale, there is a tiny foramen of uncertain homology. In ventral view (Fig. 3), in the most medial border of the alisphenoid, there are small shallow grooves that run posterolaterally–anteromedially from the level of the alisphenoid tympanic process to the level of the foramen ovale. In some metatherians, a groove medial to the alisphenoid tympanic process supports the anteromedial flange of the petrosal. Other grooves observed in the area in other metatherians connect the middle ear and the oral cavities via the auditory tube. The occipital bone is preserved separately from the rest of the skull. The occipital is formed by the supraoccipital, basioccipital, and exoccipital bones. The sutures among these bones are not observable in the specimen from Quebrada Honda, therefore, the occipital will be described as a single unit. The left occipital condyle (the only one preserved) is convex and oval with its main axis in the anterodorsal-posteroventral orientation. The entire surface of the condyle bears a smooth articular facet for the atlas. Anterior to the occipital condyle, there is a single large hypoglossal foramen, in contrast with the condition of two foramina typical of the crown group Marsupialia (SánchezVillagra, 1998, among others). A deep sulcus extends anterolaterally from the hypoglossal foramen. In better-preserved hathliacynids (i.e., Cladosictis and Sipalocyon) this sulcus is also conspicuous, bordered laterally by sharp crests, and reaches the level of the foramen for the inferior petrosal sinus. This sulcus is probably associated with a vein since the inferior petrosal sinus is venous and because a vein is transmitted with the hypoglossal nerve in marsupials (e.g., Didelphis and Monodelphis, Wible, 2003) as well as in other mammals (e.g., Homo, Jungers et al., 2003). In the lateral margin of the occipital bone, posterolateral to the anterior end of the described sulcus, there is a small notch that would correspond to the posteromedial border of a small jugular foramen. The size of this foramen suggests that, as occurs in living marsupials, the internal jugular vein was probably a minor conduit for drainage of the dural venous sinuses (Dom et al., 1970); its small size represents a plesiomorphic condition retained in metatherians (Rougier and Wible, 2006). The paracondylar process (sensu Evans, 1993; paraoccipital process sensu Muizon, 1998, 1999, among others) constitutes the ventral projection of the exoccipital lateral to the condyle (Wible, 2003); it is relatively wide and low. In this specimen, it occurs as in sparassodonts (clearly observed in Borhyaena, Arctodictis, Cladosictis, Sipalocyon, and Lycopsis) where the paracondylar process has the same ventral development as the post-tympanic process of the squamosal. It differs from didelphids (e.g., Didelphis and Monodelphis) where the paracondylar process is larger than the post-tympanic process. The posterior part of the skull is formed in metatherians gen-

FORASIEPI ET AL.—NEW HATHLIACYNID FROM BOLIVIA erally (e.g., Didelphodon, Mayulestes, didelphids; Clemens, 1966; Muizon, 1998; Wible, 2003, respectively) by the occipital, squamosal, and petrosal bones. In sparassodonts, the petrosal is excluded from the occiput and only the squamosal and occipital bones form this area (Muizon, 1999). The only exception is Callistoe, a proborhyaenid from the Eocene of Argentina, which according to Babot et al. (2002) shows the plesiomorphic condition. Unfortunately, in the present specimen the sutures in the occipital area are not clearly observable. The nuchal plane of the occiput exhibits numerous scars for the attachment of the cervical musculature (Evans, 1993), and small foramina, probably related to venous sinus drainage (Evans, 1993). A large mastoid foramen is observed in the nuchal plane above the occipital condyle. A dorsolateral portion of the lambdoidal crest has been preserved; this crest gives a roughly circular contour to the posterior view of the skull. The most conspicuous feature in posterior view is the foramen magnum. The right border is missing, but remaining parts suggest that it was oval with the main axis horizontal. The left ectotympanic is a ‘U’-shaped bone located in the inner side of the floor of the tympanic bulla (Figs. 4, 6A). The ectotympanic contacts the squamosal at both ends of the ‘U,’ medial to the postglenoid foramen. In addition to the squamosal contact, in other metatherians (e.g., Cladosictis, Didelphis, Monodelphis) the ectotympanic abuts the alisphenoid tympanic process; this specimen has a broken alisphenoid tympanic process, so contact between these two bones is assumed based on the proximity of the remaining preserved structures. In ventral view, the ectotympanic is roughly triangular with the anterior crus wider than the posterior one. As in other hathliacynids (Patterson, 1965; Muizon, 1999), the posterior crus contacts the posttympanic process of the squamosal (Figs. 4, 6A). In some living marsupials (e.g., Monodelphis; Wible, 2003) a different contact is formed with the posterior crus of the ectotympanic abutting the rostral tympanic process of the petrosal, which in turn forms the posterior wall of the tympanic bulla (Segall, 1969). Only the endocranial portion of a left petrosal is available for study (Fig. 7). The orientation of this element in the description is arbitrary. In recent marsupials, the main axis of the petrosal is placed 45º obliquely with respect to the main axis of the skull; this is assumed to be the position in Acyon myctoderos. The anterior expansion of the promontorium, the anteromedial flange (sensu Wible, 2003), is elongated and narrows anteriorly. The anteromedial flange is similar to Sipalocyon among sparassodonts, but different from Borhyaena and Lycopsis because it is extremely reduced. On the posterior edge of the anteromedial flange there is a groove that runs lateromedially (according to the orientation in this description) and is here interpreted as a sulcus for the inferior petrosal sinus. The internal acoustic meatus is a deep depression located in the anterior half of the petrosal. Two well-defined foramina open inside the internal acoustic meatus: the superior acoustic foramen on the lateral side and the larger inferior acoustic foramen on the medial one. No other internal structure within the internal acoustic meatus can be observed. On the medial margin, the petrosal has two small openings, the cochlear canaliculus and vestibular aqueduct. The cochlear canaliculus is an elongate fissure that opens in a depression at the level of the internal acoustic meatus. The vestibular aqueduct is smaller than the canaliculus and opens at the level of the posterior border of the subarcuate fossa. The crista petrosa (the crest that separates the middle and posterior crania fossae; Wible, 1990) is low and almost indistinct. The subarcuate fossa is deep, but less so than in Sipalocyon, and without a clearly defined dorsolateral border. A deep subarcuate fossa represents the primitive condition of Metatheria (Rougier et al., 1998; Sánchez-Villagra, 2002; Sánchez-Villagra and Wible, 2002). Dentary—The left dentary is almost complete, missing only the tip of the coronoid process. The right dentary is missing

677

almost the entire ascending ramus. The horizontal ramus of the dentary is long and slender (Figs. 5, 6B, C), similar to that of Acyon tricuspidatus, Lycopsis longirostrus, L. viverensis, and some specimens of Cladosicits. The ventral border of the dentary is slightly curved from the incisive alveoli to the condyle, as in other sparassodonts, except Prothylacynus in which the ventral and posterior borders form a right angle. The symphysis (Figs. 5B, 6C) is not as rugose as in other sparassodonts (i.e., it has relatively few and low ridges where the dentaries would have joined). This feature varies among sparassodonts from a relatively smooth surface present in hathliacynids in general, to the ossified symphysis found in some borhyaenids (e.g., Prothylacynus and Arctodictis munizi). The posterior-most portion of the symphysis extends to a point below the anterior border of the last premolar, as is typical for the Hathliacynidae (Marshall, 1981). In lateral view, (Figs. 5A, 6B) three mental foramina are observed: in the left dentary there is a large one below the posterior border of p1, the other two are below the middle of m1 and the posterior root of m2, respectively. In the right dentary the foramina are placed below the p1–p2 embrasure, below the posterior root of p3, and below the middle of m2. In the right dentary of the type of Acyon tricuspidatus there are at least two foramina: the largest and most anterior one opens at the level of p1, whereas the second opens at the level of the anterior root of m2. The number and relative position of mental foramina varies among species, specimens, and sometimes between the right and left dentary of the same individual. For example, many specimens of Cladosictis (MACN A 6289, A 5927, Sinclair, 1906:fig. LIX) have only two mental foramina, but the specimen illustrated by Sinclair (1906:fig. LVI) has four foramina. Prothylacynus has two (MACN A 5926) or three (Sinclair, 1906:fig. XLVII) foramina; Lycopsis longirostrus has six foramina; and the specimen MACN A 9342 of Borhyaena tuberata has three mental foramina in one dentary and four in the other. The coronoid process is broad anteroposteriorly and concave laterally where a relatively deep masseteric fossa is developed. The masseteric crest (posterior shelf of the masseteric fossa sensu Marshall and Muizon, 1995, and Wible, 2003) protrudes laterally defining the ventral and posterovental borders of the masseteric fossa. The anterior border of the coronoid process forms an angle of approximately 120º with respect to the horizontal axis of the dentary; this is a more obtuse angle than that of Cladisictis where it inclines approximately 110º. There is a relatively long retromolar space between the last molar and the anterior border of the coronoid process, larger than in Cladosictis. The condyle is oval and placed slightly above the alveolar margin in lateral view. The angular process is broken on its medial edge, but apparently was broad and inflected as in other sparassodonts (type two, intermediate angular process of Sánchez-Villagra and Smith, 1997). The mandibular foramen (Figs. 5B, 6C) is oval, faces posteriorly, and is located nearer to the base of the condyle than to the anterior base of the coronoid process; this is slightly more posterior than the condition observed in Cladosictis, Borhyaena, Arctodictis, and Prothylacynus. Dentition—The dental formula is 4.1.3.4/?3.1.3.4. (for measurements see Table). The upper incisor tooth row (Fig. 3) is almost straight following the morphology of the anterior border of the premaxilla. All the incisors have single roots. I2 and I4 are mesiodistally compressed. The crown of the distal-most upper incisor is approximately twice as large as those of similar sized I2–I3 (the upper central incisor is missing). The single-rooted upper canine is relatively slender, slightly larger and more procumbent than the lower one. Diastemata between the canine and P1 and between P1 and P2 are long and nearly equal in length. The diastema between P2 and P3 is nearly one-third that of the more mesial diastema. There is no diastema between P3 and M1. The two-rooted upper premolars are sharp, laterally compressed, and single cusped. The premolars increase in size posteriorly; P3

678

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 26, NO. 3, 2006

is much larger than P2. The main cusp of P1 is asymmetrical with the anterior crest almost straight and the posterior one concave; P2–P3 have a main cusp with symmetrically arranged anterior and posterior crests. An anterior cusp is absent on all upper premolars, but a minute posterior cuspule is present on P2–P3 (larger on P2). The upper molars (Fig. 8A) are arranged in size M1 < M2 < M3 >> M4. M1 to M3 have two labial and one lingual root, but M4 has only one labial and one lingual root. In M1–M3 the bases of the paracone and the metacone are twinned, as in most sparassodonts, being more widely separated only in the most plesiomorphic tooth pattern of the group (i.e., Patene, Allqokirus, and Mayulestes). The metacone is the largest cusp on M1–M3. It increases abruptly in size posteriorly. In contrast, on M4 the metacone is extremely reduced, and is represented as a small prominence on the posterior slope of the paracone. On all the molars the protocone is much lower than the metacone and the paracone. The protocone is distinctly basined and broader than in Cladosictis. In M1–M3 the protocone is lingual to the paracone; in contrast, in M4 it is posterolingual to the paracone. The postmetacrista is strongly developed, a characteristic feature of carnivorous mammals (Muizon and Lange-Badré, 1997). The length of the postmetacrista increases from M1 to M3, and changes from an almost straight crest (in occlusal view) on M1 to a posteriorly convex one on M3. M4 lacks any vestige of a postmetacrista. There is no carnassial notch visible on the postmetacrista of M1–M3, but this absence may be related to the advanced stage of wear. A stylar shelf is absent in M1–M2 and M4, and is only present on M3 as a posterolabial lobe. The parastylar corner is rounded in all the molars; there is not an individual stylar cusp, but a crest surrounds the anterolabial corner of the tooth. The only stylar cusp recognizable is the cusp B on M2 where it forms a small swelling posterolabial to the paracone. Cusp B on M3 forms an ectocingulum (sensu Marshall, 1978) continuous with the crest running mesially from the parastyle. The crowns of the lower incisors are not preserved. There are only two mesiodistally compressed incisor roots preserved. A third lower incisor is postulated based on the condition observed in other hathliacynids. The lower canines (Fig. 5) are slender, long, and vertically oriented; they are suboval in cross section and single rooted. Diastemata are present between the canine and p1 and between successive premolars. These diastemata are

FIGURE 8. Acyon myctoderos, MNHN-Bol-V-003668. Stereopair photographs of the occlusal view of the left upper (A) and right lower (B) molars. Scale bar equals 5 mm.

longer than in Acyon tricuspidatus. The longest diastema is between c and p1; these spaces diminish in length along the premolar series. The premolars are strongly laterally compressed and become progressively taller and longer from p1 to p3. All premolars are two rooted and the main axis is aligned with the anteroposterior axis of the dentary, as in other hathliacynids. In other sparassodonts (e.g., Prothylacynus) p1 is slightly oblique with the anterior root slightly labial and the posterior root slightly lingual. This condition is most extreme in borhyaenids (e.g., Arctodictis), where p1 is situated transverse to the anteroposterior axis of the dentary. As in the upper dentition, p1 has an asymmetrically arranged main cusp, while in p2–p3 this cusp is almost symmetrical. The p1 lacks an anterior cusp, but has a long posterior portion that ends in a minute cuspule. Premolars p2–p3 have small anterior and posterior cuspules; the posterior cuspules are larger than the anterior ones. Acyon myctoderos differs from A. tricuspidatus in having p2 comparatively more stout with a better-developed posterior cusp. The molars increase in size posteriorly (Fig. 8B), but not as abruptly as in borhyaenids. The trigonids have only two cusps, the paraconid and protoconid; there is no vestige of a metaconid, as in other hathliacynids. The paraconid increases sharply in size along the tooth row, from a minute cusp in m1 to a taller one in m4. The anterolingual corner of the paraconid of m2–m4 is flat and protrudes anteriorly, forming the interlocking mechanism of the lower molars. The lower first molar lacks this feature and has the paraconid rounded anteriorly. There is an extremely reduced anterobasal cingulum in all molars, restricted to the base of the paraconid, and even more reduced than in Acyon tricuspidatus. The protoconid is the largest cusp of the molars; it is roughly circular in cross section on m1–m2 and more triangular on m3–m4. The paracristid and preprotocistid are particularly sharp on m3–m4, with a carnassial notch developed between them. The paraconid, protoconid, and talonid are aligned in m1; in the other molars they form an obtuse angle. The talonid is present on all the molars. Lower first to third molars have wide talonids not fully basined due to the absence of an entoconid, as do the other species of the genus. Only the right m2 displays a shallow crest at this point that is reminiscent of a talonid border. The talonid of m4 is considerably reduced; a sharp crest surrounds the lingual and labial borders forming a complete basin. The hypoconid is clearly distinguished on m1–m3 at the buccal edge of the talonid (m4 has a crest in this position). The hypoconids are larger than in Acyon tricuspidatus; this feature is clearly shown on the talonid of m2. The hypoconulid in all the molars is relatively large and quite vertical, differing in this regard from the posteriorly projecting hypoconulids of Acyon tricuspidatus. Postcrania—The atlas (Fig. 9A, B) is robust, with the intercentrum fused to the rest of the vertebra as in Prothylacynus and Cladosictis. The anterior and posterior borders of the neural arch are straight and there is no neural spine. The atlantal foramen (Fig. 9A) is limited anteriorly by a bony bridge as in Sipalocyon, Cladosictis, Prothylacynus, and Lycopsis. In dorsal view (Fig. 9A), there is a foramen on the posterior portion of the dorsal arch, which probably transmitted a small branch of the vertebral artery. This foramen is also present in Sipalocyon and Borhyaena, but it is absent in Prothylacynus, Arctodictis, and probably Mayulestes (Muizon, 1998:fig. 10). The neural canal is rounded, similar to other Sparassodonta. The anterior articular facet for the occipital condyle is strongly concave and faces anteromedially; its dorsal border is distinctly curved and nearly reaches to the level of the neural arch. In dorsal view, the lateral border of the condylar facet is almost parallel to the anteroposterior axis of the vertebra. The posterior articular facet for the axis articulation is flatter than those for the condyles and faces posteromedially, as in other Sparassodonta. In ventral view (Fig. 9B), the axial facets of both sides form a wide ‘U’-shaped concavity that

FORASIEPI ET AL.—NEW HATHLIACYNID FROM BOLIVIA

FIGURE 9. Acyon myctoderos, MNHN-Bol-V-003668. Photographs of the atlas in dorsal (A), and ventral views (B), and the axis in lateral (C) and dorsal (D) views. Scale bar equals 20 mm.

enclosed the anterior portion of the axis. The transverse process has a roughly rounded outline; its anterior edge does not extend beyond the level of the atlantal foramen and its posterior edge reaches at least the level of the axial facet. A small foramen opens at the junction of the transverse process and the neural arch on both sides of the vertebra. The left foramen opens at the level of the anterior border of the transverse process, while

FIGURE 10. Cladogram showing the relationships of the sparassodonts considered in this study. Acyon represents a hathliacynid related closer to Cladosictis than to other Sparassodonta. The numbers indicate the nodes; the diagnosis of the nodes is based on unequivocal characters (character states, between parentheses, follow the character number): node 1: 3(1), 5(1), 6(1), 8(1), 10(1), 11(1), 12(1), 13(1), 17(1), 20(1), 29(1), 30(1), 36(1), 42(1), 44(1); node 2: 31(1), 32(1); node 3: 18(1), 33(1), 34(1), 37(1), 45(2); node 4: 16(1), 23(1), 36(2), 38(1), 41(1); node 5: 15(1), 22(1); nodes 6 and 7 are supported by an ambiguous character, see the discussion, and node 8: 39(2).

679

the right one opens at the level of its mid-portion. There is no transverse foramen. The axis (Fig. 9C, D) is relatively long anteroposteriorly. The neural arch is broken anteriorly, including the spinous process. This process is elongate in Sparassodonta (Sinclair, 1906; Muizon, 1998; Argot, 2003, 2004). The postzygapophyses are rounded and protrude posteriorly. The articular facets are oval and face posterolaterally. The odontoid process is broken and only its robust base is preserved. In dorsal view (Fig. 9D), there are two small foramina immediately posterior to the odontoid process. In one specimen of Prothylacynus (MACN A 5931– 5937), these foramina are similar in size to those of the specimen described here, but in another specimen (MACN A 706–720) the foramina are larger, with a median crest between them. As in extant marsupials (e.g., Didelphis), these differences may actually represent individual variation (pers. obs.). In ventral view, the anterior border of the axis, which contains the prezygapophyses, is rounded, even more than in other sparassodonts. The prezygapophyses and the dens have a clear border between them, and do not form a continuous surface as in some marsupials (character 12 of Horovitz and Sánchez-Villagra, 2003). One feature present in the Quebrada Honda specimen that is shared with other sparassodonts is the strongly developed ventral median crest of the axis (Fig. 9C), which increases in height posteriorly (Sinclair, 1906; Argot, 2003, 2004). This crest is present; the axis and the remaining anterior cervical vertebrae (see below). The suture between the axial and atlantal (anterior cotyles and odontoid) portions of the axis is partially visible; in dorsal view, it describes a wide ‘V’ opening anteriorly, but in ventral view it shows a wide ‘U’ opening posteriorly. This sutural morphology is like that of other marsupials (e.g., Didelphis). The transverse process is largely damaged and the transverse foramen is not preserved. Four other partially preserved vertebrae have been recovered; they are mainly represented by fragmentary bodies. The bodies are robust and anteroposteriorly elongate. On their dorsal surfaces open one or two foramina of relatively large caliber. Three of the elements are possible cervical vertebrae due to the strong ventral median crest. This crest ends in a bifid tubercle in two of the three preserved vertebral bodies. A median crest is commonly present in the cervicals of Sparassodonta, ending in a large single or bifid posterior tubercle. The anterior and posterior surfaces of the body are inclined anteriorly, as occurs in general in the cervicals of mammals (Slijper, 1946). The anterior surfaces are rectangular in shape as in Prothylacynus. The fourth and smallest vertebra has only a small part of the body preserved; its identification is difficult due to its fragmentary condition. The ventral side of the body has two parallel crests running anteroposteriorly. Three tarsal elements have been recovered, including a right cuboid, right ectocuneiform, and, probably, an entocuneiform. In anterior view the cuboid has the medial border longer than the lateral one, and the proximal articular surface (calcaneal facet) convex and oblique with respect to the distal one. This latter feature, present also in other sparassodonts (Sinclair, 1906; Argot, 2003, 2004), modifies the otherwise strictly cubic form. The anterior face of the cuboid is flat, but the posterior face bears a strong plantar tubercle. The distal articular surface has a smaller lateral facet for the fifth metatarsal and a larger medial facet for the fourth metatarsal; there is no crest separating the facets. The ectocuneiform is laterally flat and rectangular in cross section. The plantar tubercle is well developed near the dorsal border of the plantar surface. In medial and lateral views, the bone is roughly quadrangular. On the proximal portion of both sides it has articular facets: in medial view, a kidney-shaped facet for the articulation with the mesocuneiform, and in lateral view, a rectangular facet for the articulation with the cuboid. In distal

680

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 26, NO. 3, 2006

view the entire surface of the ectocuneiform has a concave rectangular facet for the articulation with metatarsal III. The third preserved element, here identified as a probable entocuneiform, is high, flat, and triangular in cross section. The distal articular facet for metatarsal I is oblique. These features are similar to the entocuneiform of Lycopsis longirostrus, and this is the basis for its identification. Five proximal portions of metapodials are preserved; the elements are relatively slender. Compared to those of Arctodictis sinclairi, the preserved metapodials of the specimen from Quebrada Honda are median ones: they lack the prominent medial and lateral processes characteristic of metapodials I and V, respectively, both in the hand and the foot. A more precise identification can not be given at present. PHYLOGENETIC ANALYSIS To examine the phylogenetic relationships of Acyon myctoderos and test the monophyly of Hathliacynidae, we conducted a parsimony analysis of 47 characters and ten taxa using the program NONA version 2.0 (Goloboff, 1993). Mayulestes ferox, from the Paleocene of Bolivia, is used for outgroup comparison following Muizon (1998, 1999). The ingroup is composed of Sallacyon, Notogale, Cladosictis, Sipalocyon, Lycopsis, Prothylacynus, Borhyaena, Arctodictis, and Acyon. The coding of these taxa or the modification of previous scores is based on the specimens listed in Appendix 1. Mayulestes was considered a member of the Sparassodonta (Borhyaenoidea) by Muizon (1998, 1999, among others). Nevertheless, in an analysis of mammaliaform taxa, Rougier et al. (1998) found the sparassodonts (as Borhyaenidae) to be basal to Mayulestes, the latter appearing as a sister group to a clade including Jaskhadelphys, Pucadelphys, Andinodelphys, and the crown group Marsupialia. Whatever the relationship of Mayulestes to sparassodonts, we include this taxon as the outgroup because it has one of the best-preserved skeletons of a taxon phylogenetically close to the clade of interest in this analysis. Only one outgroup was selected because the monophyly of the ingroup is assumed based on previous studies (e.g., Marshall et al., 1990; Muizon, 1999). Twenty ambiguous and unambiguous character states differentiate Mayulestes from the ingroup (characters 3, 5, 6, 8, 10–13, 17, 20, 25–27, 29, 30, 36, 39, 42, 44, 46, Appendix 2). Since Mayulestes could be more closely related to the crown group Marsupialia than to sparassodonts (Rougier et al., 1998), these characters may not be synapomorphies of Sparassodonta. For example, the absence of a postpalatine torus (character 12) and a glenoid process of the alisphenoid (character 17), shown as derived for the sparassodonts in this tree, could be plesiomorphic for the group (see Rougier et al., 1998). The character matrix (Appendices 2 and 3) is based largely on that of Muizon (1999). We took from his matrix all the characters that vary among sparassodonts or between Mayulestes and the other sparassodonts (27 characters). Other important references are Rougier et al. (1998) and Horovitz and Sánchez-Villagra (2003). The scores of the taxa were modified only in five cases. First, the precanine notch was coded as present in Prothylacynus and Sipalocyon by Muizon (1999, character 33), but based on our observations was scored as absent. Second, the ectotympanic was supposed to be enclosed by the squamosal in Borhyaena and Prothylacynus based on the presence of ridges and grooves (character 4 of Muizon, 1999; Archer, 1976). However, on the specimens we observed there is a wide area medial to the external acoustic meatus in the squamosal bone in which we assume the ectotympanic was only attached in life by ligaments (i.e., it is not ventrally support by bone). Third, the transverse canal was scored as present in Cladosictis by Muizon (1999, character 8), but in the only specimen examined by us it is absent. It is well known that some metatherians are polymorphic for the pres-

ence/absence of a transverse canal (Marshall 1977b; SánchezVillagra and Wible, 2002), and this may also be the case in Cladosictis. The transverse canal was scored absent in Sipalocyon and Prothylacynus by Muizon (1999). However, it was reported to be present by Archer (1976) in both taxa and it is also present in the specimen of Prothylacynus we examined; for that reason it is scored here as present. Fourth, the foramen ovale of Borhyaena was considered confluent with the foramen lacerum medium (foramen for greater petrosal nerve sensu Wible, 2003) by Muizon (1999, character 7). Here we considered only the conformation of the foramen ovale, and for that reason Borhyaena is coded as for Mayulestes. Another independent character should consider the variation of the foramen lacerum medium (not included in this study until more specimens are analyzed). Finally, the incisor row was codified as transverse in Sipalocyon by Muizon (1999, character 36), but based on our observations we code it as parabolic. The characters in the matrix are organized as follows: 1–32, skull; 33–34, dentary; 35–45, teeth; 46–47, postcranial skeleton. Forty-one characters are binary, while six characters are multistate; five of these are ordered (characters 7, 33, 36, 42, and 45) and one unordered (character 39). All the characters are given equal weight. The data matrix was analysed with NONA version 2.0 (Goloboff, 1993); this program of parsimony was used for heuristic searches. The Wagner trees obtained were subject to branch swapping with bisection-reconnection. The analysis resulted in one tree (Fig. 10) with a length of 71 steps, CI ⳱ 0.74, and RI ⳱ 0.74. There are two groups. One is formed by the Borhyaenidae (sensu Marshall et al., 1990), supported by two unambiguous characters: the lack of a crest separating the anterior border of the epitympanic recess from the posterior region of the alisphenoid hypotympanic sinus (character 31) and the presence of a shallow subarcuate fossa (character 32). The second group is formed by the Hathliacynidae, supported by two unambiguous characters: the presence of pneumatisation in the squamosal bone (character 15) and an ectotympanic enclosed posteriorly by the squamosal (character 22). Lycopsis is often considered a member of the Prothylacyninae (Marshall, 1979; Marshall et al., 1990); nevertheless, in this analysis and in that of Muizon (1999), it is placed as the sister group of Prothylacynus + Borhyaeninae. The inclusion of another species, ordinarily considered as belonging to Prothylacyninae, should test the monophyly of this subfamily. The general relationship of the ingroup is very similar to that presented by Muizon (1999), without considering the proborhyaenids and thylacosmilids, and differing only in the arrangement of the hathliacynids. In that paper, Sallacyon is the sister taxon of a natural group formed by (Sipalocyon (Cladosictis-Notogale)). In our most parsimonious tree, Sallacyon appears also at the base of the Hathliacynidae, but Sipalocyon and Notogale are sister taxa, and Cladosictis is the sister taxon of Acyon. The group formed by Sipalocyon and Notogale is supported by only one ambiguous character, the presence of the transverse canal (character 24); whereas Cladosictis plus Acyon are grouped by one unambiguous character, stylar cusp B on M2 extremely reduced to absent (character 39), and by three ambiguous features, a long skull (character 1), a conspicuous precanine notch (character 4), and the intercentrum of the atlas fused to the rest of the vertebra (character 46). Finally, Acyon shows one autapomorphic, unambiguous character, the presence of two cusps (i.e., hypoconid and hypoconulid) in the talonid of the lower teeth (character 45). As a final remark we note that the Hathliacynidae, as well as their internal branches, have low branch support (Bremer support of 1 in nodes 5, 6, and 7; and 2 in node 8) and are justified only by a few mostly ambiguous features in either analysis (the present review and Muizon, 1999); therefore, the topology of this family could be unstable.

FORASIEPI ET AL.—NEW HATHLIACYNID FROM BOLIVIA CONCLUSIONS Since the end of the Nineteenth century, the hathliacynids have been consistently considered as a monophyletic group (e.g., Ameghino, 1894). Recent cladistic reviews also support this grouping of the family (Marshall et al., 1990; Muizon, 1999, among others). Hathliacynids are also recognized as a natural group in our analysis by two derived, unambiguous characters, the presence of pneumatization in the squamosal bone and an ectotympanic enclosed posteriorly by the squamosal. The discovery of an almost-complete skull and associated postcranial elements of a new species, Acyon myctoderos, from the middle Miocene beds (Laventan SALMA) of Quebrada Honda (Bolivia), prompted the analysis of the systematics of this genus and its inclusion in a phylogenetic framework. Acyon myctoderos is the largest known hathliacynid. Differences with the type species of the genus, A. tricuspidatus, include longer diastemata among premolars, p2 comparatively more robust with a better developed posterior cusp and with a sharp anterior crest, lower molars with a more poorly developed anterobasal cingulum, m1–m3 with hypoconulids less salient posteriorly and more vertically oriented, and larger hypoconids at least on m2. Our phylogenetic analysis, including nine taxa of Sparassodonta and Mayulestes as the outgroup, showed that Acyon is more closely related to Cladosictis than to any other hathliacynid. Acyon and Cladosictis share a long skull, a conspicuous precanine notch, the stylar cusp B on M2 being extremely reduced to absent, and an intercentrum of the atlas fused to the rest of the vertebra. ACKNOWLEDGMENTS We thank A. Kramarz (MACN), M. Reguero and S. Bargo (MLP), and W. Simpson (FMNH) for access and assistance to specimens under their care; J. R. Wible, A. Candela, and A. Winkler for the extensive review of the manuscript; G. W. Rougier and A. G. Martinelli for comments on the original manuscript; Shawn Zack for help with fossil preparation; M. Hohloch for help with photographic work; J. Blanco for help with the figures; F. Anaya-Daza, R. Mamani, and B. Mamani for their cooperation in the field work in Quebrada Honda. Work of AMF was supported by CONICET and NSF (grant DEB 010971 to G. W. Rougier); MRSV by the Deutsche Forschungsgemeinschaft (SA 883-4-1); MT and NS by the Overseas Scientific Research Funds (no. 07041136) from the Ministry of Education, Science and Culture of Japan; and RFK by the NSF (grant EAR00-87636). LITERATURE CITED Ameghino, F. 1887. Enumeración sistemática de las especies de mamíferos fósiles coleccionados por Carlos Ameghino en los terrenos eocenos de la Patagonia austral y depositados en el Museo de La Plata. Boletín del Museo de La Plata 1:1–26. Ameghino, F. 1891. Nuevos restos de mamíferos fósiles descubiertos por Carlos Ameghino en el Eoceno inferior de la Patagonia austral— especies nuevas, adiciones y correcciones. Revista Argentina de Historia Natural 1:289–328. Ameghino, F. 1894. Enumération synoptique des espécies de mammifères fossiles des formations éocénes de Patagonia. Boletín de la Academia de Ciencias de Córdoba 13:259–452. Archer, M. 1976. The basicranial region of marsupicarnivores (Marsupialia), relationships of carnivorous marsupials, and affinities of the insectivorous marsupial peramelids. Zoological Journal of the Linnean Society 59:217–322. Argot, C. 2003. Postcranial functional adaptations in the South American Miocene borhyaenoids (Mammalia, Metatheria): Cladosictis, Pseudonotictis and Sipalocyon. Alcheringa 27:303–356. Argot, C. 2004. Evolution of South American mammalian predators

681

(Borhyaenoidea): anatomical and paleobiological implications. Zoological Journal of the Linnean Society 140:487–521. Babot, M. J., J. E. Powell, and C. de Muizon. 2002. Callistoe vincei, a new Proborhyaenidae (Borhyaenoidea, Metatheria, Mammalia) from the early Eocene of Argentina. Géobios 35:615–629. Barghusen, H. R. 1986. On the evolutionary origin of the therian tensor veli palatini and tensor tympani muscles; pp. 253–262 in N. Hotton, P. D. MacLean, J. J. Roth, and E. C. Roth (eds.), The Ecology and Biology of Mammal-like Reptiles. Smithsonian Institution Press, Washington, D.C. Berggren, W. A., D. V. Kent, C. C. Swisher, III, and M. P. Aubry. 1995. A revised Cenozoic geochronology and chronostratigraphy; in W. A. Berggren, D. V. Kent, M. P. Aubry, and J. Hardenbol (eds.), Geochronology, Time Scales and Global Stratigraphic Correlation. Society Economic Paleontologists and Mineralogists, Special Publication 54:129–212. Cabrera, A. 1927. Datos para el conocimiento de los dasyuroideos fósiles argentinos. Revista del Museo de La Plata 30:271–315. Clark, C. T., and K. K. Smith. 1993. Cranial osteogenesis in Monodelphis domestica (Didelphidae) and Macropus eugenii (Macropodidae). Journal of Morphology 215:119–149. Clemens, W. A. 1966. Fossil mammals of the type Lance Formation Part II—Marsupialia. University of California Publications in Geological Sciences 62:1–122. Dom, R. B., L. Fisher, and G. F. Martin. 1970. The venous system of the head and neck of the opossum (Didelphis virginiana). Journal of Morphology 132:487–496. Evans, H. E. 1993. Miller’s Anatomy of the Dog (3rd. edition) W. B. Saunders, Philadelphia, Pennsylvania, 1113 pp. Goloboff, P. A. 1993. NONA version 2.0. Published by the author, Tucumán, Argentina. Hershkovitz, P. 1982. The staggered marsupial lower third incisor (i3). Géobios, mémoire spécial 6:191–200. Hoffstetter, R. 1977. Un gisement de mammifères miocènes à Quebrada Honda (sud Bolivien). Comptes Rendus de l’Académie des Sciences, Paris, Série D 284:1517–1520. Hoffstetter, R., and G. Petter. 1983. Paraborhyaena boliviana et Andinogale sallensis, deux marsupiaux (Borhyaenidae) nouveaux du Déséadien (Oligocène inférieur) de Salla (Bolivie). Comptes Rendus de l’Académie des Sciences, Paris, Série II 296:205–208. Horovitz, I., and M. R. Sánchez-Villagra. 2003. A morphological analysis of marsupial mammal higher-level phylogenetic relationships. Cladistics 19:181–212. Huxley, J. S. 1880. On the application of the law of evolution to the arrangement of the Vertebrata, and more particular to the Mammalia. Proceedings of the Zoological Society of London 43:649–662. Jungers, W. L., A. A. Pokempner, R. F. Kay, and M. Cartmill. 2003. Hypoglossal canal size in living hominids and the evolution of human speech. Human Biology 74:473–484. MacFadden, B. J., and R. G. Wolff. 1981. Geological investigations of late Cenozoic vertebrate-bearing deposits in southern Bolivia. Anales del II Congreso Latinoamericano de Paleontología 2:765–778. MacFadden, B. J., F. Anaya, H. Perez, C. W. Naeser, P. K. Zeitlaer, and K. E. Campbell. 1990. Late Cenozoic paleomagnetism and chronology of Andean basins of Bolivia: evidence for possible oroclinal bending. Journal of Geology 98:541–555. Madden, R. H., J. Guerrero, R. F. Kay, J. J. Flynn, C. C. Swisher, III, and A. H. Walton. 1997. The Laventan Stage and Age; pp. 499–519 in R. F. Kay, R. H. Madden, R. L. Cifelli, and J. J. Flynn (eds.), Vertebrate Paleontology in the Neotropics—The Miocene Fauna of La Venta, Colombia. Smithsonian Institution Press, Washington, D. C. Marshall, L. G. 1977a. Evolution of the carnivorous adaptative zone in South America; pp. 709–722 in M. K. Hecht, P. C. Goody, and B. M. Hecht (eds.), Major Patterns in Vertebrate Evolution. Plenum Press, New York, New York. Marshall, L. G. 1977b. Cladistic analysis of borhyaenoid, dasyuroid, didelphoid, and thylacinid (Marsupialia, Mammalia) affinity. Systematic Zoology 26:410–425. Marshall, L. G. 1977c. A new species of Lycopsis (Borhyaenidae, Marsupialia) from the La Venta fauna (late Miocene) of Colombia, South America. Journal of Paleontology 51:633–642. Marshall, L. G. 1978. Evolution of the Borhyaenidae, extinct South American predaceous marsupials. University of California Publications in Geological Sciences 117:1–89. Marshall, L. G. 1979. Review of the Prothylacyninae, an extinct subfam-

682

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 26, NO. 3, 2006

ily of South American ‘dog-like’ marsupials. Fieldiana Geology New Series 3:1–50. Marshall, L. G. 1981. Review of the Hathlyacyninae, an extinct subfamily of South American ‘dog-like’ marsupials. Fieldiana Geology New Series 7:1–120. Marshall, L. G., and C. de Muizon. 1995. Part II: the skull; pp. 21–90 in L. G. Marshall, C. de Muizon, and D. Sigogneaou-Russel (eds.), Pucadelphys andinus (Marsupialia, Mammalia) from the early Paleocene of Bolivia. Mémoires du Muséum National d’Histoire Naturelle 165. Marshall, L. G., and T. Sempere. 1991. The Eocene to Pleistocene vertebrates of Bolivia and their stratigraphic context: a review; pp. 631–652 in R. Suarez-Soruco (ed.), Fósiles y Facies de Bolivia, Volúmen I, Vertebrados. Revista Técnica de Yacimientos Petrolíferos Fiscales Bolivianos 12. Marshall, L. G., J. A. Case, and M. O. Woodburne. 1990. Phylogenetic relationships of the families of marsupials. Current Mammalogy 2: 433–502. Mercerat, A. 1891. Caracteres diagnósticos de algunas especies de Creodonta conservadas en el Museo de La Plata. Revista del Museo de La Plata 2:51–56. Muizon, C. de 1998. Mayulestes ferox, a borhyaenoid (Metatheria, Mammalia) from the early Paleocene of Bolivia—phylogenetic and palaeobiologic implications. Geodiversitas 20:19–142. Muizon, C. de 1999. Marsupial skulls from the Deseadan (Late Oligocene) of Bolivia and phylogenetic analysis of the Borhyaenoidea (Marsupialia, Mammalia). Géobios 32:483–509. Muizon, C. de, and B. Lange-Badré. 1997. Carnivorous dental adaptations in tribosphenic mammals and phylogenetic reconstruction. Lethaia 30:351–366. Patterson, B. 1965. The auditory region of the borhyaenid marsupial Cladosictis. Breviora 217:1–9. Paula Couto, C. de. 1952. Fossil mammals from the beginning of the Cenozoic in Brazil—Marsupialia: Polydolopidae and Borhyaenidae. American Museum Novitates 1559:1–27. Rougier, G. W., and J. R. Wible. 2006. Major changes in the ear region and basicranium of early mammals; in M. Carrano, T. J. Gaudin, R. Blob, and J. R. Wible (eds.), Amniote Paleobiology: Phylogenetic and Functional Perspectives on the Evolution of Mammals, Birds and Reptiles. University of Chicago Press, Chicago, Illinois. 88:269–311. Rougier, G. W., J. R. Wible, and M. J. Novacek. 1998. Implications of Deltatheridium specimens for early marsupial history. Nature 396: 459–463. Sánchez-Villagra, M. R. 1998. Patterns of morphological change in the ontogeny and phylogeny of the marsupial skull. Ph.D. dissertation, Duke University, Durham, North Carolina, 536 pp. Sánchez-Villagra, M. R. 2002. The cerebellar paraflocculus and the subarcuate fossa in Monodelphis domestica and other marsupial mammals—the ontogeny and phylogeny of a brain-skull interaction. Acta Theriologica 47:1–14. Sánchez-Villagra, M. R., and K. K. Smith. 1997. Diversity and evolution of the marsupial mandibular angular process. Journal of Mammalian Evolution 4:119–144. Sánchez-Villagra, M. R., and J. R. Wible. 2002. Patterns of evolutionary transformation in the petrosal bone and some basicranial features in marsupial mammals, with special reference to didelphids. Journal of Zoological Systematics and Evolutionary Research 40:26–45. Segall, W. 1969. The auditory ossicles (malleous, incus) and their relationships to the tympanic: in marsupials. Acta Anatomica 73: 176–191. Simpson, G. G. 1980. Splendid Isolation—The Curious History of South American Mammals. Yale University Press, New Haven, Connecticut, 266 pp. Sinclair, W. J. 1906. Mammalia of the Santa Cruz beds: Marsupialia. Reports of the Princeton University Expeditions to Patagonia 4: 333–460. Slijper, E. J. 1946. Comparative biologic-anatomical investigations on the vertebral column and spinal musculature of mammals. Koninklijke Nederlandse Akademie van Wetenschappen, Apfling Natuurkunde 42:1–128. Troëng, B., H. Claure, L. Oliveira, R. Ballón, and G. Walser. 1993. Mapas Temáticos de Recursos Minerales de Bolivia, Hojas Tarija y Villazón, Bolivia. 1:250,000. Boletín del Servicio Geológico de Bolivia (especial) 3:1–178.

Turnbull, W. D. 1970. Mammalian masticatory apparatus. Fieldiana Geology 18:149–356. Villarroel, C., and L. G. Marshall. 1982. Geology of the Deseadan (early Oligocene) age Estratos Salla in the Salla-Luribay Basin, Bolivia with description of new Marsupialia. Géobios, mémoire spécial 6: 201–211. Villarroel, C., and L. G. Marshall. 1983. Two new late Tertiary marsupials (Hathlyacyninae and Sparassocyninae) from the Bolivian Altiplano. Journal of Paleontology 57:1061–1066. Wible, J. R. 1990. Petrosals of Late Cretaceous marsupials from North America, and a cladistic analysis of the petrosal in therian mammals. Journal of Vertebrate Paleontology 10:183–205. Wible, J. R. 2003. On the cranial osteology of the short-tailed opossum Monodelphis brevicaudata (Didelphidae, Marsupialia). Annals of Carnegie Museum 72:137–202. Submitted 15 September 2004; accepted 15 April 2006. APPENDIX 1. Taxa included in the phylogenetic analysis. Specimen numbers indicate original material examined in this study; all other data were derived from the literature sources indicated in parentheses. Sipalocyon gracilis: MACN A 691–703 (single specimen), right and left dentaries, left maxilla and premaxilla, fragmentary postcranial remains; MACN A 5938–5949 (single specimen), right and left dentaries, fragment of maxilla, fragmentary postcranial remains; both specimens from Santa Cruz Province, Argentina, Santacrucian SALMA (Sinclair, 1906; Marshall, 1981). Cladosictis patagonica: MACN A 5927–5929 (single specimen), skull, right dentary, fragmentary postcranial remains; from Santa Cruz Province, Argentina, Santacrucian SALMA (Sinclair, 1906; Marshall, 1981). Borhyaena tuberata: MACN A 5922, fragmentary skull; MACN A 6203–6265 and 10585 (single specimen), fragments of both dentaries, maxillae, postcranial remains; MACN A 9344, complete skull; MACN A 11260 postcranial elements; all specimens from Santa Cruz Province, Argentina, Santacrucian SALMA (Sinclair, 1906; Marshall, 1978). Prothylacynus patagonicus: MACN A 706–720 (single specimen), left dentary and anterior part of right dentary, left maxilla, postcranial remains; MACN A 5931–5937 (single specimen), complete skull, broken right dentary, postcranial remains; both specimens from Santa Cruz Province, Argentina, Santacrucian SALMA (Sinclair, 1906; Marshall, 1978). Arctodictis sinclairi: MLP 85-VII-3-1, nearly complete skeleton; from Chubut Province, Argentina, Colhuehuapian SALMA. Arctodictis munizi: CORD PZ 1210, left dentary and anterior part of right dentary, skull, postcranial remains; from Santa Cruz Province, Argentina, Santacrucian SALMA. Lycopsis viverensis: MNH 87-6-1, left maxilla; MNH 95-6-1, right dentary; both specimens from Buenos Aires Province, Argentina, Chasicoan SALMA. Lycopsis longirostrus: UCMP 38061, nearly complete skeleton; from La Venta, Colombia, Laventan SALMA (Marshall, 1977c). The scorings of Sallacyon hoffstetteri, Notogale mitis, and Mayulestes ferox are derived exclusively from the literature (Marshall, 1981; Villarroel and Marshall, 1982; Hoffstetter and Petter, 1983; Muizon, 1998, 1999). APPENDIX 2. Characters and character state descriptions used in the phylogenetic analysis. (1) Length of skull: anteroposterior length of skull shorter than twice width of skull (0); anteroposterior length of skull longer or as long as twice width of skull (1). (2) Length of braincase (Muizon, 1999; character 26): braincase slightly wider than long (0.7 < L/W < 0.9) (0); braincase much longer than wide (L/W < 0.6) (1). ‘L’ is distance between lambdoidal crest at height of subsquamosal foramen and medial-most point of temporal fossa; ‘W’ is width of skull at level of posterior border of posterior root of zygoma (see Muizon, 1999; character 26 for additional reference). (3) Lateral margin of paracanine fossa formed by (Rougier et al., 1998; character 81): anterolateral process of maxilla (0); premaxilla (1). (4) Precanine notch (Muizon, 1999; character 33): absent (0); present (1). (5) Palatal portion of premaxilla lateral to incisive foramen (modified from Muizon, 1999; character 32): extends to level of anterior border of canine (0); extends posterior to anterior border of canine (1). (6) Suture between nasal and frontal (Muizon, 1999; character 28): acutely ‘W’-shaped (0); very open ‘W’ or almost convex posteriorly (1).

FORASIEPI ET AL.—NEW HATHLIACYNID FROM BOLIVIA (7) Location of infraorbital foramen: dorsal to anterior border of P3 (0); dorsal to posterior border of P3 (1); dorsal to M1 (2). Although the condition for Mayulestes is uncertain (because of damage, it could be state 0 or 1) the presence of state 0 in other possible outgroups (such as Deltatheridium or Pucadelphys; Rougier et al, 1998; Marshall and Muizon, 1995) lead us to consider this character as ordered. (8) Palate and basicranium (modified from Rougier et al., 1998; character 96): palate lower than basicranium (0); both at same level (1). (9) Palatal extension (modified from Rougier et al., 1998; character 94): extends behind last molar level (0); extends to last molar level (1). The character was considered polymorphic for Prothylacynus since the specimen figured by Sinclair (1906:fig. XLVIII) has state 0 whereas MACN A 5931–5937 has state 1. This difference may be correlated with age; until that is confirmed we prefer to code it as polymorphic for Prothylacynus. (10) Major palatine foramen: present (0); absent (1). (11) Posterior end of palatal process at palatines (Muizon, 1999; character 23): straight or slightly arched posteriorly (0); conspicuously double-arched (1). (12) Postpalatine torus (modified from Muizon, 1999; character 31): present (0); absent (1). (13) Hamular process of pterygoid (Muizon, 1999; character 22): present and long (0); reduced to absent (1). Muizon (1999) stated that the hamular process in sparassodonts (referred to as borhyaenoids in his work) is absent or very reduced. In therians, the tensor veli palatini muscle inserts by a tendon on the tip of the hamulus and/or on the soft palate proper (Barghusen, 1986). In all cases however, the tendon for the insertion, part or all of the musculature uses the hamular process as a trochlea in order to change its direction to enter the soft palate from its lateral side (Barghusen, 1986). Thus, the presence of a hamular process seems to be universal among therians, and for extension it would be present in sparassodonts. In the specimens of Sparassodonta examined by us, the ventral margin of the pterygoids is broken. As we could not evaluate the morphology of this process among sparassodonts, we provisionally scored this character as did Muizon (1999), but note that it needs to be re-defined in future studies. (14) Fronto-squamosal contact (Muizon, 1999; character 13): absent (0); present (1). (15) Pneumatization of the squamosal (Muizon, 1999; character 25): absent (0); present (1). Scorings follows Muizon (1999:table 2). (16) Postglenoid process (modified from Muizon, 1999; character 18): narrow and high, width/length ratio < 3.5 (0); wide and low, width/length ratio > 3.5 (1). (17) Alisphenoid glenoid process (entoglenoid process of Muizon, 1999; character 5): present (0); absent (1). (18) Transverse width of glenoid fossa (Muizon, 1999; character 19): less than twice anteroposterior length (0); more than twice anteroposterior length (1). (19) Tympanic process of alisphenoid (Muizon, 1999; character 11): absent (0); present (1). (20) Post-tympanic process of squamosal and paracondylar process of exoccipital (paraoccipital process of Muizon, 1999; character 9): small (0); large and of same length (1). (21) Orientation of post-tympanic and paracondylar processes (paraoccipital process of Muizon, 1999; character 10): ventrally projected (0); anteroventrally projected (1). (22) Ectotympanic (modified from Muizon, 1999; character 4): not enclosed by squamosal (i.e., in life should be connected to skull only by ligaments) (0); enclosed by squamosal (1). (23) External acoustic meatus (Muizon, 1999; character 16): mediolateral width shorter than anteroposterior length (0); mediolateral width longer than anteroposterior length (1).

683

(24) Transverse canal (Muizon, 1999; character 8): absent (0); present (1). (25) Foramen ovale position (Muizon, 1999; character 7): between petrosal and alisphenoid (0); in alisphenoid (1). (26) Carotid foramen located near basisphenoid-basioccipital suture: present (0); absent (1). (27) Hypoglossal foramina: two (0); one (1). (28) Sulcus between hypoglossal foramina and foramen for inferior petrosal sinus: absent or shallow (0); deep, with prominent lateral edges (1). (29) Orientation of petrosal (Muizon, 1999; character 20): slightly oblique to sagittal plane (0); subvertical to vertical (1). (30) Mastoid exposure of petrosal (Muizon, 1999; character 6): not reduced and external; i.e., contributes to occipital shield (0); reduced and internal; i.e., excluded from occipital shield (1). Muizon (1999) considered an additional feature, the contribution of the squamosal to the occiput: absent or small versus large (Muizon, 1999; character 21). We noted that the contribution of the squamosal to the occiput is related to that of the petrosal. When the petrosal is exposed on the occipital area, the squamosal is absent (see Mayulestes in Muizon, 1998) or extremely reduced (see Didelphis as an example). When the petrosal does not contribute to the occiput, the squamosal occupies its place; consequently, it is well developed in occipital view (e.g., sparassodonts in general). (31) Crest between epitympanic recess and alisphenoid hypotympanic sinus (petrosal crest of Muizon, 1999, character 15): present (0); absent (1). (32) Subarcuate fossa (Muizon, 1999; character 12): deep (0); shallow (1). (33) Dentary dorsoventral depth: low, depth/length ratio of dentary 0.15 or smaller (0); intermediate, 0.16 to 0.19 (1); high, 0.20 or larger (2). Depth of dentary was taken at m3–m4 level, in labial view. (34) Mandibular symphysis (modified from Muizon, 1999; character 35): dentaries weakly attached at symphysis (0); dentaries strongly attached or fused at symphysis (1). (35) Incisor row (Muizon, 1999; character 36): parabolic (0); transverse (1). (36) Incisor number (Muizon, 1999; character 37): 5/4 (0); 4/3 (1); 3/3 (2). (37) P/p1 position (modified from Muizon, 1999; character 24): P/p1 nearly parallel to dentary row (0); oblique or transversal to dentary row (1). (38) Shape of P3: laterally compressed (0); bulbous and ovate (1). (39) Size of stylar cusp B on M2: large (0); reduced and flat (i.e., forming ectocingulum labial to paracone) (1); vestigial or absent (2). (40) Paraconule and metaconule on molars: present (0); absent (1). (41) Protocone on M3: present (0); extremely reduced to absent (1). (42) Metacone on M4: present and distinct from metastylar corner of tooth (0); extremely reduced, forming small prominence posterior to paracone (1); absent (2). (43) Height of lower premolars: height increases gradually (0); gaps in size between p1 and p2–p3 (i.e., even though p3 is taller than p2, they are more similar in height than p2 and p1) (1). (44) Metaconid on lower molars: present (0), absent (1). (45) Number of cusps on m1–m3 talonid: three cusps (0); two cusps (1); one cusp (2). (46) Atlas ventral arch (intercentrum): not fused to rest of vertebra (0); fused to rest of vertebra (1). In the only known atlas of Lycopsis (UCMP 38061), the intercentrum is not fused to the rest of the vertebra; nevertheless, as the specimen is a juvenile we codified uncertain for this taxon. (47) Atlantal foramen: absent (0); present (1).

684

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 26, NO. 3, 2006

APPENDIX 3. Data matrix of 47 osteological character states distributed among 10 metatherian taxa. Special Codes: ?, missing information or uncertain character-states; a, 0/1; b, 1/2 (polymorphic conditions). Taxa Mayulestes Sallacyon Notogale Cladosictis Sipalocyon Lycopsis Prothylacynus Borhyaena Arctodictis Acyon

Characters 000000a000 ?0???1??0? ?1???????? 111111a101 011011a101 111???0101 00101111a1 0011111111 0011112101 1111111101

0000000000 11?01010?? ???01???11 1110101011 1110101011 111??01001 1111001101 1111011101 11?1?11101 11???01011

0000000?00 ?1001???11 110111??11 110a111111 1101111111 1001111011 0001111011 0010000011 00100000?1 11001?11??

00?0000000 00??????10 00??????1? 00a0110021 0000010010 11001100b0 1111111001 1111121111 1121121111 ?000110020

00?0000 0??10?? 0??10?? 0201011 0201001 01010?1 0111b11 1201200 1211200 0101111