Primates, Adapiformes) Skull from the Uintan (Middle Eocene) of San Diego County, California

Home , Adapidae, Adapiformes, Anchomomys, Cantius, Copelemur, Donrussellia

AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 98:447-470 (1995

New Notharctine (Primates, Adapiformes) Skull From the Uintan (Middle Eocene) of San Diego County, California

GREGG F. GUNNELL Museum of Paleontology, University of Michigan, Ann Arbor, Michigan 481 09-1079 KEY WORDS Californian primates, Cranial morphology, Haplorhine-strepsirhine dichotomy

ABSTRACT A new genus and species of notharctine primate, Hespero- lemur actius, is described from Uintan (middle Eocene) aged rocks of San Diego County, California. Hesperolemur differs from all previously described adapiforms in having the anterior third of the ectotympanic anulus fused to the internal lateral wall of the auditory bulla. In this feature Hesperolemur superficially resembles extant cheirogaleids. Hesperolemur also differs from previously known adapiforms in lacking bony canals that transmit the internal carotid artery through the tympanic cavity. Hesperolemur, like the later occurring North American cercamoniine Mahgarita steuensi, appears to have lacked a stapedial artery. Evidence from newly discovered skulls ofNotharctus and Smilodectes, along with Hesperolemur, Mahgarita, and Adapis, indicates that the tympanic arterial circulatory pattern of these adapiforms is charac- terized by stapedial arteries that are smaller than promontory arteries, a feature shared with extant tarsiers and anthropoids and one of the characteristics often used to support the existence of a haplorhine-strepsirhine dichotomy among extant primates. The existence of such a dichotomy among Eocene primates is not supported by any compelling evidence. Hesperolemur is the latest occurring notharctine primate known from North America and is the only notharctine represented among a relatively diverse primate fauna from southern California. The coastal lowlands of southern California presumably served as a refuge area for primates during the middle and later Eocene as climates deteriorated in the continental interior. Hesperolemur probably was an immigrant taxon that entered California from either the northern (Wyo- mingmtah) or southern (New Mexico) western interior during the middle Eocene o 1995 Wiley-Liss, Inc.

Terrestrial fossil mammals from the more recent reviews of Eocene terrestrial Eocene Sespe Formation of southern Califor- mammals from southern California. nia were first described in a series of papers The San Diego Natural History Museum by Stock (1932,1933a-d, 1934a-d, 1935a-d, has been conducting fieldwork in the greater 1936a-c) and Wilson (1935). Shortly after- San Diego area for the past several years ward, mammals were described from Eocene (Walsh, 1991). During the course of this rocks in the greater San Diego area (Stock 1937, 1938; Wilson, 194Oa-c). Golz (1976), Golz and Lillegraven (1977), Lillegraven Received September 26, 1994; accepted June 21, 1995. (1976), Kelly (1990), Mason (1990), Kelly Address reprint requests to Gregg F. Gunnell, Museum ofpale- et al. (19911, and Walsh (1991) have provided ontology, University of Michigan, Ann Arbor, MI 48109-1079.

0 1995 WILEY-LISS, INC 448 G.F. GUNNELL fieldwork, a primate skull was discovered tube, while those that are not enclosed in from rocks of early Uintan age. The purpose bone are referred to as grooves. of this paper is to describe the new primate, to compare it with other notharctid and ada- Comparative samples pid primates from North America and Eu- Comparisons of the new California pri- rope, and to provide a summary of primate mate were made with a wide range of adapi- faunas from San Diego and Ventura counties. forms, either with original specimens or high Institutional abbreviations and designa- quality casts. Included in these comparisons tions used in the text are as follows: AMNH, were the following: European adapiforms: American Museum of Natural History, New Adapis parisiensis (UM 63301, 63302, Cam- York, Ny; BMNH, Natural History Museum, bridge M.538, Montauban-4); Agerinia ro- London, United Kingdom; CM, Carnegie selli (Unnumbered holotype and Cecilie Museum of Natural History, Pittsburgh, PA; 4241); Anchomomys gaillardi (UL L-46bis); Halle, Geiseltal Museum, Halle, Germany; Anchomomys stehlini (Basel En-1); Caeno- Louis, Private collection of P. Louis, Cormicy, pithecus lernuroides (NHB Eh 597-728); France; MNHN, Museum National d’His- Cantius eppsi (BMNH 13773, 15145, 15147, toire Naturelle, Paris, France; MPM, Mil- 29639); Cantius sauagei (Louis Collection waukee Public Museum, Milwaukee, WI; Mu 155-158, 160, MNHN Av 4846, 5907, NHB, Basel, Naturhistorisches Museum, 7702, Gr 98); Cercamonius brachyrhynchus Basel, Switzerland; PLV, Laboratorium voor (Basel Qv 619);Donrussellia gallica (MNHN Actuopaleontologie, Katholieke Universi- Av many unnumbered teeth); Donrussellia teit, Louvain, Belgium; SDSNH, San Diego louisi (MNHN Av 4731,4845,5664 and un- Society of Natural History, San Diego, CA; numbered teeth from Avenay and Grauves); TMM, Texas Memorial Museum, Austin, TX; Europolemur klatti (Halle-unnumbered UALP, University of Arizona, Laboratory of dentaryfrom Geiseltal and Halle 4238,4292, Paleontology, Tucson, AZ;UCMP, University 4304, 7325, 7396); Leptadapis magnus (Ba- of California, Museum of Paleontology, sel Qv 545,920, MNHN Qu 10943);Pericono- Berkeley, CA; UL, University of Lyon, Lyon, don (= Anchomomys?)pygmaeus (Basel En France; UM, University of Michigan, Mu- 367, Halle 7418, MNHN Bchs 494);Pronycti- seum of Paleontology, Ann Arbor, MI; USGS, cebus gaudryi (MNHN unnumbered holo- United States Geological Survey, Denver, type skull, Qu 11057); Protoadapis curuicus- CO; USNM, United States National Mu- pidens (AL 5182, 5719); Protoadapis filholi seum, Washington, DC; YPM, Yale Peabody (PLV 35); Protoadapis russelli (MNHN Av Museum, New Haven, CT. 4644,5759);Protoadapis sp. (MNHN Gr 150 and unnumbered teeth from Bouxwiller); MATERIALS AND METHODS North American adapiforms: Cantius abditus (many specimens in UM collections, Measurements and nomenclature AMNH 4734); Cantius angulatus (AMNH All tooth measurements were taken with 55505, 55510, 55515); Cantius frugiuorus dial calipers under a binocular microscope (AMNH 16210, 55501, 86296, CM 37448); and recorded to the nearest tenth millimeter. Cantius mckennai (many specimens in UM Length (L) measurements record maximum collections); Cantius ralstoni (many speci- mesiodistal tooth dimensions, and width (W) mens in UM collections); Cantius tor- measurements record maximum buccolin- resi (many specimens in UM collections); gual breadth. Upper teeth are designated by Copelemur australotutus (USNM 22261, tooth position with superscripts and lower 411833); Copelemur praetutus (USNM 411- teeth by subscripts with I = incisor, 882, 411886, YPM 14698); Copelemur tutus C = canine, P = premolar, and M = molar. (AMNH 16205, 55462, UALP 11377-8); All measurements of arterial canal widths Mahgarita steuensi (TMM 41578-9);Nothar- were made using a binocular microscope ctus robinsoni (many specimens in UM col- with an optical micrometer at X 10. Arterial lections); Notharctus tenebrosus (many spec- pathways referred to as canals are those sur- imens in UM collections); Pelycodus jarrouii rounded by bone and manifest as a hollow (CM 37453, USGS 6549);Smilodectes graci- CALIFORNIAN NOTHARCTINE SKULL 449

lis (USNM 17994,21815,YPM 12904, many Etymology. Latin, hesperus, west, and le- specimens in UM collections); Smilodectes mur, ghost of the departed. mcgrewi (many specimens in UM collec- Discussion. Franzen (1987) has pro- tions). Comparisons were also made with the posed resurrecting Trouessart’s (1879) fam- plesiadapiform taxa Plesiadapis cookei (UM ily Notharctidae to include all adapiforms 87990) and Ignacius gruybullianus (UM except Adapis, Leptadapis, and Cryptad- 68006). apis, which remain in the family Adapidae. This arrangement accounts for the divergent characteristics of Adupis, Leptadapzs, and RESULTS Cryptadapis while recognizing the funda- mental similarities of other adapiforms. Systematic paleontology Most North American notharctids are placed Order Primates Linnaeus, 1758 in the subfamily Notharctinae, differen- Suborder Prosimii Illiger, 1811 tiated from European notharctids (Cerca- Infraorder Adapiformes Szalay and moniinae) mainly by the presence of proto- Delson, 1979 cone folds, metaconules, and mesostyles on Family Notharctidae Pouessart, 1879 upper molars. The lone North American ex- Subfamily Notharctinae Trouessart, ception is Muhgarita, which is placed within 1879 cercamoniines based on the presence of a cingular (“true”) hypocone and an absence of metaconules and mesostyles on upper Hesperolemur, gen. nov. molars. While Hesperolemur is generically distinct Notharctus (Lillegraven, 1980). from other North American notharctines, it ope species. Hesperolemur actius, sp. clearly shares common ancestry with the nov. taxa in this subfamily (Cantius, Pelycodus, Diagnosis. Hesperolernur differs from all Copelemur, Notharctus, and Smilodectes). other known adapiforms in lacking canals Like these taxa, Hesperolemur has upper enclosing the internal carotid arterial sys- molars with protocone folds and strong met- tem within the tympanic cavity and in hav- aconules, both characteristics that stand in ing the anterior third of the ectotympanic contrast to cercamoniines and adapids. Cra- and the anterior crus fused to the internal nially, Hesperolemur is similar to Notharctus surface of the lateral wall of the auditory in most features, differing substantially only bulla; it differs from adapids (Adupis and in the disposition of the tympanic cavity. Leptadapis) in having a smaller sagittal crest, in having less massive zygomatics, in Hesperolemur actius, sp. nov. having a premolariform P4, and in having upper molars with a protocone fold (= “pseu- Notharctus sp. near N. robustior (Lille- dohypocone”) and a well-developed metaco- graven, 1980). Holotype. SDSNH 35233, compressed nule; it differs from cercamoniines (Euro- skull with left and right P4-M3, collected by pean notharctids and North American R.A. Cerutti, December 7, 1987, from Mahgurita) in having upper molars with a SDSNH locality 3413, Azuaga 11, site 5. protocone fold and well-developed metaco- Referred specimens. SDSNH 42415, nule and lower third molars with basally right MJ, from SDSNH locality 3380, Carmel inflated, very robust protoconids and meta- Mountain Ranch, Unit 16, site 3; UCMP conids and a flexed cristid obliqua; differs 113256, broken left M2?from UCMP locality from North American notharctines Cantius, V-72157. Notharctus, Copelemur, and Pelycodus in Horizon. The type and referred speci- lacking lower molar paraconids; it differs mens were collected from early Uintan (mid- from Smilodectes and Copelemur in having dle Eocene) strata originally mapped as Mis- closed lower molar trigonids and inflated sion Valley Formation by Kennedy and protoconid and metaconid cusps. Moore (1975) and regarded as a new litho- 450 G.F. GUNNELL stratigraphic unit by Walsh (1991). These strata are now tentatively correlated by Walsh (personal communication) with the type Friars Formation. Diagnosis. As for genus. Etymology. Latin, aktios, coastal, refer- ring to the coastal setting of the San Diego lo- calities. Description. SDSNH 35233 is a dorsoventrally compressed and somewhat plasti- cally deformed skull of a relatively large notharctine primate (Fig. 1). The anterior portion of the skull is missing in front of P4. The skull now measures 70.2 mm in length and 48.2 mm in maximum width but obvi- ously would have been both longer and wider had it not been crushed and broken. No su- tures are evident anywhere on the skull because of crushing and step-fracturing of cra- B nial elements. Lateral view (Fig. 1A). Only a few features of Hesperolemur can be discerned in lateral view due to crushing and breakage. The maxillary, from the base of the orbit to the gingival margin, is relatively narrow dorsoventrally, more so than in other North American notharctines. It is much narrower than in Adapis. The infraorbital foramina are relatively larger than in Notharctus, Smilodectes, Cantius (UM 939381, and Ad- apis and are positioned above P4as in those C taxa. The root of the jugal originates dorsal to the posterior root of M', making it more anteriorly placed than in Notharctus, Smilo- dectes, or Adapis but about as in Cantius. Fig. 1. Skull ofHesperoZenur actius (SDSNH 35233, holotype). A Lateral view. Small black triangle to left The orbits are crushed, but, judging from indicates position of infraorbital foramen, white arrow- relatively undistorted portions of the orbital head indicates position of confluence of frontal lines, margin, they would have been about the rel- and black arrow indicates position of parietosquamosal ative size of those in North American noth- foramen. B: Dorsal view. Small black arrow to left indicates position of confluence of frontal lines; larger arrow arctines, relatively larger than those exhib- to right indicates position of parietosquamosal foramen. ited in Adapis. The lacrimal canal appears C: Ventral view. Small black triangle indicates position to have been positioned just inside the orbit of basisphenoid-basioccipital suture. Anterior to left in as in Notharctus, Smilodectes, and Adapis, all views. Scale in 1 cm increments. where it interrupts the continuity of the anterior orbital margin. This condition differs Dorsal view (Fig. 1B). The skull exhib- from Cantius (UM 93938 (Fig. 21, where the its relatively strong postorbital constriction lacrimal canal is positioned anterior to the dividing it into distinct splanchnocranium orbital margin such that the anterior orbital and neurocranium. The nasals are not pre- rim forms a continuous, unbroken border, served. The frontals are relatively broad as The morphology of the orbital fossa in in Smilodectes (Gazin, 1958) andNotharctus Hesperolemur is completely obscured by bro- (Gregory, 1920) but unlike Adapis, in which ken and overlapping bone fragments. the frontals are constricted and concave CALIFORNIAN NOTHARCTINE SKULL 45 1 broader than it is in Adapisparisiensis (11.3 mm). The braincase is not inflated anteriorly as it is in Smilodectes but tapers more gradu- ally to the postorbital constriction. The anterior, posterior, and ventral extent of the parietals is obscured by crushing. Hesperolemur appears to have distinct parietosquamosal foramina as in Smilodectes, Notharctus, and Adapis, but they are positioned more anteriorly than in those taxa. If interpreted correctly, these foramina are located more dorsally than in Smilodectes or Notharctus, which suggests that the dorsolateral portion of the squamosal was relatively more extensive than in either of these taxa, more like the condition exhibited in Adapis. The root of the zygomatic arch in Hesperolemur is relatively broad anteropost- Fig. 2. Rostrum of Cantius abditus (UM 93938) in eriorly as in Notharctus and Smilodectes. Al- lateral view, anterior to left. Small black arrow above though broken, it does not appear to have indicates position of the lacrimal canal; larger arrow extended posteriorly to form de'ep temporal below indicates position of infraorbital foramen. Scale gutters as in Adapis (Stehlin, 1912). in 1 cm increments. Ventral view (Fig. 1C). The skull of Hesperolemur preserves the left and right maxillary tooth rows (P4to M3), small por- (Stehlin, 1912; Gingerich, 1981a; Gingerich tions of the basisphenoid, crushed and plas- and Martin, 1981). There is no evidence to tically deformed petrosals, and a somewhat suggest that the frontals were inflated as in distorted basioccipital. The anterior portions Smilodectes (Gazin, 1958), but instead they of the basisphenoid are missing. Posteriorly, were probably flattened as in Notharctus. the basisphenoid contacts the basioccipital The frontals taper laterally to form rela- at about the anteriormost point of the audi- tively broad superior rami of the postorbital tory bullae as in Notharctus and Smilode- bars, very similar to the condition exhibited ctes, relatively more posteriorly than in in Smilodectes. The frontal lines converge at Adapis. The basisphenoid contacts the the midline just posterior to the postorbital anteromedial portion of the bullae anterior constriction much as in Adapis and unlike to its contact with the basioccipital as in No- Smilodectes or Notharctus, where the frontal tharctus. The basisphenoid does not appear lines join farther posteriorly. to overlap the bullae as in Smilodectes. The Hesperolemur has a well-developed sagit- central ridge of the basisphenoid is broad tal crest extending from the confluence of and relatively flat as in Notharctus, not nar- the frontal lines posteriorly to the nuchal rower and more elevated as in Adapis or crest. The sagittal crest is less developed laterally compressed and sharply defined as than is typical ofAdapis (Gingerich, 1981a) in Smilodectes. The mesopterygoid fossa ap- but is better developed than it is in most pears to have been only moderately exca- specimens of Smilodectes (which almost vated laterally as in Adapis and Notharctus, never has a sagittal crest) or Notharctus not deeply pocketed as in Smilodectes (where a weak to moderate sagittal crest of- (Gazin, 1958). ten is present). The braincase is pear- The basioccipital of Hesperolemur is nar- shaped, being relatively broad and rounded row anteriorly and widens posteriorly, form- posteriorly, tapering anteriorly to the postor- ing a relatively broad surface approaching bital constriction. The anterior biparietal the condition in Notharctus but not as broad width (20.9 mm) is comparable to that of as in Adapis and not as laterally impinged Smilodectes gracilis (21.0 mm) but much upon by the bullae as in Smilodectes. The 452 G.F. GUNNELL basioccipital does not overlap the medial lemur can be discerned by examining both wall of the auditory bullae as it does in Noth- left and right ear regions (Fig. 3C). The pro- arctus and especially Smilodectes but does montorium (best seen in the right ear region contact it more broadly than in Adapis. The [Fig. 3A]) is a rounded eminence. At the dor- basioccipital has a relatively weak median sal base of the posterior portion of the pro- ridge extending from its anterior contact montorium is the cochlear fenestra, some- with the basisphenoid to the margin of the what hidden by a posterior extension of the foramen magnum much as in Notharctus but promontorium. Just lateral to the cochlear unlike Adapis, where this ridge appears to fenestra is a small fossa that may represent be absent, or Smilodectes, where a median the point of origin for the stapedius muscle. ridge is well developed. The vestibular fenestra (best seen in the left The auditory bullae are relatively large, ear region [Figs. 3B, 41) is located along the continuous with the petrosal, and rounded lateral margin of the promontorium. The anteromedially. These taper slightly pos- epitympanic recess is lateral to the vestibu- terolaterally but do not form an ossified ex- lar fenestra. The tensor tympani fossa is lo- ternal auditory meatus. The anteromedial cated just anterior to the epitympanic recess. extent of the petrosal reaches beyond the Relatively large stylomastoid foramina contact of the basisphenoid and basioccipital are preserved on both left and right bullae as in Notharctus, but this is not as anteriorly located posterolaterally. It is possible to fol- extended as in Adapis. Anteriorly, the bullae low the course of the facial canal through the overlap the posterior margins of the basis- stylomastoid foramen and into the tympanic phenoids, unlike Smilodectes, where the ba- cavity. However, the course of the facial canal sisphenoids overlap the bullae. cannot be traced farther due to breakage. It ompanic cavity. Morphological fea- presumably crossed the tympanic cavity at tures of the tympanic cavity can be discerned the medial margin of the epitympanic recess from both the left and right side of the skull as in other notharctines, but this is impossi- of Hesperolemur. The right tympanic cavity ble to determine with certainty. has been crushed anteriorly, tilting all of the The circulatory pattern of the internal ca- preserved features anterodorsally to pos- rotid arterial system in Hesperolemur ap- teroventrally (Fig. 3A). The left tympanic pears quite different from that documented cavity has been crushed dorsoventrally and for all previously known notharctines (Greg- anteriorly, pushing most of the middle ear ory, 1920; Szalay, 1975; Szalay and Delson, dorsally (Figs. 3B, 4). Even with such distor- 1979). Like other notharctines, the internal tion, many of the relevant features of middle carotid artery presumably entered the bulla ear morphology are preserved on one side or posteriorly, ventral and medial to the stylo- the other. mastoid foramen (neither ear region pre- The tympanic cavity of Hesperolemur is serves the posterior carotid foramen). Unlike separable into four distinct regions. There other known adapiforms (and other Eocene is a relatively large anteromedial cavity (Si- primates), there is no evidence of any bony mons and Rasmussen, 1989; Rasmussen, canals enclosing arteries within the tym- 1990) roofed by the epitympanic wing of the panic cavity (Figs. 3A,B, 4). It is difficult petrosal. This cavity is divided into anterior to determine the exact path of the internal and posterior areas by a medial secondary carotid artery until it reaches the promont- septum (MacPhee, 1981). A relatively large orium. There is a large, deep groove (not a anterolateral cavity is present, roofed by the canal) along the lateral aspect of the pro- “tegmen tympani.” Posterolaterally, the montorium that represents the promontorial tympanic cavity is mediodorsally limited by pathway of the promontory artery, but no the facial canal and laterally by a deep epi- tube or canal is present that is continuous tympanic recess. The posteromedial portion with this groove either anteriorly or posteri- of the tympanic cavity is occupied by the orly. Some specimens of Smilodectes and No- promontorium and associated structures. tharctus have canals that are open across Although crushed and broken, many of the the lateral part of the promontorium (mani- features of the tympanic roof of Hespero- fest as a groove), but well-developed, closed Fig. 3. Stereophotographs of tympanic regions of top and lateral to right showing stylomastoid foramen Hesperolemur actius (SDSNH 35233, holotype). A Ven- (white arrowhead at bottom right), groove across lateral tral view of right tympanic cavity with anterior to top surface of the promontorium for the promontory artery and lateral to left showing promontorium (p), stylomas- (white arrowhead to left), anterior portion of the ecto- toid foramen (white arrow), groove across lateral surface tympanic anulus (white arrow at top), and the position of promontorium for promontory artery (white arrow- of the auditory meatus (large white arrow to right). C: head), and medial secondary septum (black arrowhead). Photograph of entire basicranial region of H. actius, an- B: Ventral view of left tympanic cavity with anterior to terior to top. Scale = 1 cm. 454 G.F. GUNNELL

Fig. 4. Stereophotographs of left ear region of anulus. CF, cochlear fenestra: EC, ectotympanic; PC, Hesperolemur actius (SDSNH 35233, holotype) and line groove for promontory branch of internal carotid artery; drawing indicating relevant features, anterior to top and PGF, postglenoid foramen; SmF, stylomastoid foramen; lateral to right. On photographs, white arrowhead at VF, vestibular fenestra. Photographic scale = 1cm. Sty- bottom indicates the stylomastoid foramen, white arrow- lomastoid foramen appears deceptively large in drawing head to left indicates cochlear fenestra, and black arrow- due to breakage and oblique position. head at top indicates anterior portion of ectotympanic

canals are present throughout the rest of areas where such canals would have con- the internal carotid circulatory pathway nected are smooth and show no signs of any through the tympanic cavity. It is possible bony portions having been broken off. In ad- that these canals are simply missing in dition, there were no broken bone fragments Hesperolemur because of breakage, but the within the tympanic cavities that could be CALIFORNIAN NOTHARCTINE SKULL 455

TABLE 1. Promontory and stapedial canal diameters in notharetines Specimen Genus/species Promontory (narrowest) Stapedial (broadest) Aspect UM 98828 Notharetus tenebrosus 1.0 mm 0.50 mm External UM 98828 Notharetus tenebrosus - 0.40 mm Internal YPM 11466 Notharetus tenebrosus‘ 0.60 mm 0.40 mm Internal UM 100000 Smilodectes gracilis 1.0 mm 0.50 mm Internal UM 100044 Smilodeetes gracilis 1.3 mm 0.80 mm External UM 100521 Smilodeetes gracilis’ 1.7 mm 0.90 mm Ext e r n a 1 UM 101212 Smilodeetes gracilis 1.3 mm 0.80 mm External UM 100603 Smilodectes megrewi 1.2 mm 0.80 mm External SDSNH 35233 Hesperolemur actius 0.9 mm - Internal ‘From Gingerich (1973). ‘Measured at bifurcation of internal carotid artery. attributed to these canals, which seems un- skulls of Smilodectes (MPM 2612 and 5409) likely given that other delicate structures both have promontory canals much (their (ear ossicles, portions of the ectotympanic italics) larger than stapedial canals. Field- anulus) were found within the tympanic work in the Bridger Formation of southwest- cavities. ern Wyoming has produced six additional Hesperolemur also differs from known no- notharctine skulls (UM numbers 98828 tharctids in apparently lacking a canal or (Notharctus tenebrosus), 100000, 100044, groove for the stapedial artery. It is difficult 100521, 101212 (Smilodectes gracilis), and to be certain of this because of the condition 100603 (Smilodectes mcgrewi) (Figs. 5, 6) of the ear regions, but there is no evidence that provide additional support for the mor- apparent for the existence of a stapedial ar- phology cited by Gingerich (1973) and tery (Figs. 3A,B, 4). The lack of a stapedial MacPhee and Cartmill (1986). In all cases, artery in Hesperolemur seems incongruous by either external or internal diameters, the with previous notions of the disposition of promontory canal is larger than the stape- arterial circulation through the middle ear dial canal in these notharctine specimens. in “extinct strepsirhines” (Szalay, 1975; Sza- While this does not necessarily prove that lay and Delson, 1979). It has been suggested the arteries in these canals reflected the that the strepsirhine condition is one in same relationship, it does call into question which the stapedial artery is larger than the the usefulness of such a character for recog- promontory artery, while the opposite holds nition of a haplorhine-strepsirhine dichot- in haplorhines (Gregory, 1920; Szalay, 1975; omy among middle Eocene primates. If rela- Szalay and Delson, 1979; but also see MacPhee and Cartmill, 1986; Ross, 1994). tive size of promontory and stapedial There has been some evidence to suggest arteries is a valid character differentiating that the supposed strepsirhine condition is strepsirhines from haplorhines (this is in not representative of notharctines (Gin- dispute, see Beard and MacPhee, 19941,then gerich, 1973; MacPhee and Cartmill, 1986), adapiforms are haplorhines, a conclusion Adapis (Gingerich and Martin, 1981), or few proponents of Strepsirhini-Haplorhini Mahgarita (Rasmussen, 1990), but this evi- would support (although see Cartmill and dence has been disputed (Wilson and Szalay, Kay, 1978). Other lines of evidence do not 1976; Szalay and Delson, 1979). support the existence of a strepsirhine- Gingerich (1973) provided measurements haplorhine dichotomy among Eocene pri- of diameters of bony canals in one specimen mates (Simpson, 1940; Gingerich, 1981b; of Notharctus (YPM 11466) that indicated Rasmussen, 1986, 1990, 1994; Beard, 1988; that the promontory artery was, in fact, Cartmill, 1994). The reality is that such a larger than the stapedial artery (see Table dichotomy simply is not usefully applied to 1). Gingerich and Martin (1981) provided the early primate radiation. In any event, measurements of the Cambridge skull ofAd- the relatively large size of the promontory apis where the same condition holds. arterial groove and the apparent lack of a MacPhee and Cartmill (1986) note that two stapedial artery in Hesperolemur is not out 456 G.F. GUNNELL

Fig. 5. Stereophotographs of right middle ear region of stapedial canal as it crosses the facial canal (below); of Notharctus tenebrosus (UM 98828) and line drawing black arrow at top indicates internal carotid canal. FC, indicating relevant features, ventral to top and anterior facial canal; ICC, internal carotid canal; P, promontor- to left. On photographs, white arrowhead to left indi- ium; PC, promontory canal (with bone removed laterally cates the promontory canal on the lateral aspect of the to show caliber of opening); SC, stapedial canal. Photo- promontorium with canal partially opened to show cali- graphic scale = 4 mm. ber; smaller white arrowhead to right indicates opening of step with findings based on other along the anterodorsal margin of the audi- known adapiforms. tory meatus. It is attached to the bullar wall Broken pieces of the left ectotympanic by a solid, bony anular bridge (MacPhee and anulus, incus, and malleus and a nearly Cartmill, 1986; MacPhee, 1987). There is a complete right incus were found within the small suture or crack running along the lat- tympanic cavities of Hesperolemur. The ecto- eral margin of the ectotympanic suggesting tympanic fragment consists of approxi- the presence of a recessus dehiscence mately the anterior third of the anulus (Figs. (MacPhee, 1987) and indicating that the an- 3B, 4). This ectotympanic portion appears to ular bridge is petrosal in origin. A grooved have been solidly fused to the internal bullar surface that served as the attachment area wall just medial to the postglenoid foramen for the tympanic membrane is formed by a CALIFORNIAN NOTHARCTINE SKULL 457

Fig. 6. Stereophotographs of left ear region ofSmilo- orium; PC, promontory canal (with bone removed anteri- decks gracilis (UM 100000) and line drawing indicating orly to show caliber of opening); PGF, postglenoid fora- relevant features, anterior to right and ventral to top. men; SC, stapedial canal (with bone removed laterally Black arrow on photograph indicates position and open- to show caliber of opening); SmF, stylomastoid foramen. ing of internal carotid artery. FC, facial canal; FM, fora- Photographic scale = 1 cm. men magnum; ICC, internal carotid artery; P, promont- strong ridge (crista tympani) located along mill, 19861, the anterior portion of the ecto- the medial margin of the ectotympanic tympanic of Hesperolemur is fused to the fragment. lateral bullar wall throughout its course. Unlike Smilodectes (MacPhee and Cart- UM 100521 and 101212 (Fig. 7) show the 458 G.F. GUNNELL

Fig. 7. Stereophotographs of right ear region of Smi- fused to anular bridge, larger white arrowhead indicates lodectes gracilis showing disposition of ectotympanic. A: broken “free”segment of ectotympanic, and black arrow UM 100521 in lateral view with anterior to right and indicates anterior crus located along dorsal surface of dorsal to top. Small white arrowheads indicate posterior the auditory meatus. C: UM 101212 in ventral view with crus (left) and broken “free” segment (right) of ectotym- anterior to top and lateral to left. Black arrow indicates panic; white arrow at lower left indicates position of the position of internal carotid foramen; white arrowhead internal carotid foramen. B: UM 100521 in ventrolateral indicates broken “free” portion of ectotympanic lying view with anterior to top and lateral to left. Small white anteromedial to promontorium. Scale = 1 cm. arrowhead indicates posterior crus of ectotympanic disposition of the ectotympanic in Smilo- it is no longer attached to the bulla, exhib- dectes gracilis. The posterior crus of Smilo- iting the “free” ring condition of extant Le- dectes is connected to the lateral bullar wall mur (anular bridge incomplete sensu via an anular bridge. After the ectotympanic MacPhee and Cartmill, 1986). The ectotym- passes the level ofthe internal carotid canal, panic anulus may contact a small portion CALIFORNIAN NOTHARCTINE SKULL 459 of the lateral bullar wall along its anterior the left incus is missing both the long and curvature as in Lemur but is not fused to short crura. Comparisons with a left incus the lateral wall anteriorly as in Hespero- of Smilodectesgracilis (UM 100000) indicate lemur. The anterior crus of Smilodectes rests that Hesperolemur and Smilodectes shared in the auditory meatal fossa (the crus is most features of incudal morphology (Fig. clearly visible in UM 100521 [Fig. 7B1). The 8A,B). The body of the incus in both taxa anterior crus of the ectotympanic ofHespero- has distinct medial and lateral mallear fac- lemur does not appear to lie in a meatal fossa ets that are oriented at nearly 90 degrees to but is instead attached to the internal sur- one another. The medial mallear facet is face of the lateral bullar wall at the auditory more concave than the lateral facet, espe- meatus similar to extant Otolemur. cially in Smilodectes, although the lateral The presence or absence of a continuous facet is somewhat saddle-shaped in both bony connection between the ectotympanic taxa. The short crus is relatively shorter and and the lateral bullar wall (a complete anu- less robust in Hesperolemur, and, although lar bridge [see MacPhee and Cartmill, 1986; the long crus is missing, judging by the angle MacPhee, 1987; Beard and MacPhee, 19941) of the broken surface, it must have been cannot be determined. If Hesperolemur was more inferiorly oriented than in Smilodectes, like other known notharctids it would have resulting in the short and long crura being lacked a complete anular bridge (MacPhee relatively farther apart in Hesperolemur. and Cartmill, 1986; MacPhee, 1987; Fran- Other characteristics of the basicranium zen, 1987). However, the fact that the ante- are also evident on the skull of Hespero- rior third of the ectotympanic is solidly lemur. The mastoid region, while not rela- attached to the lateral bullar wall via an tively large, did enclose at least two substan- anular bridge suggests that the entire ecto- tial air cells. The mastoids are not as inflated tympanic may have been similarly con- as in Adapis but approach the condition ex- nected to the bulla. There is a small piece hibited in Smilodectes and Notharctus. The of bullar wall from the ventrolateral portion glenoid fossa is relatively wider mediolater- of the right bulla of Hesperolemur that has ally than in Smilodectes but is approached a smooth, shallow groove running antero- by Adapis in this characteristic. The postgle- posteriorly across its dorsal surface. This noid process is robust and angled slightly may represent a band or groove for the ecto- posteriorly as in Smilodectes. It may have tympanic and if so suggests that the ecto- contacted the bulla, unlike Smilodectes, but tympanic was in close proximity to the dorsal does not appear to have been fused to the bullar surface, but there is no evidence to bulla anterior to the external auditory me- indicate that the ectotympanic was directly atus as inAdapis. There is a distinct postgle- connected to the bulla in this area. If Hes- noid foramen located on the posteromedial perolemur had an ectotympanic anulus aspect of the postglenoid process abutting attached to the dorsal surface of the bulla against the anterior portion of the audi- throughout its course by a complete or nearly tory meatus. complete anular bridge, it would have re- Dentition. The skull of Hesperolemur sembled the condition exhibited by some ex- preserves P4-M3 on both sides (Figs. lC, 9B). tant cheirogaleids (Cartmill, 1975; MacPhee P4 is a mesiodistally compressed tooth with and Cartmill, 1986). a well-developed protocone that is positioned The malleus fragment of Hesperolemur is near the mesiolingual base of the paracone. represented only by the incudal articular The paracone is nearly twice the height of surface and provides little information about the protocone. The paracone is centrally the complete disposition of this ossicle. Com- placed along the buccal margin producing a pared to Smilodectes gracilis (UM 101212), preparacrista and postparacrista of nearly the malleus ofH. actius is somewhat smaller, equal length. There is no metacone. P4 has and the incudal articular surface is flatter a small parastyle and no metastyle. There and less saddle-shaped. is a weak cingulum that encircles the tooth. The right incus of Hesperolemur is nearly The molars are quite worn across the lin- complete, missing only the long crus, while gual half of their occlusal surfaces. M2 is 460 G.F. CUNNELL

Fig. 8. Line drawings of (A) right incus of Hesperolemur actius (SDSNH 35233, holotype) and (B) left incus of Smilodectes gracilis (UM 100000). In A, lateral view is bottom left, medial view is bottom right, and anterior view is top. In B, medial view is bottom left, lateral view is bottom right, and anterior view is top. Small arrows point dorsally in each view. Scale = 2 mm.

A- B- Fig. 9. Stereophotographs of left upper dentitions of (A) Notharctus tenebrosus (UM 32335, F3-M7) and (B) Hesperolemur actius (SDSNH 35233, holotype, P4-M3) in occlusal view. Scale = 2 mm. the largest molar, with M1 being noticeably rather wide, shallow valley formed by gently smaller than MZand M3slightly smaller than sloping postparacristae and premetacristae. MI. It is difficult to determine the position M3 has a very weak, low metacone. Rela- and relationships of upper molar protocones tively robust metaconules are present on and hypocones with precision, but extensive M1-2,while M3has a relatively smaller meta- comparisons with other adapiforms indicate conule. The presence and disposition of mo- that Hesperolemur almost certainly had a lar paraconules cannot be determined due protocone fold and lacked a cingular hypo- to obscuring wear. MI has a weakly formed cone. MI" have well-developed paracones mesostyle that is somewhat better developed and metacones that are separated by a on M2 and absent on M3. Relatively heavy CALIFORNIAN NOTHARCTINE SKULL 46 1

mesial and distal cingula are present on all the metaconid taller, more basally inflated, three molars as well as relatively weaker and much more massive. The protoconid and buccal and lingual cingula. Stylar shelves metaconid are connected buccally by a well- are essentially absent. M' is longest buccally developed crest. There is no paraconid, but and tapers somewhat lin,wally but retains a short, arcuate paracristid connects the me- a generally square outline. M2 is wider and sial flanks of the protoconid and metaconid more rectangular in outline. M3 is generally forming the mesial wall of a distinct trigonid squared but the mesiobuccal corner of the fovea. There is a distinct, robust hypoconid tooth is slightly distended, and the hypocone that is connected to the postprotocristid by lobe is reduced compared to the other a sigmoid cristid obliqua. A small ectocin- molars. gulid is present beneath the hypoflexid. A Comparing the upper dentition with other small entoconid and an elongate, faintly bilo- North American notharctines indicates that bate hypoconulid are present. Hesperolemur is generally similar to most The M3 of Hesperolemur differs from that taxa but differs in important ways from each of Notharctus and Cantius (Fig. 10D,E) in (Figs. 9, 10). Hesperolemur differs from having a smaller trigonid fovea with a Notharctus in having P4 with a more mesi- shorter, more robust paracristid and no par- ally and buccally placed protocone and a cen- aconid, in having a very massive metaconid trally placed paracone. Upper molars of and a more inflated protoconid, in having a Hesperolemur differ from those of Notharc- weaker buccal cingulid, in lacking multiple tus in having less well-developed mesostyles entoconid cuspules, and in lacking the dis- with little or no stylar shelf development and tinctive lingual extension of the postproto- in lacking distinct hypocone lobes that are cristid at its juncture with the cristid obli- separated from the protocone by a distinct qua. Hesperolemur differs from Copelemur lingual groove (especially true of more de- in lacking a distinct paraconid (C. tutus has rived Notharctus species). Hesperolemur dif- a paraconid: C. praetutus does not), in lack- fers from Smilodectes in having weaker mes- ing an entoconid notch, in lacking the lin- ostyles, in having very shallow trigon basins, gually positioned entoconid typical of Copel- and in having relatively larger hypocone emur, and in having more massive and lobes. inflated protoconid and metaconid. C. prae- Hesperolemur differs from derived Can- tutus shares a sigmoidal cristid obliqua- tius species in much the same manner as it postprotocristid flexure with Hesperolemur, does from Notharctus, but more primitive while C. tutus has this complex less well de- species of Cantius have weaker mesostyles veloped. and more mesial P4 protocones like those of The M3 of Hesperolemur differs from that Hesperolemur. Hesperolemur also has a more of Pelycodus jarrovii in lacking a mesiodis- mesiodistally compressed P4 than most spe- tally compressed trigonid, in lacking a para- cies of Cantius. Hesperolemur differs from conid, in having a sigmoid cristid obliqua- Copelemur (C. tutus only) in lacking a well- postprotocristid flexure, in having a massive developed mesostylar cusp (Hesperolemur metaconid, and in having a better developed has a crest-like mesostyle instead of an iso- trigonid fovea. Hesperolemur differs from lated cuspule), in lacking a lingually ex- Smilodectes (Fig. 1OC) in having the trigonid panded protocone lobe, in having a weaker of M, closed lingually by the paracristid, in stylar shelf, in having a shallower trigon ba- having a much more massive metaconid and sin, and in having more bulbous and robust an inflated protoconid, in having a sigmoidal metaconules and stronger upper molar cin- cristid obliqua-postprotocristid flexure, in gula. Hesperolemur differs from Pelycodus having a shallower talonid basin, and in hav- jarrouii in having upper molar mesostyles ing a relatively smaller entoconid. and in having the M3 paracone less buc- UCMP 113256, a broken left M2?,was de- cally distended. scribed and figured by Lillegraven (1980) as SDSNH 42415 is a right M3 referred to Notharctus sp. near N. robustior. Although Hesperolemur actius (Fig. 10F). The proto- the talonid is damaged, the trigonid is nearly conid and metaconid are widely spaced, with complete. The trigonid consists of a rela- Fig. 10 CALIFORNIAN NOTHARCTINE SKULL 463

TABLE 2. Dental measurements Hesperolemur actius compared with selected North American notharctines and the cercamoniine Mahgarita stevensi’ Genushpecies P‘LXW MLLxW M2LxW MsLxW M,LXW

Hesperolemur actius 4.4 X 6.0 (1) 5.2 X 6.2 (1) 5.3 x 7.3 (1) 5.0 X 6.2 (1) 6.9 X 4.5 (1) Notharctus tenebrosus 4.8 X 6.7 (12) 6.4 X 7.8 (16) 6.4 X 8.2 (14) 5.2 X 6.9 (10) 6.9 X 4.4 (11) Notharctus robinsoni 4.5 x 6.6 (3) 6.0 X 7.6 (4) 6.1 X 8.0 (3) - 6.8 X 4.2 (10) Smilodectes gracilis 3.4 X 4.2 (5) 4.4 x 5.5 (7) 4.6 X 6.0 (9) 4.0 X 5.3 (10) 5.7 X 3.5 (10) Smilodectes mcgreuli - 4.4 X 5.6 (2) 4.8 X 6.1 (3) 4.2 X 5.4 (2) 5.1 X 3.6 (2) Cantius abditus 3.9 X 5.3 (5) 4.9 X 6.4 (3) 5.1 X 7.6 (4) 4.0 X 5.9 (3) 6.6 X 4.3 (7) Copelemur tutus - - 5.5 X 8.1 (1) - - Copelemur praetutus - - - - 5.3 x 3.1 (1) Pelycodus jarrooii - 5.5 X 6.5 (4) 5.4 X 7.2 (2) 4.8 X 6.1 (3) 6.6 X 4.4 (2) Mahgarita steoensi 2.9 X 3.8 (1) 4.6 X 4.5 (1) 4.0 X 4.7 (1) 3.2 X 4.4 (1) 4.2 X 2.5 (1)

’The number in parentheses = N, measurement mean is given for samples larger than 1. tively robust protoconid widely separated poids such as Aegyptopithecus (Rasmussen, from a massive, basally inflated metaconid 1990; Simons and Rasmussen, 1989;but also and a short, relatively straight paracristid see Ross, 1994). However, as pointed out that connects the mesial flanks of the proto- above, relatively small stapedial arteries conid and metaconid. There is no paraconid also apear to be the rule in most known ada- developed. All of these features are shared piforms, so this character state could be with SDSNH 42415, and therefore this spec- viewed as primitive (symplesiomorphic) for imen is referred to H. actius. the infraorder. The apparent presence of an Measurements of the teeth of Hespero- anteriorly fused ectotympanic and the lack lemur actius are presented in Table 2. of a stapedial artery may both represent sy- napomorphies for Mahgarita and Hespero- RELATIONSHIPS OF HESPEROLEMUR lemur, but the evidence is such that defini- Hesperolemur is the latest surviving noth- tive interpretation remains difficult. arctine primate known. The only later adapi- In nearly all other dental and cranial fea- form known from North America is the cer- tures, Hesperolernur and Mahgarita appear camoniine notharctid Mahgarita stevensi quite different (Wilson and Szalay, 1976; from the Skyline Channels, Devil’s Grave- Rasmussen, 1990). Muhgarita, like all other yard Formation, Duchesnean Land-Mam- known Eocene primates, had a bony tube ma1 Age (late Eocene) of southwestern Texas that carried the promontory artery through (Wilson and Szalay, 1976;Rasmussen, 1990). the tympanic cavity. Mahgarita has true Comparison of Hesperolemur and Mahgarita (cingular) hypocones on upper molars in con- does not appear to support a close relation- trast with all other North American notharc- ship. Mahgarita and Hesperolemur do share tids and lacks upper molar mesostyles (un- an apparent lack of a stapedial artery along like Hesperolemur). Mahgarita also differs with a large promontory artery and possibly from Hesperolemur in having the following: the presence of a “fused” ectotympanic anu- more cuspate teeth that lack robust, bulbous lus (a complete anular bridge [see Rasmus- cusps; P4 with a centrally placed protocone sen, 19901). These character states are and more steeply sloping pre- and postpara- shared in common with primitive anthro- crista; lower molars with metaconid positioned posterior to the protoconid and lacking basal inflation; lower molar trigonids that are open lineally with a short, sloping Fig. 10. Stereophotographs of representative notharctine primate dentitions in occlusal view. A: Right paracristid and a very shallow, small tri- P4-M3 of Cantius ubditus (UM 88239). B: Left P3-M2 gonid fovea; and lower molars with a straight of Smilodectes grucilis (UM 100000). C: Right MH of cristid obliqua and a well-developed ento- Srnilodectes grucilis (UM 100000). D: Left M2.3 of Noth- conid positioned posterior to the hypoconid. arctus robinsoni (UM 94853). E: Right M2.3 of Cantius abditus (UM 97091).F Right M, ofHesperolemuractius Hesperolemur shares some character (SDSNH 42415). Scale = 2 mm. states with adapids. Like Adapis parisiensis 464 G.F. GUNNELL and Leptadapis magnus, the skull of H. ac- flated lower molar protoconids and metacon- tius has frontal lines that converge anteri- ids and relatively weak upper molar meso- orly and are confluent with a well-developed styles with some derived species of Cantius sagittal crest. Hesperolemur also shares a (from both North America and Europe). relatively large dorsolateral expansion of the Hesperolemur seems more distantly related squamosal with Adapis. Hesperolemur dif- to Smilodectes and Copelemur among North fers dentally from Adapis and Leptadapis in American notharctines. having upper molars with postprotocingula An analysis of the relationships among the and well-developed metaconules, a premo- seven North American adapiforms was car- lariform P4, closed lower molar trigonids, ried out using the branch-and-bound option and basally inflated protoconids and meta- of PAUP (Phylogenetic Analysis Using Par- conids that are opposite one another, not off- simony, version 3.1.1 [Swofford, 19931). This set with the metaconid posterior to the pro- analysis is based on 21 unordered cranial toconid. and dental characters (Tables 3,4)using the Hesperolemur is also similar to European primitive European cercamoniine Donrus- cercamoniines and European Cantius eppsi sellia as the outgroup to root trees. No a and Cantius savagei in some ways. Like Eu- priori or a posteriori weighting of characters ropean Cantius species, Hesperolemur has was attempted. Three most parsimonious upper molars with well-developed metaco- trees were derived from the character ma- nules and lower third molars with closed tri- trix, each consisting of 29 steps with a consis- gonids, small entoconids, flexed cristid obli- tency index of 0.724 and a retention index quae, and relatively inflated trigonid cusps. of 0.600. Figure 11 presents one of the three As in all cercamoniines except Donrussellia, hypothesized branching sequences for North Hesperolemur has lower molars that lack a American notharctines and also represents paraconid, but it differs from most cerca- the 50% Majority Rule Consensus Tree. This moniines (except Periconodon and Agerinia) cladogram differs from the strict consensus by having lower molar trigonids closed tree by resolving a polytomy consisting of all lingually instead of being broadly open as North American notharctines except Can- in the former. Caenopithecus lemuroides tius. In all trees Mahgarita and Adapis form shares upper molar mesostyles and lower a clade that is the sister group of North molar hypoconulids with Hesperolemur but American notharctines, and Cantius is the differs in most other dental details. Like Pro- sister.taxon to all other North American no- toadapis, Hesperolemur has relatively bul- tharctines. Smilodectes and Copelemur are bous cusps, but Protoadapis lacks the basal more closely related to each other than to inflation of lower molar cusps exhibited by any other notharctine in all trees. In two of Hesperolemur. Hesperolemur differs from all the three trees, Hesperolemur is the sister cercamoniines in having a protocone fold and taxon of Pelycodus, while in the third a well-developed metaconule on upper Hesperolemur is the sister to a clade con- molars. sisting of Notharctus, Smilodectes, and Co- Among North American notharctines, pelemur. Hesperolemur seems to be most similar to Pelycodus, Notharctus, and Cantius. Like Notharctus, H. actius has relatively broad SOUTHERN CALIFORNIAN PRIMATE frontals and shares similar basisphenoid FAUNAS and basioccipital morphologies. Dentally, H. Middle to late Eocene primates are well actius shares a flexed cristid obliqua on M, represented in southern California from with Notharctus, Cantius, Pelycodus, and both San Diego and Ventura counties (Stock, some species of Copelemur. Hesperolemur 1933a, 1934a; Gazin, 1958; Szalay, 1976; Lil- shares relatively heavy upper molar cingula legraven, 1980; Kelly, 1990; Mason, 1990; with Pelycodus as well as relatively robust Honey, 1990; Gunnell, 1995). Two separate and basally inflated lower molar protoconids regions in San Diego County have middle to and metaconids and a small, single-cusped late Eocene rocks that have produced pri- entoconid on M3. Hesperolemur shares in- mate specimens. In the northwestern part CALIFORNIAN NOTHARCTINE SKULL 465

TABLE 3. Characters and character states used to construct matrix shown in Table 4l Character number Character Alternative state 0 Alternative state 1 c1 Bony internal carotid canals Absent Present c2 Petrosal anular bridge Partial Complete c3 Ectotympanic "Free ring" "Partially fused c4 Lacrimal canal Outside orbit Within orbit c5 Frontals (width) Narrow Broad C6 Frontals (shape) Flat Inflated c7 Basisphenoid No bullar overlap Overlaps bulla C8 Basioccipital No bullar overlap Overlaps bulla c9 P4 protocone Widely separated' Appressed to paracone c10 Molar hypocone Cingular' Protocone fold c11 Molar metaconule Present' Absent c12 Molar mesostyle Absent' Present C13 Mesostyle form Cuspate Crescentic-arcuate C14 Molar hypocone lobe Continuous Separate from protocone C15 Molar protoconid Not basally inflatedl Basally inflated C16 Molar metaconid Not basally inflated' Basally inflated C17 Molar paraconid Present' Absent C18 Molar paracristid Not continuous' Continuous with metaconid c19 M, cristid obliqua Straight' Sigmoidal c20 P4 shape Premolariform' Molariform c21 Molar entoconid notch Absent' Present 'Character state alternative present in outgroup Donrussellia

TABLE 4. Character state matrix for the seven North American adapiform genera based on 21 cranial and dental rh.aracters (see Table 3)'

~ Taxon C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12 C13 C14 C15 C16 C17 C18 C19 C20 C21 Hesperolemur 011110001 101101 111100 Notharctus 100110010 101110 100100 Smilodectes 1001111101 0 1100010001 Cantius ???OOO??O 10O?OO 000100 Copelemur ????????O 101000 000101 Pelycodus ????????O 100?01 100100 Mahgarita 1110??0000 0 O??O 010000 Adapis 1001000000 1 O??O 010010 Donrussellia ????????OO 0 O??O000000 'The primitive European cercamoniine adapiform Donrussellia is used as the outgroup of San Diego County the Lower and Upper counties. By far the greatest number of pri- units of the Santiago Formation contain five mate taxa known from southern California local faunas ranging in age from early Uin- come from the Poway fauna of early Uintan tan to early Duchesnean (Walsh, 1991). In age in San Diego County. At least nine differ- the PowayIGreater San Diego area a series ent primates are found in these rocks. All but of at least five rock units spans the early two taxa are referable to omomyid primates, to late Uintan (Kennedy and Moore, 1971; and most are represented by genera also Peterson and Kennedy, 1974; Kennedy and known from earlier (Bridgerian)rocks in the Peterson, 1975; Walsh, 1991, personal com- interior of western North America (the ex- munication). In Ventura County, Kelly ceptions being Stockia powayensis, whose (1990) and Kelly et al. (1991) have recently ancestry can be linked to taxa in Wyoming revised the biostratigraphy of the Eocene and Utah [Honey, 1990; Gunnell, 19951, and portion of the Sespe Formation. These au- Ourayia, a taxon also known from the Uin- thors recognize five local faunas spanning tan of the western interior). the late Uintan and Duchesnean, each of One adapiform primate was described which contains primate taxa. from the early Uintan Poway faunas by Lille- Figure 12 summarizes the distribution of graven (1980). It is represented by a single primate taxa in San Diego and Ventura maxilla (UCMP 113210) containing MI-' and 466 G.F. GU “ELL

Donrussellia ostyles on upper molars (Robinson, 1968; Krishtalka, 1978). In other features noted / by Lillegraven (1980)-the broadly rounded anterior and posterior buccal corners, the lower, more rounded cusps and crests, the cingular development of the hypocone, and Mahgarita the presence of small pericones on MI-’-- UCMP 113210 is very similar to Macrotar- sius. In addition, both UCMP 113210 and

Cantius Macrotarsius share a broad, well-developed stylar shelf. The absence of molar mesostyles does not

Hesperolernur rule out inclusion ofUCMP 113210 in Macro- tarsius. M. roederi from the Brea Canyon Local Fauna, while only known from a single lower jaw, has molar morphology suggesting Pelymdus that upper molars may have lacked or had only small mesostyles. During mastication the postcristid and hypoconulid regions of Notharuus the lower molars shear along surfaces formed by the premetacrista and mesostyle of the upper molars. In Macrotarsius siegerti Srnilodectes (CM 15056, 186461, where a well-developed mesostyle exists, the postcristid and hypoconulid of the lower molars are well developed, closing off the posterior end of the talonid (Krishtalka, 1978). Kelly (1990) de- scribes M. roederi lower molars as lacking Fig. 11. Cladogram depicting one possible branching sequence for North American notharctine primates us- hypoconulids (except on M3) and instead as ing the primitive cercamoniine Donrussellia as out- having a notch in the postcristid where the group. This also represents the 50% Majority Rule Con- hypoconulid would normally be found. This sensus Tree. The analysis is based on 21 unordered suggests that M. roederi upper molars may cranial and dental characters (see Table 3) and represents one of three most parsimonious trees obtained by not have had mesostyles. If interpreted cor- PAUP 3.1.1 using the branch-and-bound option on the rectly, this suggests that UCMP 113210 may resulting data matrix (see Table 4). Tree length is 29 well represent a specimen of M. roederi or a steps; consistency index = 0.724; retention index = closely related taxon. In any event, UCMP 0.600. 113210 is better interpreted as an omomyid than an adapiform, leaving Hesperolemur as a broken M3. Lillegraven (1980) compared the sole adapiform primate known from this specimen carefully with Herniacodon southern California. and “Pelycodus” (now Cantius) and con- cluded that UCMP 113210 showed greater overall similarity to “Pelycod~s”(= Can- BIOGEOGRAPHY AND PALEOCLIMATES tius). Upon further consideration, I believe Lillegraven (1979) has hypothesized that a third possibility exists. Comparisons of mammalian migration in the middle Eocene stereophotographs of UCMP 113210 with between southern California and the west- other omomyid taxa leads to the conclusion ern interior was accomplished via a lowland that this specimen represents a species of route, the Sevier orogenic belt, that extended Macrotarsius, close to the recently described from southern California through northern M. roederi (Kelly, 1990) from the Sespe For- Arizona and southern Nevada into Utah and mation. Wyoming during the middle Eocene. Early UCMP 113210 differs from Macrotarsius Uintan mammalian faunas are similar in most notably in lacking well-developed mes- taxonomic makeup from both the western CALIFORNIAN NOTHARCTINE SKULL 467

Fig. 12. Stratigraphic distribution of middle Eocene primate taxa from San Diego and Ventura counties (Golz and Lillegraven, 1977; Lillegraven, 1980; Kelly et al., 1991; Walsh, 1991). interior and southern California. By the with both California and the western inte- later Uintan, notable faunal differences oc- rior, suggesting that some mammalian mi- cur between these areas, suggesting the clo- gration was possible between these areas, at sure of the Sevier orogenic belt leading to least in the Bridgerian. the development of more endemic faunas in The climate along the southern California southern California and the western interior coastal lowlands during the middle Eocene (Lillegraven, 1979). Emry (1990) has re- ranged from tropical in the early Uintan to cently described a middle Bridgerian fauna somewhat more arid and subtropical by the from Nevada that shares taxa in common latter part of the Uintan (Lillegraven, 1979). 468 G.F. GUNNELL

The western interior paleoclimate had reviewers provided many helpful sugges- begun to deteriorate by the early Uintan tions and comments. I thank P.D. Gingerich from a tropical, moist environment in the for access to his large collection of North Bridgerian to a drier, more open environ- American and European adapiform casts ment in the Uintan, followed by a collapse and K.C. Beard (Carnegie Museum), T.M. of paleotemperatures in the latest Eocene/ Bown (USGS, Denver), J.W. Froelich (Uni- earliest Oligocene (Berggren and Prothero, versity of New Mexico), M.C. McKenna 1992; Leopold et al., 1992; Wolfe, 1992). The (American Museum of Natural History), and warmer coastal lowlands may have served as K.D. Rose (Johns Hopkins University) for a refuge area for tropical dwelling primates providing casts of North American adapi- during the early Uintan. As paleoclimates forms. B. Miljour and P. D. Gingerich aided continued to deteriorate towards the end of in the preparation of photographs, and B. the Uintan in California, primates became Miljour prepared all line drawings. M.D. less and less important members of paleo- Uhen provided assistance in phylogenetic communities. By the latest Eocene primates analysis. This research was supported in were essentially gone from North America part by a grant from the National Geo- with only the rare Mahgarita and Rooneyia, graphic Society to the author. along with isolated occurrences of Omomys, Macrotarsius, and Ourayia (West, 1982; LITERATURE CITED Westgate, 1988, 19901, being represented Beard KC (1988)The phylogenetic significance of streps- from the southern part of Texas. The only irhinism in Paleogene primates. Int. J. Primatol. possible later occurrence of a North Ameri- 9:83-96. can primate is the enigmatic Arikareean ge- Beard KC, and MacPhee RDE (1994) Cranial anatomy of Shoshonius and the antiquity of Anthropoidea. In nus Ekgmowechashala (Macdonald, 1963, JF Fleagle and RF Kay (eds.): Anthropoid Origins. 1970; Rose and Rensberger, 1983) from New York: Plenum Press, pp. 55-97. South Dakota and Oregon (recently placed in Berggren WA, and Prothero DR (1992)Eocene-Oligocene the order ?Dermoptera by McKenna, 1990). climatic and biotic evolution: An overview. In DR Pro- Hesperolemur, as with the many omomyid thero and WA Berggren (eds.): Eocene-Oligocene Cli- matic and Biotic Evolution. Princeton: Princeton Uni- taxa known from southern California, seems versity Press, pp. 1-28. to have found refuge in this warm, coastal Cartmill M (1975) Strepsirhine basicranial structures region during the later portion of the middle and the affinities ofthe Cheirogaleidae. In WP Luckett Eocene. Hesperolemur may have migrated and FS Szalay (eds.): Phylogeny of the Primates: A multidisciplinary approach. New York: Plenum, pp. from the western interior, either from the 313-354. WyomingLJtaWColorado area or perhaps Cartmill M (1994) Anatomy, antinomies, and the prob- from New Mexico along the Sevier orogenic lem of anthropoid origins. In JF Fleagle and RF Kay belt during the late Bridgerian, reaching (eds.): Anthropoid Origins. New York: Plenum Press, California by the early Uintan. Alterna- pp. 549-566. Cartmill M, and Kay RF (1978) Cranio-dental morphol- tively, Hesperolemur may have evolved from ogy, tarsier affinities, and primate suborders. In DJ an unknown adapiform stock previously Chivers and KA Joysey (eds.): Recent Advances in present in southern California. Primatology, Vol. 3. London: Academic Press, pp, 205-213. Emry RJ (1990) Mammals of the Bridgerian (middle ACKNOWLEDGMENTS Eocene) Elderberry Canyon local fauna of eastern Ne- vada. In TM Bown and KD Rose (eds.): Dawn of the I thank T.A. Dembre for providing me the Age of Mammals in the northern part of the Rocky opportunity to study and describe Hespero- Mountain Interior, North America. Spec. Pap. Geol. lemur and S. Walsh for bringing the speci- SOC.Am. 243:187-210. Franzen JL (1987) Ein neuer Primate aus dem Mitteleo- mens of Hesperolemur to my attention, I zan der Grube Messel (Deutschland, S-Hessen). In S thank both for providing information con- Schaal (ed.): Forschungsergebnisse zu Grabungen in cerning the stratigraphy and faunas of the der Grube Messel bei Darmstadt. Cour. Forsch.-Inst. San Diego area. I thank P.D. Gingerich, K.D. Senckenberg 91t151-187. Gazin CL (1958) A review of the middle and upper Rose, and D.T. Rasmussen for discussions Eocene primates of North America. Smithson. Misc. concerning relationships and morphology Collns. 136:l-112. among Eocene adapiforms. Two anonymous Gingerich PD (1973) Anatomy of the temporal bone in CALIFORNIAN NOTHARCTINE SKULL 469

the Oligocene anthropoid Apidium and the origin of Boucot (eds.): Historical Biogeography, Plate Tecton- Anthropoidea. Folia Primatol. 19:329-337. ics, and the Changing Environment. Corvallis: Oregon Gingerich PD (1981a) Cranial morphology and adapta- State University Press, pp. 333-347. tions in Eocene Adapidae. I. Sexual dimorphism in Lillegraven JA (1980) Primates from later Eocene rocks Adapis magnus and Adapis parisiensis. Am. J. Phys. of southern California. J. Mammal. 61:181-204. Anthropol. 56:2 17-234. Macdonald JR (1963) The Miocene faunas from the Gingerich PD (1981b) Early Cenozoic Omomyidae and Wounded Knee area of western South Dakota. Bull. the evolutionary history of tarsiiform primates. J. Am. Mus. Nat. Hist. 125r139-238. Hum. Evol. 1Or345-374. Macdonald JR (1970). Review of the Miocene Wounded Gingerich PD, and Martin RD (1981) Cranial morphol- Knee faunas of southwestern South Dakota. Sci. Bull. ogy and adaptations in Eocene Adapidae. 11. The Cam- Mus. Nat. Hist. Los Angeles Co. 8:l-82. bridge skull of Adapis parisiensis. Am. J. Phys. An- MacPhee RDE (1981) Auditory regions of primates and thropol. 56~235-257. eutherian insectivores. Morphology, ontogeny, and Golz DJ (1976) Eocene Artiodactyla of southern Califor- character analysis. Contr. Primatol. 18:l-282. nia. Sci. Bull. Nat. Hist. Mus. Los Angeles Co. 26:l-85. MacPhee RDE (1987) Basicranial morphology and on- Golz DJ, and Lillegraven JA (1977) Summary of known togeny of the extinct giant lemur Megaladapis. Am. occurrences of terrestrial vertebrates from Eocene J. Phys. Anthropol. 74:333-355. strata of southern California. Contr. Geol., Univ. MacPhee RDE, and Cartmill M (1986)Basicranial struc- Wyo. 15:43-65. tures and primate systematics. In DR Swindler and Gregory WK (1920) On the structure and relations of J Erwin (eds.): Comparative Primate Biology, Vol. 1: Notharctus, an American Eocene primate. Mem. Am. Systematics, Evolution, and Anatomy. New York: Alan Mus. Nat. Hist. 3:49-243. R Liss, Inc., pp. 219-275. Gunnel1 GF (1995) Omomyid primates (Tarsiiformes) Mason MA (1990) New fossil primates from the Uintan from the Bridger Formation, middle Eocene, southern (Eocene) of southern California. PaleoBios 13:l-7. Green River Basin, Wyoming. J. Hum. Evol. 28:147- McKenna MC (1990) Plagiomenids (Mammalia: ?Der- 187. moptera) from the Oligocene of Oregon, Montana, and Honey JG (1990)New washakiin primates (Omomyidae) South Dakota, and middle Eocene of northwestern from the Eocene of Wyoming and Colorado, and com- Wyoming. In TM Bown and KD Rose (eds.): Dawn of ments on the evolution of the Washakiini. J. Vert. the Age of Mammals in the northern part of the Rocky Paleontol. 10r206-221, Mountain Interior, North America. Spec. Pap. Geol. Kelly TS (1990) Biostratigraphy of Uintan and Duches- SOC.Am. 243:211-234. nean land mammal assemblages from the Middle PetersonGL, and KennedyMP(1974) Lithostratigraphic Member of the Sespe Formation, Simi Valley, Califor- variations in the Poway Group near San Diego, Califor- nia. Contr. Sci. Nat. Hist. Mus. Los Angeles Co. nia. Trans. San Diego SOC.Nat. Hist. 17t251-258. 419:l-42. Rasmussen DT (1986) Anthropoid origins: A possible Kelly TS, Lander EB, Whistler DP, Roeder MA, and solution to the Adapidae-Omomyidae paradox. J. Reynolds RE (1991)Preliminary report on a paleonto- Hum. Evol. 15:l-12. logic investigation of the Lower and Middle Members, Rasmussen DT (1990) The phylogenetic position of Sespe Formation, Simi Valley Landfill, Ventura Mahgarita steuensi: Protoanthropoid or Lemuroid? County, California. PaleoBios 13:l-13. Int. J. Primatol. 11:439-469. Kennedy MP, and Moore GW (1971) Stratigraphic rela- Rasmussen DT (1994)The different meanings of a tarsi- tions of Upper Cretaceous and Eocene formations, San oid-anthropoid clade and a new model of anthropoid Diego coastal region, California. Bull. Am. Assoc. Pet- origins. In JF Fleagle and RF Kay (eds.): Anthropoid rol. Geol. 55:709-722. Origins. New York Plenum Press, pp. 335-360. Kennedy MP, and Peterson GL (1975) Geology of the Robinson P (1968) The Paleontology and geology of the eastern San Diego Metropolitan area, California. Bull. Badwater Creek Area, central Wyoming. Part 4. Late Calif. Div. Mines and Geol. 20O-B:45-56. Eocene primates from Badwater, Wyoming, with a dis- Krishtalka L (1978)Paleontology and geology ofthe Bad- cussion of material from Utah. Ann. Carnegie Mus. water Creek Area, central Wyoming. Part 15. Review 39:307-326. of the late Eocene primates from Wyoming and Utah, Rose KD, and Rensberger JM (1983) Upper dentition of and the Plesitarsiiformes. Ann. Carnegie Mus. Ekgmowechashala (Omomyid Primate) from the John 47:335-360. Day Formation, Oligo-Miocene of Oregon. Folia Pri- Leopold EB, Liu G, and Clay-Poole S (1992) Low-bio- matol. 41:102-111. mass vegetation in the Oligocene? In DR Prothero and Ross C (1994) The craniofacial evidence for anthropoid WA Berggren (eds.): Eocene-Oligocene Climatic and and tarsier relationships. In JF Fleagle and RF Kay Biotic Evolution. Princeton: Princeton University (eds.): Anthropoid Origins. New York Plenum Press, Press, pp. 399-420. pp. 469-547. Lillegraven JA (1976) Didelphids (Marsupialia) and Simons EL, and Rasmussen DT (1989)Cranial morphol- Uintasorex (?Primates) from later Eocene sediments ogy of Aegyptopithecus and Tarsius and the question of San Diego County, California. Trans. San Diego SOC. of the tarsier-anthropoidean clade. Am. J. Phys. An- Nat. Hist. 18:85-112. thropol. 79:l-23. Lillegraven JA (1979)A biogeographical problem involv- Simpson GG (1940) Studies on the earliest primates. ing comparisons of later Eocene terrestrial vertebrate Bull. Am. Mus. Nat. Hist. 77:185-212. faunas of western North America. In J Gray and AJ Stehlin HG (1912) Die Saugetiere des schweizerischen 470 G.F. GUNNELL

Eocaens: Adapis. Abh. Schweiz. Palaont. Ges. 38: uted by Illinois History Survey, Champagne, IL (pro- 1165-1298. gram licensed to DC Fisher). Stock C (1932) Eocene land mammals on the Pacific Szalay FS (1975) Phylogeny of primate higher taxa: The coast. Proc. Natl. Acad. Sci. U.S.A. 18:518-523. basicranial evidence. In WP Luckett and FS Szalay Stock C (1933a) Hyaenodontidae of the upper Eocene of (eds.): Phylogeny of the Primates: A multidisciplinary California. Proc. Natl. Acad. Sci. U.S.A. I9:434-440. approach. New York: Plenum, pp. 91-125. Stock C (1933b) A miacid from the Sespe upper Eocene, Szalay FS (1976) Systematics of the Omomyidae (Tarsii- California. Proc. Natl. Acad. Sci. U.S.A. 19:481-486. formes, Primates). Taxonomy, phylogeny, and adapta- Stock C (1933~)An amynodont skull from the Sespe tions. Bull. Am. Mus. Nat. Hist. 156:163449. deposits, California. Proc. Natl. Acad. Sci. U.S.A. Szalay FS, and Delson E (1979) Evolutionary History of 19:762-767. the Primates. New York: Academic Press. Stock C (1933d3 An Eocene primate from California. Trouessart E-L (1879) Catalogue Systematique, Syn- Proc. Natl. Acad. Sci. U.S.A. 19t954-959. onymique et Geographique des Mammiferes Vivants Stock C (1934a) A second Eocene primate from Califor- et Fossiles. I: Primates. (Simiae, Prosimiae, Chirop- nia. Proc. Natl. Acad. Sci. U.S.A. 20t150-154. tera); 11: Insectivora. Paris: E. Deyrolles. Stock C (1934b) Microsyopsinae and Hyopsodontidae in Walsh SL (1991) Eocene mammal faunas of San Diego the Sespe upper Eocene, California. Proc. Natl. Acad. County. In PL Abbott and JA May (eds.):Eocene Geo- Sci. U.S.A. 20t349-354. logic History San Diego Region, Pacific Section. Stock C (1934~)New Creodonta from the Sespe upper SEPM 68:161-178. Eocene, California. Proc. Natl. Acad. Sci. U.S.A. West RM (1982) Fossil mammals from the lower Buck 20:423-427. Hill Group, Eocene of Trans-Pecos Texas: Marsupicar- Stock C (1934d) A hypertragulid from the Sespe upper nivora, primates, Taeniodonta, Condylarthra, buno- Eocene, California. Proc. Natl. Acad. Sci. U.S.A. dont Artiodactyla, and Dinocerata. The Pearce-Sel- 20:625-629. lards Series 35:l-20. Stock C (1935a) New genus of rodent from the Sespe Westgate JW (1988) Biostratigraphic implications of the Eocene. Bull. Geol. SOC.Am. 46:61-68. first Eocene land-mammal fauna from the North Stock C (193513) Plesiomiacis, a new creodont from the American coastal plain. Geology 16:995-998. Sespe upper Eocene, California. Proc. Natl. Acad. Sci. Westgate JW (1990) Uintan land mammals (excluding U.S.A. 21:119-122. rodents) from an estuarine facies of the Laredo For- Stock C (1935~)Insectivora from the Sespe uppermost mation (Middle Eocene, Claiborne Group) of Webb Eocene, California. Proc. Natl. Acad. Sci. U.S.A. County, Texas. J. Paleontol. 64t454-468. 21:214-219. Wilson JA, and SzalayFS (1976) New adapid primate of Stock C (1935d) Titanothere remains from the Sespe of European affinities from Texas. Folia Primatol. California. Proc. Natl. Acad. Sci. U.S.A. 2I:456-462. 25294-312. Stock C (1936a) Sespe Eocene didelphids. Proc. Natl. Wilson RW (1935) Cricetine-like rodents from the Sespe Acad. Sci. U.S.A. 22t122-124. Eocene of California. Proc. Natl. Acad. Sci. U.S.A. Stock C (1936b) Hesperomeryr, a new artiodactyl from 21t26-32. the Sespe Eocene, California. Proc. Natl. Acad. Sci. Wilson RW 11940a) Californian paramyid rodents. Publ. U.S.A. 22:177-182. Carnegie Inst. Wash. 514t59-83. Stock C (1936~)Perissodactyla ofthe Sespe Eocene, Cali- Wilson RW (1940b) Two new rodents from California. fornia. Proc. Natl. Acad. Sci. U.S.A. 22:260-265. Publ. Carnegie Inst. Wash. 514%-95. Stock C (1937) An Eocene titanothere from San Diego Wilson RW (1940~)Pareumys remains from the later County, California, with remarks on the age of the Eocene of California. Publ. Carnegie Inst. Wash. Poway Conglomerate. Proc. Natl. Acad. Sci. U.S.A. 514:97-108. 23t48-53. Wolfe JA (1992) Climatic, floristic, and vegetational Stock C (1938) A tarsiid primate and a mixodectid from changes near the Eocene/Oligocene boundary in North the PowayEocene, California. Proc. Natl. Acad. Sci. America. In DR Prothero and WA Berggren (eds.): U.S.A. 24:288-293. Eocene-Oligocene Climatic and Biotic Evolution. Swofford D (1993) PAUP (Phylogenetic Analysis Using Princeton: Princeton University Press, pp. 421436. Parsimony), version 3.1.1. Computer program distrib-