A GENERIC REVISION OF THE PROTOLABIDINI (MAMMALIA, CAMELIDAE), WITH A DESCRIPTION OF TWO NEW PROTOLABIDINES

JAMES G. HONEY AND BERYL E. TAYLOR

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY VOLUME 161 : ARTICLE 3 NEW YORK: 1978 IN III i A GENERIC REVISION OF THE PROTOLABIDINI (MAMMALIA, CAMELIDAE) WITH A DESCRIPTION OF TWO NEW PROTOLABIDINES

JAMES G. HONEY Present Address: United States Geological Survey Denver, Colorado

BERYL E. TAYLOR Frick Associate Curator Department of Vertebrate Paleontology The American Museum of Natural History

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY VOLUME 161 : ARTICLE 3 NEW YORK: 1978 BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY Volume 161, article 3, pages 367-426, figures 1-13, tables 1-12

Issued October 24, 1978 Price. $4.25 a copy

ISSN 0003-0090

Copyright © The American Museum of Natural History 1978 C O N T E N TS

Abstract...... 371

Introduction...... 371

Acknowledgments...... 372 Abbreviations ...... 372 Stratigraphy...... 372

Systematics...... 373

Order Artiodactyla Owen, 1848 ...... 373

Suborder Tylopoda Illiger, 1811 ...... 373 Family Camelidae Gray, 1821 ...... 373 Subfamily Camelinae Gray, 1821 ...... 373

Tribe Protolabidini Zittel, 1895 ...... 373

Tanymykter, New Genus ...... 375

Tanymykter brachyodontus (Peterson, 1904) ...... 375 Protolabis Cope, 1876 ...... 382 Protolabis coartatus (Stirton, 1929) ...... 386

Michenia Frick and Taylor, 1971 ...... 406 Michenia exilis (Matthew, 1960) ...... 407

Michenia yavapaiensis, New Species ...... 411

Summary of Relationships ...... 419

Literature Cited ...... 423

369

ABSTRACT A revised diagnosis of the Protolabidini character- tolabis. The small size of the cranium and coronoid izes the tribe as having a narrow rostrum and process is correlated with the strong lateral and mes- laterally expanded anterior nares. Tanymykter, new ial flare of the mandibular angle in P. coartatus. genus, possess the above derived features and is This reflects increased emphasis on the masseter and included in the Protolabidini as the primitive sister pterygoid muscles in mastication and less emphasis taxon to Protolabis. The morphological features that on the temporalis. In contrast, the large coronoid separate Tanymykter from Protolabis are all primi- process in Procamelus indicates greater emphasis on tive except for the derived condition of closely ap- the temporalis muscle. pressed PI roots. Based upon new material from the Oxydactylus exilis Matthew is transferred to Milk Creek Formation, morphological distinctions Michenia because it shares the derived features of a between Protolabis and Procamelus are clarified. narrow rostrum, incisiform upper canine, and a shal- The principal derived features of Protolabis coar- low symphysis. A new species, Michenia yavapaien- tatus that differ from Procamelus are extreme con- sis, with Pi unerupted, P2 small, unerupted or striction of the rostrum, lateral expansion of the absent, and tall-crowned molars is described from anterior nares, P1 and P2 small and sometimes ab- the Milk Creek Formation. Finally, a cladogram of sent, and a mandibular angle with extreme lateral proposed relationships of the Protolabidini is pre- flare and a prominent mesial tuberosity. Although sented, including Tanymykter, new genus, Protolabis the above derived features are less pronounced, they and Michenia with supporting evidence for this tribe are still present in Protolabis heterodontus and other being a monophyletic group. The relationship of undescribed samples in the collection of the Ameri- other members of the Camelinae to the Protolabidini can Museum of Natural History. Thus, Stirton's spe- is also discussed. A short section detailing the strat- cies, coartatus, which has previously confused the igraphic framework of the Milk Creek Formation, generic concept of Procamelus, is transferred to Pro- Yavapai County, Arizona is included. INTRODUCTION During the Miocene, camels of the Tribe Protolabis has frequently been confused with Protolabidini were common in western North Procamelus and Miolabis, a fact pointed out by America. Members of this tribe have been Frick and Taylor (1971, pp. 4-5). In that paper found in deposits ranging in age from late the authors briefly discussed the characters that Arikareean to Clarendonian. The Protolabidini distinguish Protolabis from Procamelus and may be characterized by progressive constric- Miolabis, and clarified the taxonomy with re- tion of the rostrum, a weak to strong buccinator spect to Miolabis. Michenia is less well known fossa, marked lateral expansion of the anterior due to its recent description, although it in- nares, and a progressive shortening of the met- cludes material from the Avawatz Mountains apodials with the length of the metacarpust referred to P. coartatus, and one species (M. ranging from subequal to significantly shorter exilis) previously referred to that taxonomic than the metatarsus. scrapbasket, Oxydactylus. Judging from the The genera of the Protolabidini include Pro- abundant material of Michenia in the F:AM tolabis Cope, 1876, Michenia Frick and Tay- collection, this genus is probably well repre- lor, 1971, and Tanymykter, new genus. sented but unrecognized in other collections. Although well known since its first description, Tanymykter is based on a species also long referred to Oxydactylus, despite significant 'Although Frick and Taylor (1971, p. 21) reported that morphological differences with the genotypic the length of the metacarpus (AMNH 14259) slightly ex- ceeded that of the metatarsus in Michenia agatensis, a species, 0. longipes. reexamination of the metacarpus shows, when an allowance Our initial objective in writing the present is made for its crushed and restored condition, the length of paper was to describe the protolabidines from the metacarpus could be either equivalent or greater than the Milk Creek Formation in Arizona. that of the metatarsus. However, after an analysis of the character po-

371 372 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161 larities in Michenia and Protolabis, it became partment of Vertebrate Paleontology of the apparent that the primitive sister taxa to both of American Museum of Natural History for their these genera could be found only in taxa re- assistance and especially thank Messrs. George ferred to Oxydactylus. Recognition of certain Krochak, Otto Simonis, and Ernst Heying for derived features in species referred to Oxydac- their help. Gratitude is expressed to Mr. Ray- tylus necessitated their transfer to the Pro- mond Gooris for his skillful preparation of the tolabidini and a broadening of the concept of figures and to Mrs. Alejandra Lora and Mrs. this tribe. A final result of this study is, there- Janice Quinter for typing the manuscript. fore, a revised definition of the Protolabidini as a monophyletic group, based on shared.derived ABBREVIATIONS characters. The following abbreviations are used for in- ACKNOWLEDGMENTS stitutions throughout the text: AMNH, Department of Vertebrate Paleontology, the Recognition is due the late Childs Frick for American Museum of Natural History his interest in the Milk Creek camelids and for AMNH (M), the American Museum of Natural His- directing the F:AM field work in that area. We tory, Modern Mammals especially thank Dr. Richard H. Tedford for CIT, formerly the California Institute of Technology; permission to study the AMNH and F:AM col- the collection is now at the Los Angeles County lections. Deep appreciation is expressed to Drs. Museum Richard H. Tedford and Eugene S. Gaffney for CM, Carnegie Museum a critical reading of the manuscript and for F:AM, Frick American Mammals, Department of helpful suggestions. Special thanks are due to Vertebrate Paleontology, the American Museum Dr. Everett Lindsay for the UALP col- of Natural History making UALP, University of Arizona Laboratory of Paleon- lection available for our study and for help and tology counsel in the field. We are indebted to Mr. UCMP, University of California, Museum of Paleon- Ted Galusha for collecting the F:AM material tology, Berkeley from the Milk Creek area and making his strat- UTBEG, University of Texas, Bureau of Economic igraphic sections available for our use. It is a Geology pleasure to acknowledge the benefits gained from discussions with Dr. Malcolm C. MEASUREMENTS AND STATISTICS McKenna and Mr. Earl Manning. We are grateful to Drs. Donald E. Savage of the Uni- N, sample size OR, observed range versity of California at Berkeley, David P. P, probability from students t-distribution Whistler of the Los Angeles County Museum, S, standard deviation and Sydney Anderson of the American Mu- T, students t-statistic seum of Natural History for the loan of critical V, coefficient of variation specimens. X, mean We are grateful to all members of the De- ( ) indicate approximate measurement STRATIGRAPHY The Milk Creek Formation crops out in the Blacet (1972). Anderson and Blacet (1972) for- Walnut Grove Basin about 15 miles directly malized the usage of Milk Creek Formation, a south of Prescott, Arizona. The basin is drained name first proposed by Hook (MS). Hook's by Milk Creek, a tributary of Hassayampa Milk Creek Formation included only the Mio- Creek. The Milk Creek Formation has been Pliocene sediments of the Walnut Grove Basin; partially mapped on the Kirkland and Mount Anderson and Blacet's (1972) Milk Creek For- Union Quadrangles by Plafker (MS) and Hook mation includes sediments and intercalated (MS), in unpublished master's theses, and by basalts and latites. McKee and Anderson (1971) and Anderson and Hook (MS) gave a detailed description and 1978 HONEY AND TAYLOR: PROTOLABIDINI 373 stratigraphic section of the sediments of the marker sandstone. The correlation at the base Milk Creek Formation. According to Hook, the of the unit was established by tracing beds sediments include approximately 2300 ft. of from the Shield's Ranch Quarry, at the level of "predominantly siltstone and fine-grained sand- the Manzanita Quarry, to Goldie's Honeypot stone and interbedded layers of medium and Quarry, a distance of about one-half mile. coarsegrained sandstone and tuff [p. 9]." On Manzanita Quarry lies about 10 ft. strat- the basis of lithology and structure, Hook di- igraphically below Goldie's Honeypot Quarry, vided his Milk Creek Formation into three and thus the section thickens by about 33 ft. members. The lower member consists predomi- between Manzanita and Goldie's Honeypot. nantly of 496 ft. of interbedded conglomerate, We believe that the alternating siltstones, conglomeratic and silty sandstone, and tuff. sandstones, and occasional mudstones in these The lower member is in places conformably two sections are largely fluvial. The 37 ft. of overlain by basalt with a maximum thickness of tabular planar crossbedded sandstone atop sec- 342 ft. Hook's middle member consists of 1062 tion A, and the 17.5 ft. of slope-forming sand- ft. of argillaceous siltstones and sandstones stone below this, however, is aeolian in origin with beds of clay and coarse, resistant sand- (Honey, MS). In Section B, the 10 ft. of sand- stone and tuff. The upper member is 747 ft. stones both above and below the white limey thick and composed of "interbedded layers of layer at Deep Springs Quarry are physically siltstone, sandy and argillaceous siltstone, correlative with the aeolian sands to the west in coarse-grained, resistant sandstone, and con- Section A, and the sands immediately below glomerate [p. 12]." Locally, the sandstones of the white-limey layer exhibit large-scale tabular the upper member are strongly crossbedded. planar crossbedding. This large-scale crossbed- The vertebrate fossils described in this paper ding disappears east of Deep Springs Quarry. were obtained from Hook's upper member. In several places in Section A, isolated bones The accompanying stratigraphic columns were found weathering out of the cliffs just (fig. 1) are partial sections of Hook's upper below the gray-white resistant marker sand- member of the Milk Creek Formation, and stone, providing strong evidence for direct as- show the relative positions of the various quar- sociation of some fossils with aeolian ries. Section A was measured by the senior sediments. The fossils of Milk Creek Quarry author and Section B was drawn from informa- also occur in sandstone. The fossils in Shield's tion supplied by Ted Galusha. The sections are Ranch, Manzanita, and Goldie's Honeypot in the SW. 1/4, SW. 1/4, sect. 5, T. 11 N., R. Quarries occur in claystones and siltstones, 21 W. Mount Union Quadrangle, Arizona. Sec- however. Further, the fossils in the last quarry tion A is about 500 ft. northwest of Section B. show evidence for sorting in a current of water The upper part of the two sections are easily (Honey, MS), possibly a current associated correlated by means of a gray-white resistant with overbank deposition. S Y S T E M A T I C S ORDER ARTIODACTYLA OWEN, 1848 from the Camelopini, Camelini, and Lamini in SUBORDER TYLOPODA ILLIGER, 1811 shorter cranial (postorbital) length relative to facial length; greater rostral constriction; lateral FAMILY CAMELIDAE GRAY, 1821 expansion of anterior nares; maxillary fossa SUBFAMILY CAMELINAE GRAY, 1821 never pocketed; paroccipital process rarely ex- TRIBE PROTOLABIDINI ZITTEL, 1895 tends below level of auditory bulla in contrast with other Camelinae except Camelus; distance INCLUDED GENERA: Protolabis Cope, 1876; between upper canines equal to or less than Michenia Frick and Taylor, 1971; and Tan- distance between upper I3s; 11-2 present in prim- ymykter, new genus. itive species; PI with double roots versus single REVISED DIAGNOSIS: The Protolabidini differ root in other Camelinae; and length of metacar-

1978 HONEY AND TAYLOR: PROTOLABIDINI 375 pus either subequal or less than metatarsus ex- upon the characters of the genotypic species 0. cept in the recent Lama and Camelus.1 longipes, and a referred species, 0. brachyo- dontus. In many skull, dental, and mandibular TANYMYKTER, NEW GENUS features the two species are primitive and TYPE: Tanymykter brachyodontus (Peterson, closely resemble each other, and in addition 1904) possess primitive phalanges and unfused meta- ETYMOLOGY: From the Greek tany, long; podials. These (and other) shared primitive mykter, nose. characters were used to diagnose the genus INCLUDED SPECIES: Type species only. Oxydactylus. We feel, however, that the differ- DISTRIBUTION: Early Hemingfordian of Wy- ences in rostral, vertebral, and limb proportions oming and the medial Hemingfordian of in 0. longipes and "O." brachyodontus have Nebraska. considerable phylogenetic significance, and jus- DIAGNOSIS: Differs from Oxydactylus in tify separating the two species at the generic longer, narrower rostrum, with anterior palatine level. We agree with McKenna (1966, p. 6), vacuities usually more elongate; 11-2 generally that the morphological boundaries of Oxydac- weaker and often more laterally situated; C/ tylus have been stretched "past the breaking always larger than I3; P1 2 diastema longer; point," and we have separated 0. brachyodon- mandible with weaker mesial inflection of tus as the primitive sister taxon to Protolabis. angle; cervical vertebrae less elongate; limbs proportionally shorter with humerus longer rela- Tanymykter brachyodontus (Peterson, 1904) tive to radius, phalanges longer relative to met- Figures 3-4; Table 1-2 apodials, and length of metapodials less than Oxydactylus brachyodontus Peterson, 1904, p. 469, basal length of skull. pl. 11, pl. 12, fig. 4, pl. 13, pl. 14, figs. 3-5. Differs from Protolabis in generally weaker TYPE: CM 664, skull with 12-3, C, P1-4, M'-3; buccinator fossa and generally wider rostrum; mandible with complete dentition; cervical ver- basisphenoid with two strong, elongate tu- tebrae 1-7; distal ends of right and left humeri; berosities; P2 with stronger metastyle and para- right and left ulnae and radii; all carpals, meta- style; P3 lingual cingulum usually stronger and carpals, and phalanges, left femur; right tibia; unbroken; upper and lower molars more calcaneum; astragalus, cuboid, and metatarsals, brachyodont and anteroposteriorly shorter, with from the Upper Harrison beds2 of Peterson, stronger buccal ribs above and stronger meta- 1906 (= the Marsland Formation of Schultz, stylids and lingual ribs below; mandibular angle 1938) Nebraska: (figured by Peterson 1904, pl. ventrally unexpanded; metacarpals unfused and 11, pl. 12, fig. 4, pl. 13, pl. 14, figs. 3-5). metatarsals weakly fused; proximal phalanx DISTRIBUTION: From the Upper Harrison more convex dorsally and distal articular sur- beds of Peterson, 1906 (= Marsland Formation face more grooved and less anteriorly ex- of Schultz, 1938) Sioux County, Nebraska and panded. from deposits approximately temporally equiva- Differs from Michenia in less constricted lent to the Upper Harrison beds of Peterson in rostrum and shallower buccinator fossa; rela- Goshen and Platte Counties, Wyoming. tively longer braincase; stronger I3 and C/C; REFERRED MATERIAL: From the Upper Har- more robust symphysis; and stouter meta- rison beds, Sioux County, Nebraska CM 562, podials. "nearly complete skeleton" (Peterson 1904, p. DISCUSSION: Peterson (1904, p. 434) named 469). the genus Oxydactylus, basing his diagnosis From deposits approximately temporally equivalent to the Upper Harrison beds in Wy- 'In both the recent Lama and Camelus the metacarpus oming: F:AM 36594, 16 mi. district, S of is sometimes shorter than the metatarsus and the meta- Lusk, Goshen County, skull and mandible with podials tend to be subequal in length. Following Webb's (1965, pp. 44-45) subdivision of the Camelinae into four complete dentition; atlas, partial axis, two frag- tribes, two of which included Lama and Camelus, we mentary cervical vertebrae, proximal right assume that these two Recent taxa are derived from mem- 2The Upper Harrison beds are here presumed to be bers of this subfamily in which the length of the metacar- equal to the "Lower Marsland" as interpreted by McKenna pus always exceeds that of the metatarsus. in a footnote (1965, p. 14). 376 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161 femur, distal left femur, right fragmentary met- broken in the type skull of T. brachyodontus, acarpus; left tibia; right astragalus; two proxi- we measured the distance from the back of the mal phalanges; F:AM 36597, east side of road I3 to the external auditory meatus, and com- to Jay-EM Goshen County, skull with complete pared this length with the total lengths of vari- dentition, atlas; distal left humerus; left radius- ous limb elements. In the type of T. ulna; right metacarpus; proximal right femur; brachyodontus the 13-external auditory proximal left tibia; left calcaneum; proximal meatus/radius ratio is 88 vs. 68 in 0. longipes. right metatarsus and distal metapodial; F:AM Similarly, the 13-external auditory meatus dis- 36542, 3 mi. SW of Guernsey, Platte County, tance is 93 percent of the length of the femur partial skull with I1-M3, axis, third cervical, and 78 percent of the length of the tibia in the left tibia, left tarsals except for astragalus, left type of T. brachyodontus, vs. 68 and 60 per- metatarsus, and proximal phalanx. cent in the type of 0. longipes. Thus, the limbs DESCRIPTION AND COMPAIUSON: Peterson of 0. longipes are proportionally longer relative (1904, p. 469-471), provided a brief descrip- to the skull than are those of T. brachyodontus. tion of Oxydactylus brachyodontus and com- In the type of 0. longipes the metatarsus is pared this species with Oxydactylus longipes. associated with the skull, while the type of T. Here, we point out the significant differences brachyodontus has the metacarpus associated whereby Tanymykter brachyodontus (Peterson) with the skull. Thus, the metapodial/skull differs with Oxydactylus longipes (figs. 2-4). length ratios cannot be directly compared in the Tanymykter brachyodontus and 0. longipes types. However, the type of 0. longipes is differ mainly in their relative limb and vertebral closely matched by a skull (F:AM 50753) proportions. Tanymykter brachyodontus retains which has an associated metacarpus and meta- the primitive condition of relatively short cervi- tarsus. In F:AM 50753 the metacarpus and cal vertebrae and limbs. Peterson (1904, p. metatarsus are virtually equal in length, and the 469) noted the "comparatively short and metacarpus is approximately 18 percent longer heavy" cervicals of "O." brachyodontus as dis- than the basal length of the skull (from the tinctive from the elongate vertebrae of 0. long- anterior border of the premaxillary to the ante- ipes. Peterson also noted that the cervical rior edge of the foramen-magnum). In contrast, spines were heavier in "O." brachyodontus and the length of the metacarpus of the type of T. the spine of the axis was also more steeply brachyodontus is less than the distance from inclined than in 0. longipes. We have not seen the I3 to the external auditory meatus which these features duplicated in specimens in the makes it less than the basal length of the skull. F:AM collection. However, on elongate axes Specimens in the F:AM collection show that referred to Oxydactylus, we have noted that the the metacarpus in Oxydactylus ranges from 5 to inferior keel extends forward, as a low but 21 percent longer than the basal length of the sharp ridge, well anterior to the opening of the skull, whereas the metacarpus of Tanymykter vertebrarterial canal. On referred cervicals of ranges from 3 to 10 percent shorter than the Tanymykter the inferior keel terminates as a basal length of the skull. Moreover, specimens rounded ridge slightly posterior to the ver- in the F:AM collection show that the humerus tebrarterial canal. is longer relative to the radius and the first Peterson (1904, p. 470) stated that "the phalanx is longer relative to the metapodial in limbs of 0. brachyodontus (No. 664) are Tanymykter. In the F:AM collection metatarsal elongated and rather lighter in comparison with V is variably present in both genera. the cranium, than in 0. longipes." Although The type skulls and dentitions of T. this statement is true, we think that it is more brachyodontus and 0. longipes possess many significant that the limbs of T. brachyodontus primitive features and are morphologically sim- are proportionally shorter than those of 0. ilar. For this reason plus the fact that the skull longipes. These proportional differences are of T. brachyodontus is incomplete, Peterson seen most clearly when the lengths of the vari- could point out only a few features whereby ous limb elements are compared with those of they differed. Our analysis reveals additional the skulls of the two species. Because the oc- distinguishing features of the skull of T. ciput and anterior border of the premaxilla are brachyodontus, which are described below. 1978 HONEY AND TAYLOR: PROTOLABIDINI 377

The type skull of T. brachyodontus has a The auditory bullae of the referred T. brachyo- relatively longer, narrower rostrum than that of dontus are slightly more inflated than those of the type skull of 0. longipes. Referred speci- the type of 0. longipes. In the type of 0. mens show that the rostral width/length ration longipes the paroccipital process projects in Oxydactylus is nearly always greater than in slightly below the ventral border of the lateral Tanymykter. Similarly, the rostral length/cheek plate of the bulla. In the referred specimens of tooth length ratio is also nearly always greater T. brachyodontus the paroccipital process is in Tanymykter, again demonstrating the relative level with, or slightly above, the ventral border elongation of the rostrum in this genus. Much of the lateral bullar plate. of this elongation in Tanymykter occurs in the The basioccipital of the referred specimens premaxillaries; thus, the lengths of the anterior of T. brachyodontus is notable in possessing a palatine foramina also aveaage longer than in strong, narrow median ridge that separates the Oxydactylus. The lateral surface of the maxilla moderately to well-developed depressions for above the PI on the type of 0. longipes is the origin of the rectus capitus ventralis mus- inflated and shows little evidence of a buccina- cle. Immediately anterior to these depressions tor fossa. In contrast, the lateral surface of the are two rugose, elongate tubercles. The basioc- maxilla of the type of T. brachyodontus is cipital of 0. longipes is somewhat distorted, more depressed and shows a weak development but it apparently did not possess the strong of the buccinator fossa. Referred specimens of median ridge and well-developed rectus capitus T. brachyodontus (F:AM 36594, fig. 4) show ventralis troughs. The two anterior tuberosities, greater development of this fossa and conse- however, were apparently present. In other cra- quently have an even narrower rostrum. The nial features T. brachyodontus and 0. longipes above skull also shows marked lateral expan- are closely comparable. sion of the anterior nares, a derived feature in The incisors of both Tanymykter and Oxy- common with other protolabidines. dactylus are small. However, those of Tan- The maxillary of the type skull of T. ymykter are always weak, laterally compressed, brachyodontus is broken in the region of the and concave lingually. Many specimens re- maxillary fossa. Referred specimens (F:AM ferred to Oxydactylus have incisors which are 36534, 36597) exhibit a large, well-developed larger and subcylindrical in cross section, and maxillary fossa. This fossa is subcircular and the I3 is usually larger than the C. Referred has well-defined posterior, dorsal, and ventral specimens of Tanymykter often have the in- rims. Although the fossa is confined entirely to cisors more laterally situated than in Oxydac- the maxillary, the dorsal border crowds against tylus. the nasals. A broad, low elevation separates the There are no significant differences in the maxillary fossa from the much shallower buc- cheek teeth morphologies in the two genera. cinator fossa. A marked concavity immediately Although Peterson (1904, p. 469) stated that beneath the nasals on the type skull of T. the PI of T. brachyodontus had a single root, brachyodontus indicates that the maxillary fossa reexamination of the type specimen reveals that was similarly strongly developed in this speci- the PI roots are fused. The length of the P1-P2 men. The maxillary fossa of the type of 0. diastema (26.2 mm.) in T. brachyodontus is longipes is much shallower and lacks well-de- greater than that (14.4 mm.) in the type 0. fined posterior dorsal and ventral rims. We sus- longipes. Although the length of the diastema pect, however, that the depth of the maxillary is variable, it is almost always longer in Tan- fossa is variable in both taxa. For example, one ymykter than in Oxydactylus. Furthermore, the skull (F:AM 50733), which is referred to 0. type mandible of 0. longipes possesses a lin- longipes, has the fossa more strongly developed gual inflection on the angle which is stronger and possesses a well-defined dorsal border. than that in the type of T. brachyodontus, and Similar variation in fossa depth are seen in referred specimens of Oxydactylus also show Protolabis. stronger mesial inflections of the mandibular The sagittal and lambdoidal crests are angle. slightly weaker on the referred skulls of T. DISCUSSION: In discussing the affinities of brachyodontus than on those of 0. longipes. Oxydactylus, Peterson (1904, p. 472) con- z

V3

x

-3

._

CT C0 ._

ut ._

OO

D-,z 0->

E0

.0 1 3

378 E

-

0n

'ef

.

0

379 .4-

<0

S~~~ CZ Q

-4

a c:

LL -

m 1978 HONEY AND TAYLOR: PROTOLABIDINI 381

TABLE 1 Measurements (in Millimeters) of the Skull and Upper Dentition of Tanymykter brachyodontus (Peterson) from Nebraska and Wyoming CM F:AM F:AM F:AM 664a 36597b 36594 36542 Occipital crest to incisive border (vertex) 301.0 293.0 Foramen magnum to incisive border (basilar) 290.0 277.0 Postorbit to anterior incisive border (facial) 210.0 199.0 202.0 Postorbit to occipital crest (cranial) (92.) (89.) Anterior border foramen magnum to posterior edge M3 109.7 94.1 Palatal width between 13 19.0 21.3 19.7 Palatal width between P2 (21.) (27.9) 20.9 Palatal width between M3 posterior moiety 42.3 Palatal width narrowest part of rostrum (20.2) 19.0 18.3 13-C diastema length 12.7 16.5 13.1 12.5 C-P1 diastema length 15.8 14.0 15.0 14.4 Pl-2 diastema length 20.2 18.8 18.4 21.8 Upper canine length (7.2) 6.3 7.8 5.7 P1 length 5.8 8.1 7.1 6.0 P2 length 11.8 11.7 11.6 9.9 P2 width 5.3 5.2 6.3 4.6 P3 length 11.2 9.5 11.7 11.9 P3 width 6.9 7.5 6.8 PI length 11.5 11.1 12.0 10.9 P4 width 11.7 11.1 11.5 10.2 M1 length 14.8 13.7 15.0 13.0 M1 width 16.4 16.3 16.2 15.7 M2 length 17.8 15.6 18.8 17.1 M2 width 17.0 16.8 17.6 15.6 M3 length 19.2 18.5 19.5 19.6' M3 width 17.5 17.1 16.9 15.4 P24 length 35.0 32.7 32.6 32.1 M1-3 length 54.5 51.0 55.8 51.0 P2-M3 length 88.8 83.5 86.0 81.6 aCM 664 is the type of Tanymykter brachyodontus (Peterson). bF:AM 36597, F:AM 36594, and F:AM 36542 are referred. cluded that 0. longipes was closely allied with of Aepycamelus. However, the characters of Aepycamelus (=Alticamelus) by virtue of the "Oxydactylus" brachyodontus show a close re- small cranium and elongated cervicals and lationship to other camelids, specifically Pro- limbs. Authors since Peterson have accepted tolabis, and we have, therefore, removed this view and have classified Oxydactylus, in- "Oxydactylus" brachyodontus from the cluding 0. brachyodontus, within the Aepy- Aepycamelinae. camelinae (=Alticamelinae). The present au- Evidence irndicates that Tanymykter brachyo- thors concur with the original view of Peterson dontus is the primitive sister taxon to Pro- and believe that 0. longipes is the sister taxon tolabis. Unlike Oxydactylus longipes the 382 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161

TABLE 2 skull. If fused, the metapodials of T. brachyo- Measurements (in Millimeters) of the Mandible dontus would in fact be indistinguishable from and Lower Dentition of Tanymykter those of many Protolabis specimens. Another brachyodontus (Peterson) from Nebraska and feature reflecting relationship to Protolabis is Wyoming seen in the shape of the II-2. As mentioned CM F:AM before, the incisors of Tanymykter are weak, 664a 36594 laterally compressed, and strongly concave lin- gually. These incisors are virtually identical Mandible length (240.) 235.1 with those of Protolabis. In some Distance mental foramen below addition, top of ramus 15.3 14.1 referred specimens of T. brachyodontus (F:AM Depth jaw between P1 and P2 22.0 22.4 36594, fig. 4) have the incisors laterally situ- Depth jaw at P1 18.1 23.5 ated, forming near parallel rows, as in Pro- Depth jaw below anterior tolabis. Interestingly, were it not for the moiety of M3 37.6 30.0 brachyodont molars, the palate of T. brachyo- C-P1 diastema 19.8 18.7 dontus (F:AM 36594) would be indistinguisha- P1-2 diastema 26.5 20.5 ble from that of Protolabis. The more elongate Lower canine length 7.8 9.8 P 2 diastema and relatively deep rectus capitus P1 length 6.7 7.8 ventralis troughs are also similar to these fea- P2 length 10.4 11.7 tures in P2 width 4.1 4.5 Protolabis. P3 length 11.6 12.1 The only feature in which T. brachyodontus P3 width 4.7 5.7 is derived relative to primitive species of Pro- P4 length 12.3 13.2 tolabis is in the appressed roots of PI. We P4 width 6.7 7.3 conclude, on the basis of their shared derived M1 length 15.0 14.7 characters of narrow rostrum and the lateral ml width 11.8 11.1 expansion of the anterior nares, that Tanymyk- M2 length 18.5 17.5 ter and Protolabis are sister taxa. M width 13.0 13.1 M3 length (30.) 25.7 M3 width 13.4 13.7 PROTOLABIS COPE, 1876 P2_4 length 34.2 38.6 TYPE SPECIES: Protolabis heterodontus length 63.8 60.5 (Cope, 1874). P2-M3 length 99.0 100.3 INCLUDED SPECIES: Protolabis coartatus aCM 664 is the type of Tanymykter brachyodontus (Pe- (Stirton, 1929); Protolabis barstowensis Lewis, terson). 1968. KNOWN DISTRIBUTION: From early Hem- proportionally longer rostrum of T. brachyo- ingfordian to Clarendonian of the western dontus with longer premaxillaries and longer United States. anterior palatine foramina and a weakly devel- REVISED DIAGNOSIS: Protolabis differs from oped buccinator fossa are derived features, seen Procamelus in smaller size; less inflated brain- in the camelines. The narrower rostrum and the case; weaker sagittal and lambdoidal crests; more laterally expanded anterior nares in T. maxillary fossa not pocket-like; shorter paroc- brachyodontus are derived features that are cipital processes terminating above ventral shared with Protolabis. Furthermore, T. border of auditory bullae; greater rostral con- brachyodontus lacks the derived features seen striction; retention of I1-I2 in primitive species; in Oxydactylus longipes which link it with P3 and C rarely transversely compressed and Aepycamelus, namely, the elongate cervicals without sharp anterior and posterior edges; PI and limbs, with metapodials longer than the double rooted, recurved, and caniniform, but basal length of the skull. Tanymykter brachyo- never bladelike; P2-P4 shorter and more dontus and Protolabis retain the primitive con- brachyodont with P2 sometimes lost in most dition of short cervicals and limbs, with derived species; molars lower crowned, nar- metapodials shorter than the basal length of the rower and less anteroposteriorly expanded; 1978 HONEY AND TAYLOR: PROTOLABIDINI 383 weak to strong lateral flare and weak to strong of the two genera is quite beyond the scope of mesial tuberosity or shelf on angle of mandible; the present paper. angular process more subdued; coronoid proc- The skulls of Protolabis average smaller ess shorter with less rugose muscle attach- than those of Procamelus. The sagittal and ments; and metapodials proportionally lighter lambdoidal crests are usually weaker, and the with metatarsus longer than metacarpus. braincase is less inflated, in Protolabis. The Protolabis differs from Tanymykter in stronger crests and more inflated braincase in stronger buccinator fossa and generally nar- Procamelus result in a greater postorbital length rower rostrum; basisphenoid without well-de- relative to facial length. In the specimens meas- veloped elongate tuberosities; P2 with weaker ured little overlap was found in the postorbital/ metastyle and parastyle and discontinuous lin- preorbital ratio. This ratio ranged from 34 per- gual cingulum; P3 lingual cingulum weaker and cent to 44 percent in Protolabis, vs. 44 percent generally discontinuous, upper and lower mo- to 59 percent in Procamelus. lars taller crowned and anteroposteriorly longer, Both Protolabis and Procamelus possess a with weaker buccal ribs above and weaker met- maxillary fossa. The development of this fossa, astylids and lingual ribs below; mandibular however, differs between the two genera. In angle more ventrally expanded; metapodials Protolabis the fossa varies in shape from ante- fused; proximal phalanx less convex dorsally roventrally elongate to subcircular, and may and distal articular surface less grooved and cover a wide area on the maxillary centered more anteriorly expanded. above the infraorbital foramen, and just beneath Differs from Michenia in larger size; longer the nasal. This fossa varies from a deep depres- braincase; stronger I1-I3 in primitive species; sion with well-defined rims, to a very shallow stronger and more caniniform C; more antero- concavity. It is never pocket-like. In contrast, posteriorly expanded MI; deeper, more ven- the fossa in Procamelus is usually deeply de- trally produced symphysis; deeper horizontal pressed and pocket-like. A few Procamelus, ramus; and relatively stouter metapodials. See however, have only a shallow trough with a p. 418. posterodorsal slit at the position of the pocket. DISCUSSION: As discussed by Frick and Tay- The maxillary fossa in Procamelus is also over lor (1971, p. 3), Protolabis was originally diag- the infraorbital foramen, but is generally situ- nosed by Cope (1876 p. 144) as a camel ated lower on the maxillary than in Protolabis. retaining a full set of upper incisors, whereas Both genera usually have a shallow trough Procamelus Leidy (1858 p. 89), was defined by trending anteroventrally from the maxillary Cope as a camel that had lost the first two fossa. upper incisors. This criterion of the presence or The paroccipital process of Protolabis is the absence of upper incisors has been consid- more weakly developed than in Procamelus and ered diagnostic by authors since Cope; indeed, almost never projects below the ventral edge of other dental and skeletal features have usually the lateral plate of the auditory bulla. In con- assumed secondary significance. The assump- trast, the paroccipital process of Procamelus is tion that Protolabis always retains a complete much more strongly developed and almost al- set of upper incisors has led to a problem in ways projects below the lateral plate of the distinguishing between Protolabis and "Pro- bulla. The troughs on the basioccipital for the camelus" coartatus. Recently, Frick and Taylor origin of the rectus capitus ventralis muscle are (1971, p. 5) have added the differences in the generally more strongly developed and deeper metapodial proportions and molar expansion to in Protolabis. In later occurring Protolabis the differential diagnosis of Protolabis and Pro- skulls the troughs often occur as deep, elongate camelus. In the present paper these and other depressions on either side of a sharp median differences will be examined and evaluated ridge. Procamelus tends to have shallower more fully, to clarify the distinctions between troughs separated by a wider, more rounded these two genera. Skull, dental, mandibular, median ridge. Immediately anterior to each and metapodial morphological features are trough is a tuberosity, probably for the origin herein considered; a complete anatomical study of the longus capitus. These tuberosities are 384 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161 often elongate and strongly developed in Pro- always lost, or unerupted, but it is clear that in camelus, whereas in Protolabis they tend to be many adult individuals of Protolabis the first weakly developed. two superior incisors are absent. Their former Constriction of the rostrum posterior to the positions are occasionally indicated by healed canine is a progressive feature in Protolabis. alveoli. Extreme constriction of the rostrum oc- This constriction is associated with the pro- curs in association with the loss of I1 and 12; nounced development of a buccinator fossa on both are derived features of Protolabis that at- the ventrolateral border of the maxillary ante- tain maximum specialization in P. coartatus. tior to the P2. Examination of many specimens In his analysis of variation in Procamelus in the F:AM collection reveals that rostral con- grandis, Webb (1969, p. 158) noted differences striction and buccinator fossa development is in the character of the caniniform teeth which least in the early, primitive samples and great- he attributed to sexual variation. Specifically, est in the late, derived samples. The palate Webb considered specimens with large, coni- narrows rapidly immediately posterior to the cal, recurved teeth to be males, and those with canine, reaching maximum constriction be- laterally compressed, bladelike, caniniform tween the PI and P2. This is in part correlated teeth females. These observations were con- with the deepening of the buccinator fossa to a firmed by the examination of a large number of maximum depth between the PI and P2. Palatal Procamelus specimens in the Frick collection, constriction is less pronounced and occurs more but no comparable consistent dichotomy was posteriorly in Procamelus. This correlates both observed for Protolabis. In all the Protolabis with the weaker development of the buccinator specimens examined, the I3 and C are conical fossa and a much larger P1. In Procamelus, and recurved, and never as compressed as in only slight constriction occurs immediately be- Procamelus; moreover, they lack the sharp an- hind the canine, and most of the narrowing is terior and posterior flanges that are characteris- behind the large PI. This results in the rostrum tic of the females of Procamelus. The lower of Procamelus being more constant in width canine of Protolabis is occasionally bladelike anterior to the PI, whereas the rostrum of Pro- and consequently, small procamelines are tolabis flares greatly anterior to this point. Gen- sometimes difficult to distinguish from Pro- erally, in Procamelus the rostral width between tolabis by this character alone. In all Pro- the canines is either equal to or greater than the camelus examined, the P1 is bladelike, width between the 13s. In contrast, the inter- caniniform, and recurved, whereas in Pro- canine distance in Protolabis is always equal to tolabis it is low crowned, blunt, and less re- or less than the inter-I3 distance. An examina- curved. Although lateral compression of the tion of many specimens shows that the rostrum caniniform teeth does not appear to be a sexual of Procamelus is almost always wider than that character in Protolabis, the size of these teeth of Protolabis; even specimens of primitive taxa is believed to reflect sexual dimorphism, with of Protolabis have rostra relatively narrower the larger, heavier toothed specimens from each than those of most Procamelus. sample presumed to the males. Protolabis was originally diagnosed as hav- The presence of a double-rooted P' in Pro- ing a full set of upper incisors, and these teeth tolabis and a single-rooted P' in Procamelus is are present in all but the most derived taxa. In useful in separating the two genera. Webb general, the I1 and 12 are small and peglike. (1969, p. 148) included the presence of a dou- Occasional loss of I1 or both I1 and 12 occurs as ble-rooted P' in his diagnosis of Protolabis, but individual variation in earlier samples, but by because he considered coartatus to be a mem- the Clarendonian both of these teeth are usually ber of Procamelus, Webb could not always use lost or unerupted. This was recognized by Frick this character to separate the two genera. In and Taylor (1971, p. 4), when they referred Protolabis the roots vary from widely flared to Cope's specimen of "Procamelus occidentalis" closely appressed or fused. If the roots are from the "Santa Fe marls" of New Mexico to fused, however, a mesial or lateral groove indi- Protolabis. It is uncertain if these teeth are cates this fusion, and a cross section of the 1978 HONEY AND TAYLOR: PROTOLABIDINI 385 tooth usually shows the individual roots. In can be consistently used to distinguish the mo- contrast, Procamelus has a single-rooted, can- lars of Protolabis and Procamelus. Both genera iform P1. Generally, the PI of Procamelus is overlap to some degree in the development of transversely compressed with sharp anterior and ribs, styles, and stylids, and in the flatness of posterior edges, though in some of the larger the buccal walls above and the lingual walls specimens the PI is robust and conical. In Pro- below. However, there are significant differ- camelus the PI is often as large as the I3, which ences in the proportions of the molars. Frick itself is only slightly smaller than the canine. and Taylor (1971, p. 5) pointed out that the On the other hand, the PI (like the P1) of molars of Procamelus are more anteroposte- Protolabis is not caniniform, and never as large riorly elongate and higher crowned than those as the I3, and has a low, blunt crown. of Protolabis. Specifically, in a similar wear Prototabis exhibits a progressive reduction stage MI and M2 of Procamelus are longer in the size of the P2-P4. P2 shows the greatest relative to the length of M3 than those of Pro- reduction, and in its most derived state, is very tolabis. The molars of Procamelus also tend to small and occasionally lost. Patton (1969, pp. be relatively wider than those of Protolabis. 151, 162-163) noted premolar reduction as Mandibular rami of Protolabis, particularly characteristic of the Protolabis lineage, but those of earlier samples, tend to have shorter, based much of his hypothesis on species now more upright symphyses than those of Pro- referred to Miolabis by Frick and Taylor (1971, camelus. In general, the mandible of Pro- p. 4). camelus has a long symphysis which forms a In general, P2 of Protolabis is lower rather shallow angle with the horizontal body crowned than that of Procamelus, and has a of the ramus. In dorsal view, the posterior low, central protoconid. In Procamelus, the border of the relatively short symphysis in Pro- protoconid is higher and has steep crests de- tolabis is found to be either beneath or anterior scending anteriorly and posteriorly. Patton to PI; on the longer, shallower symphysis of (1969, p. 156) noted that Procamelus occasion- Procamelus the posterior border unites either ally has reduced premolars, as seen in his fig- beneath or posterior to P1. However, there are ure 16 of Procamelus occidentalis. Never- occasional exceptions with some individuals of theless, Clarendonian specimens of Protolabis Protolabis having a long, shallow symphysis. usually have a more highly reduced P2. Reduc- The ventral border of the mandibular ramus in tion in size is also usually applicable to P3 and Protolabis is generally less convex and P4, P3 and P4 of Protolabis are also lower straighter than in Procamelus. Immediately be- crowned and generally have a less expanded hind the symphysis the lower border of the talonid than those of Procamelus. With wear, ramus is concave in both genera, but in Pro- the talonid of P3 and P4 of Procamelus acquires tolabis this concavity extends posteriorly for a a more pronounced buccal bulge than in Pro- greater distance than in Procamelus. camelus. In Procamelus the width of the pos- Lateral flare of the posteroventral border of terior cusps of P3 and P4 is almost always the mandibular angle occurs repeatedly to vary- greater than that of the middle cusps, a fact ing degrees in many individuals of Protolabis. pointed out by Webb (1969, p. 158). In Pro- This flare is greatest in latest known samples. tolabis the central cusp tends to be equal in An occasional jaw of Procamelus in the Frick width or wider than the posterior cusps. There collection also exhibits a slight flare of the are exceptions though, including the type of P. mandibular angle, but this condition occurs angustidens, in which the posterior cusps of P4 rarely and never to the same degree as in Pro- (but not P3) are wider than the middle cusp. tolabis. An even more consistent feature in Also, the degree of lingual inflection of the Protolabis is the development of a mesial flar- anterior cusp of P3 and P4 was found to be ing or tuberosity immediately anterior to the variable in both genera; however greater inflec- lateral flaring on the posteroventral border of tion was generally present in Protolabis. the mandibular angle. This mesial flare appears No morphological features are known that to be invariably present in later samples. Mes- 386 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161 ial inflection of the mandibular angle occurs proximal and median phalanges from Stewart very rarely in Procamelus, and when present, it Valley (early Clarendonian), Cedar Mountain is manifested only as a slight bending of the region, Nevada. bottom part of the ramus. Never is the develop- KNOWN DISTRIBUTION: Early Clarendonian ment of the mesial tuberosity as strong as in of Stewart Valley and Fish Lake Valley, Ne- Protolabis, which in its derived state occurs as vada, Clarendonian of Lapara Creek, Texas, a prominent anterodorsally trending shelf. This and Milk Creek, Arizona. mesial tuberosity is evidenced on the lateral REFERRED MATERIAL: From Fish Lake Val- surface of the mandible by a small concavity- ley beds, Esmeralda County: UCMP 29608, on the lower border just anterior to the lateral palate with roots of P2-p4, partial M1 and M2 flare. This concavity is absent in Procamelus. (Stirton, 1929, fig. 4); UCMP 29605, fragmen- The angular process of the mandible in Pro- tary right ramus; UCMP 29612, palate with left tolabis is usually less developed than in Pro- P3-P4, left M1-M3, right P2-P4, right MI-M3 camelus. In Procamelus it is often very (Stirton, 1929, fig. 5). knoblike and mesially inflected, whereas in From Lapara Member of Goliad Formation: Protolabis it tends to be subdued with only a UTBEG 31152-332, mandible with P4-M3, roots slight mesial inflection in some specimens. of P2-P3, P1 alveolus (Patton, 1969, figs. 14-15); Consequently, the notch on the back part of the UTBEG 31081-261, 31081-295, 31081-612, jaw in Protolabis is shallower than in Pro- 31081-525, UTBEG 31132-319, all metapodials. camelus. Furthermore, the coronoid process in From Manzanita Quarry, 10 feet below Protolabis is anteroposteriorly narrower and Goldie's Honeypot Quarry, Milk Creek Forma- does not extend as high above the co;Ndyle as in tion (Clarendonian), Milk Creek area, Yavapai Procamelus. In Protolabis, the coronoid proc- County, Arizona: F:AM 73290, palate with P2 ess generally has less rugose muscle attach- (broken)-M2; F:AM 73300, partial mandible ments along its length, and this is correlated with 1-13, C-P3 (all broken) and P4-M3; F:AM with a relatively smaller temporal fossa. 73257, left partial ramus with P3-M3 (P4 As discussed by Frick and Taylor (1971, p. broken); F:AM 73266, partial mandible with I, 5), the length of the metatarsus in Protolabis is (root) -I2, 13-P, (all broken) and P3 (root)-M3; always greater than the length of the metacar- F:AM 73265, right partial ramus with P3 pus, whereas the reverse is true in Procamelus. (broken)-M3; F:AM 73297, right partial ramus Moreover, the metapodials of Protolabis are with P2-P3 alveoli and P4-M3; F:AM 73299, much shorter and less robust than those of Pro- symphysis with I, (broken)-C and P, (broken); camelus, with the length of the metapodials in F:AM 73366, right partial ramus with C-P1 and Protolabis always less than the basal length of P3-M3 (P4 broken); F:AM 73375, right partial the skull and proportionally shorter than in Pro- ramus with incomplete symphysis with camelus. C(alveolus) and P2 (unerupted)-M1 and M2-M3 broken; F:AM 99634, left partial ramus with symphysis with (broken alveoli), P3 (root)- Protolabis coartatus (Stirton, 1929) I,-C Tables 3-6 M3; F:AM 73374, left partial ramus with P2-M2 Figures 5-9, and M3 broken; F:AM 73299, left ramal frag- Procamelus near gracilis Merriam, 1916, p. 189, ment with M3; F:AM 73333, immature palate figs. 30a-b, 31a-c, 32, 33. with dP2-M1; F:AM 73332, right immature par- Procamelus coartatus Stirton, 1929, p. 295, figs. tial mandible with dP2-dP4; F:AM 99631, right 2-4. immature partial maxilla with dP2-M1 (all bro- Procamelus cf. gracilis Stirton, 1929, p. 297, fig. 5. ken); F:AM 73335, right immature partial max- Protolabis notiochorinos Patton, 1969, p. 152, figs. 14-15, table 11. illa with dP2-M'; F:AM 73337, right immature partial maxilla with dP2-M' (MN broken) and TYPE: UCMP 19820, rostral part of skull right and left immature partial rami with dP2- with F, C, and PI; mandible with complete M2; F:AM 99629, left immature partial maxilla dentition; distal end of humerus; radius-ulna; all with dP3-M' (all broken); F:AM 73346, left carpals except for cuneiform; metacarpal; and immature partial maxilla with M'-M2 broken 1978 HONEY AND TAYLOR: PROTOLABIDINI 387 and right and left immature partial rami with with symphysis and Ii-C; F:AM 73334, right dP3-dP4 and MI-M2 broken; F:AM 73347, right immature partial maxilla with dP3 (alveolus)-MI immature ramus with dP2dP3 and dP4 broken; and right immature partial ramus with dP3-M1 F:AM 73342, right and left immature partial all broken; F:AM 99630, right and left imma- rami with incomplete symphysis with dI2-dI3, ture maxilla with dP2-M2; F:AM 99628, right dC root and dP2 (aleveolus)-dP4; F:AM 73339, immature partial maxilla with dP2 (root)-dP4; right immature partial ramus with dP2 alveolus- F:AM 73331, left immature partial maxilla with dP4 and M1 broken; F:AM 73343, left immature dP2 (root)-dP4; F:AM 73344, right immature partial ramus with symphysis with dII-dC, dP3 partial ramus with dP2 (root)-MI; F:AM 99625, (broken)-MI; F:AM 73341, left immature partial right immature partial ramus with dP3-Ml; ramus with dP2-Mi and M2 broken germ; F:AM F:AM 99621, right immature partial ramus with 99624, right immature partial ramus with dP4- dP3 (alveolus)-M3 (erupting); F:AM 73357, MI; F:AM 99619, right immature partial ramus right partial ramus with dP3 (alveolus)-M2; with dP3-dP4 and M1 broken; F:AM 73345, F:AM 99641, right immature partial ramus with right and left immature partial rami with dP3- dP4-M2 all broken; F:AM 73423, left metacar- dP4 and M, broken; F:AM 99623, left immature pus; F:AM 99647, left incomplete metacarpus; partial ramus with dP3(broken)-dP4 and M1 F:AM 99648, left proximal one half metacar- broken germ; F:AM 73348, left immature par- pus; F:AM 99651, distal part of metacarpus; tial ramus with dP2-dP4 and Ml broken; F:AM F:AM 73412, left metatarsus; F:AM 73412A, 99620, right and left immature partial rami right proximal one half metatarsus; F:AM with dP4-M, and M2 broken; F:AM 73355, im- 73317, right metatarsus; F:AM 99646, right in- mature partial mandible with ?I1 germ and dP3- complete metatarsus; UALP 9404, partial right M1 and M2broken germ; F:AM 73310, right ramus with P49 M1, M2 (broken); UALP 9405, incomplete metacarpus; F:AM 73311, left meta- mandibular symphysis with 11-13, C, P, (roots); carpus; F:AM 73312, left metacarpus; F:AM UALP 9386, right M'-M2; UALP 9374, right 73313, full length broken mesial half of left articulated astragalus, calcaneum, navicular and metacarpus; F:AM 73315, left metacarpus; cuboid; UALP 3628, right metatarsal; UALP F:AM 73323, right metacarpus; F:AM 73343, 9375, right distal tibia and metatarsal; UALP right incomplete metatarsus; F:AM 73297, right 9376, left metatarsal; UALP 9378, right proxi- metatarsus; F:AM 73360, left metatarsus; mal metacarpal; UALP 9379, distal metapodial; F:AM 73318, left distal end humerus and radius UALP 9380, left astragalus; UALP 9391, left and ulna; and F:AM 73319, right crushed ra- distal humerus; UALP 9392, left distal dius and ulna, and associated broken proximal humerus; UALP 9393, left proximal radius- end metacarpus. ulna, right distal radius-ulna, right distal tibia; Shields Ranch Quarry (=Stein Gallery, UALP 9394, left astragalus; UALP 9395, right UALP locality 1-7), Milk Creek Formation, distal radius-ulna; UALP 9397, right radius- same horizon as Manzanita Quarry (Clarendo- ulna; UALP 9398, left distal humerus; UALP nian), Milk Creek area, Yavapai County, Ari- 9399, left proximal radius; UALP 9400, left zona: F:AM 73296, left premaxilla-maxillary distal tibia, proximal phalanx; UALP 9401, fragment with 12(alveolus)-P1; F:AM 99626, right proximal radius-ulna; UALP 9396, left right partial maxilla with M1-M2 and M3 distal tibia; UALP 9402, left distal radius-ulna; broken; F:AM 99627, right partial maxilla with and UALP 9403, left distal radius-ulna. M1-M2; F:AM 73369, right partial ramus with Trail Prospect, Milk Creek Formation, same P3-M3; F:AM 73268, right and left partial rami horizon as Shields Ranch Quarry (Clarendo- with incomplete symphysis with C(broken)-P1, nian), Milk Creek area, Yavapai County, Ari- P3-M1 and M2-M3 both broken; F:AM 74468, zona: F:AM 73291, right partial maxilla with mandible with I,-M3; F:AM 73392, left partial P2-P3 broken and P4-M3 (M' and M3 broken); ramus with symphysis, Ij-I3 (all broken), C (al- and F:AM 73356, left immature ramus with II, veolus)-P1, P3 (root)-P4 and M,-M2 broken; dl2-dC roots and dP3-dP4 broken and MI. F:AM 99632, left partial ramus with M1M3 all Goldie's Honeypot Quarry (UALP locality broken; F:AM 99639, right ramal fragment 7498), Milk Creek Formation, 5 feet above 388 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161

Shields Ranch horizon, Milk Creek area, ble with II, I2-C alveoli and roots and P1-M3, Yavapai County, Arizona: UALP 9022, partial and atlas, axis, three other cervical vertebrae, skull with 13 (root), C, PI, P3-P4, MI (broken), metatarsus and phalanges; F:AM 73378, skull M2 and M3 (broken); UALP 9023, right partial with 13-M3; F:AM 73378A, right femur; F:AM palate with M'-M3; UALP 9034, left partial 73385, skull with I-M3; F:AM 73285, skull palate with dP3-dP4 and M1; UALP 9010, right with I-M3; F:AM 73390, skull with I3 (al- partial dentary with P2-P4 and Ml (broken); veolus)-M3; F:AM 73377, skull with F-C al- UALP 9032, mandibular symphysis with I1_L3 veoli, PI broken and P2-M3; F:AM 73439, left roots and C; UALP 9016, left partial ramus partial maxilla with P2-PI all erupting and M1; with dP3 (roots), dP4 and MI; UALP 9005, left F:AM 73309, mandible with II alveolus, 12-13, dentary and symphysis with right dI-dl2, right C alveolus and PC-M3; F:AM 73384, mandible II, left dP3 (roots), left dP4 and left M1-M2; with 129 C and P3-M3; F:AM, 73376, mandible UALP 9012, right dentary with dP3-dP4 and M1- with l1-C alveoli, PI unerupted and P2-M3; M2; UALP 9006, right fragmentary dentary F:AM 73306, right ramus with symphysis, I1-PI with dP2-dP4; UALP 9015, right fragmentary and P3-M3; F:AM 73308, mandible with 13-M3; dentary with dP3-dP4 and Ml; UALP 9019, right F:AM 73383, mandible with incomplete sym- fragmentary dentary with dl and dP3-dP4; physis and P3-M3; F:AM 73307, left partial UALP 9007, left fragmentary dentary with dP3- ramus with P3 (broken)-M3; F:AM 99635, left dP4; UALP 9009, mandible with d12-d13, I,, 12 partial ramus with P4-M3 all broken; F:AM (erupting), dC, dP3 (roots), dP4, Ml and M2 99638, left partial ramus with P4-M3 all broken; (broken); UALP 9011, right partial dentary with F:AM 99640, right ramal fragment with M3; P1, P3-P4 M1-M2 and M3 (broken); UALP 9017, F:AM 99637, right partial ramus with P2-M2 all left partial dentary with C (root), PI, P2 broken; F:AM 99633, right ramal fragment (alveolus), P3-P4; (broken) and M,-M3; UALP with P4-M2 all broken; F:AM 99636, left ramal 9013, left partial dentary with dP2 (alveolus), fragment with incomplete symphysis, I-I3 dP3 (broken), dP4 and MI-M2; UALP 9014, alveoli and C-P2 all broken; F:AM 99625, left right partial dentary with dP2 (alveolus), dP3 immature partial ramus with dP3-Ml; F:AM (broken), dP4 and MI; UALP 9008, right partial 73415, right and left articulated radii through dentary with P3-P4 and M,-M3; UALP 9024, left third phalanx; F:AM 73413, left articulated partial palate with dP3-dP4 and M1-M2; UALP metacarpus through third phalanx; F:AM 9443, partial rostrum with I3, C, PI, and P3-P4; 73414, right metacarpus; F:AM 73416, right UALP 9114, left astragalus; UALP 9108, left and left metacarpi and two or more associated proximal metacarpal; UALP 9107, left proximal individuals including F:AM 73416A, crushed metacarpal; UALP 9129, left astragalus; UALP and broken metapodial; F:AM 73416B, left 9106, distal metapodial; UALP 9110, left distal humerus; F:AM 73416C, left radius and ulna; radius-ulna; and UALP 9125, right distal tibia. F:AM 73416D, crushed radius and ulna; F:AM Cliff Prospect, Milk Creek Formation, strat- 73420, right metacarpus and articulated pha- igraphically between the Milk Creek Quarry langes; F:AM 73422, left metacarpus; F:AM and Shields Ranch Quarry horizons (Clarendo- 73421, left metacarpus; F:AM 73417, left meta- nian), Milk Creek area, Yavapai County, Ari- carpus; F:AM 73418, left metacarpus; F:AM zona: F:AM 93314, left metacarpus. 73419, right metacarpus; F:AM 73428, left Deep Spring Quarry (=Johny Post, UALP metatarsus; F:AM 99645, left broken metatar- locality 1-la), about 69 feet above Manzanita sus; F:AM 73429, right metatarsus and right Quarry and 34 feet below Milk Creek Quarry, calcaneum; F:AM 73431, right metatarsus; Milk Creek Formation (Clarendonian), Milk F:AM 73430, right metatarsus; F:AM 73435, Creek area, Yavapai County, Arizona: F:AM right metatarsus; F:AM 73434, right metatar- 73438, articulated skull with 13-M3 and mandi- sus; F:AM 73433, left metatarsus; F:AM ble with I1-I3 alveoli, C-P, and P3-M3; F:AM 99644, broken metatarsus; F:AM 99643, 73379, articulated skull with 13-P' and P2-M3 crushed metatarsus; F:AM 73432, broken meta- and mandible with II (alveolus)-M3; F:AM tarsus; F:AM 73425, right humerus; F:AM 73437, skull with 1P-P1 alveoli, P2-M3, mandi- 73424, right humerus; F:AM 73426, left radius 1978 HONEY AND TAYLOR: PROTOLABIDINI 389 and ulna; F:AM 73427, left crushed femur; M3 broken; F:AM 73295, partial palate with F:AM 99652, right tibia; F:AM 73413, PI-M2; F:AM 73303, partial mandible with C, scaphoid and magnum; F:AM 73436, partial P1 root and P2(root)-M3 (P3-M, broken); F:AM pelvis; F:AM 99655, left partial humerus; 73302, left partial ramus with symphysis, IF-M2 F:AM 99656, right and left radius and ulna; and M3 broken; F:AM 73267, right partial F:AM 99657, left femur; F:AM 101454, right ramus with symphysis, I-P, and P3-M3; F:AM crushed femur; F:AM 101472, right femur; 73298, left partial ramus with P, (root)-M2 and F:AM 101473, right femur; F:AM 101474, right M3 broken; F:AM 73301, left partial ramus and left crushed femora, associated with F:AM with P3-M, roots and M2-M3; F:AM 73304, 101473 of a larger individual; F:AM 101475, right partial ramus with P2-M2 and M3 broken; left crushed femur; F:AM 101476, left femur; F:AM 73263, right and left partial rami with UALP 5023, partial skull with Pl-P4 and M1- P3-M3; F:AM 73264, right partial ramus with P3 M3; UALP 1441a, right articulated humerus, (broken)-M3; F:AM 73264A, left partial ramus radius-ulna, metacarpal, 2 proximal phalanges with P2(germ), P3-M2 and M3 (broken); F:AM and sesamoid; UALP 1441b, left articulated 99642, right partial ramus with P2 (alveolus)- scapula (glenoid fossa), humerus, radius-ulna, M2 and M3 broken; F:AM 73289A, left partial scaphoid, lunar, cuneiform, magnum, un- ramus with P2 (broken)-M3; F:AM 73271, left ciform, trapezoid, pisiform, metacarpal, 2 partial ramus with P3-Mi and M2-M3 broken; proximal phalanges, 2 medial phalanges, 2 dis- F:AM 73274, symphysis with I,-C and P1 tal phalanges, and 3 sesamoids; UALP 9355, broken; F:AM 73272, symphysis with IF-I3 and distal left tibia; UALP 9356, left astragalus and C-PI broken; F:AM 73273, symphysis with II calcaneum; UALP 9357, right proximal femur; (broken)-C and P1 root; F:AM 73271B, sym- UALP 9358, left proximal and distal femur; physis with C(root)-P1 broken; F:AM 73336, UALP 9359, right? metatarsal; UALP 9360, right immature partial maxilla with dP2-dP3 and left distal humerus; UALP 9361, 2 proximal dP4-Ml broken; F:AM 73389, right immature phalanges, UALP 9355-UALP 9361 apparently ramal fragment with dP2(broken)-dP3 and dP4 associated; UALP 9362, right proximal broken; F:AM 73303, right incomplete meta- humerus; UALP 9363, left proximal metatarsal, carpus; F:AM 73316, right and left incomplete 1 proximal phalanx, 2 medial phalanges, 2 dis- metacarpi; F:AM 73289B, left incomplete met- tal phalanges; UALP 9364, right metacarpal, 2 atarsus; F:AM 73321, right articulated distal proximal phalanges, 2 medial phalanges and 1 end humerus and proximal half radius and ulna; distal phalanx; UALP 9365, right distal radius- F:AM 73320, left crushed radius and ulna; ulna, right proximal metacarpal, 2 proximal F:AM 73624, left proximal part of radius and phalanges; UALP 9366, left astragalus and cal- ulna; and F:AM 73322, left distal end tibia. caneum, 1 distal metapodial, 5 proximal pha- REVISED DIAGNOSIS: Differs from P. hetero- langes, 2 medial phalanges and 1 distal dontus and P. barstowensis in extreme trans- phalanx; UALP 9367, right astragalus and cal- verse constriction of rostrum posterior to PI; 1- caneum, left distal tibia and 2 distal phalanges; 12 lost; p2_p3 without parastyle; molars with less UALP 9369, right proximal femur; UALP sharp styles and less prominent ribs; P1 short, 9370, right astragalus; UALP 9371, right cal- peglike, occasionally suppressed or absent, caneum; and UALP 9372, right astragalus, with closely appressed roots; P2 very small and proximal left metatarsal, left cuneiform, proxi- occasionally absent, with closely appressed and mal phalanx and medial phalanx. sometimes fused roots; P3-P4 relatively smaller Milk Creek Quarry, about 103 feet above the and simpler; molars (especially M3) anteropos- Manzanita Quarry level, Milk Creek Formation teriorly shorter; mandibular ramus with promi- (Clarendonian), Milk Creek area, Yavapai nent lateral flare on angle of mandible; and County, Arizona: F:AM 73292, anterior part of prominent mesial tuberosity or shelf on ventral skull with C-M3 and F:AM 73292A, associated angle of mandible. left partial ramus with P2-M3; F:AM 73293, DESCRIPTION AND COMPARISON: Knowledge crushed anterior part of skull with 13-M3; F:AM of the cranial structure of Protolabis coartatus 73284, partial palate with P2(broken)-MW and has heretofore been based on the fragmentary 390 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161 rostral and palatal remains from Stewart and is unexpanded and short relative to the length Fish Lake valleys. The many complete and par- of the skull; it is only slightly more inflated tial skulls which exist in the collections at the than that of P. heterodontus and is approx- American Museum of Natural History and the imately the same size as the above specimen University of Arizona now make possible a that Cope described from New Mexico. more thorough description of the cranial anat- The anterior rostral region is best preserved omy of this species. Because all skulls have in the type (UCMP 19820) from Stewart Val- suffered some postmortem distortion, the fol- ley. As Stirton (1929, p. 295) observed, the lowing description is based upon the examina- rostrum is distinctive for its extreme lateral tion of all specimens. constriction. This constriction is most apparent The length of the skull of P. coartatus is in the palatal view, where the roof of the palate intermediate between that of P. barstowensis narrows continuously from I3 to a point about and the skull described from New Mexico by midway between P' and P2. The narrowest Cope (1877, p. 329) under the name of Pro- point of the rostrum posterior to P1 is marked camelus occidentalis. The basal length averages by a sharp maxillary crest on each side of the 318.4 mm. All the Milk Creek specimens ex- roof of the mouth; it is between these crests hibit some dorsoventral flattening, but the face that the palatal width was measured. On the appears to have been relatively deep with little type specimen of P. coartatus the narrowest maxillary inflation anterior to the orbits. Ante- palatal width is 8.5 mm., whereas the width of rior to the cheek teeth, the muzzle is strongly some referred specimens from Milk Creek constricted. The skull is wide across the orbits measures as little as 5.0 mm. These measure- with marked postorbital constriction imme- ments are less than half of those of P. hetero- diately in front of the braincase. The braincase dontus and P. barstowensis. The palate of the

A

FIG. 5. A-C. Protolabis coartatus from Manzanita Quarry (Clarendonian), Arizona. A. F:AM 73315, metacarpus, anterior view. B. F:AM 73300, lateral view. C. F:AM 73266, lateral view, reversed. All X1/2. 1978 HONEY AND TAYLOR: PROTOLABIDINI 391 type and several of the referred specimens are other along a suture 86.0 mm. in length (F:AM undistorted, demonstrating that this constriction 73438). Near the posterior end of each nasal is real. This palatal narrowing is associated are one or two small foramina. Skulls of P. with the pronounced development of the buc- coartatus measure from 145.0 to 160.0 mm. cinator fossa on the ventrolateral border of the across the frontals between the orbits. On the maxilla, above and behind Pl. When the fossa anterodorsal border of the orbits is a mesially is undistorted, it is arcuate in outline, begin- directed notch about 16.0 mm. long. Because ning above the canine and arching posteriorly of crushing the true dimensions of the orbits in a smooth curve about halfway up the face, cannot be determined, but they appear to be descending and disappearing above P4. The about the same as those of P. heterodontus. fossa is deepest between PI and p2, and its The zygomatic arch in P. coartatus meas- surface tends to be relatively smooth. A partial ures about 15.0 mm. dorsoventrally at the skull (AMNH 9118) from northeastern Color- widest point and is of similar proportions to ado referred to P. heterodontus also shows this that of P. heterodontus. It varies from rela- fossa well developed. tively flat to slightly arched. Anteriorly, a Above the buccinator fossa the premaxillary strong zygomatic ridge extends onto the face to and maxillary bones flare and roll outward and a point above the posterior moiety of M2. This upward, reaching their maximum expansion im- ridge is higher on the face and more strongly mediately behind the external nares. In dorsal developed than in P. heterodontus (AMNH view the rostrum narrows rapidly posteriorly, 9426). The anterior edge of the zygomatic reaching its maximum constriction above the process of the squamosal contacts the maxilla maxillary fossa and anterior to the nasofrontal less than halfway forward along the ventral suture. Vertical crushing has obscured the de- edge of the orbit. In some skulls (F:AM 73438 tails of this fossa in most specimens except and UALP 1449), a single large subsquamosal F:AM 73379 from Milk Creek. The maxillary foramen is present on the posterior root of the fossa is high on the maxillary bone, beginning zygomatic arch. at a point above the anterior edge of p2. The The supraoccipital tilts posterodorsally and fossa increases in depth posteriorly, reaching a widens laterally, with the maximum width be- maximum depth above and immediately behind tween the paroccipital processes. The the infraorbital foramen. Posteriorly, the max- lambdoidal crest is strong. The median occipi- illary fossa decreases in depth and disappears tal ridge is narrow at the top and widens ven- above the anterior lobe of M3. The superior trally to equal the width across the top of the border of the fossa is marked by a shallow foramen magnum. At its junction with the fora- groove separating the maxillary and nasal men magnum, the median ridge occasionally bones. The maxillary fossa of P. coartatus is bears two small tubercles. On a nearly un- broad and open and resembles that of the par- distorted occiput (UALP 1449), the height of tial skull of P. heterodontus (AMNH 9118). the foramen magnum measures 22.3 mm., and Some Protolabis specimens in the F:AM col- the width 25.0 mm. In form, the occipital con- lection were observed to have a more deeply dyles of P. coartatus are similar to those of P. depressed maxillary fossa with a well-defined heterodontus. The basal articular surfaces are rim. The infraorbital foramen is situated above convex posteriorly and concave anteriorly. the anterior moiety of M1, as in the type of P. They fail to meet at the midline and are slightly heterodontus. As mentioned above, P. coar- upturned at their junction with the basioccipital. tatus has a relatively deep face, measuring ap- The deeply excavated mastoid fossae lie later- proximately 72.0 mm. from the palatal border ally to the condyles, and are separated by a at a point behind PI to the top of the nasals distance of 34.0 mm. in an undistorted occiput (F:AM 73379). (F:AM 73285). This compares with a measure- In the type rostrum of P. coartatus the pre- ment of 47.0 mm. in a referred skull of P. maxillary-nasal suture measures 52.5 mm.; in heterodontus (AMNH 4432). some of the Milk Creek specimens it may be as The basisphenoid in P. coartatus possesses short as 28.0 mm. The nasals contact each two troughs for the insertion of the rectus cap- 392 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161

C

D < ,

FIG. 6. A-D. Protolabis coartatus from Milk Creek Quarry (Clarendonian), Arizona. A. F:AM 73292, lateral view. B-C. F:AM 73302. B. Occlusal view. C. Lateral view. D. F:AM 73267, lateral view. All X1/2. itus ventralis muscle. The troughs are separated Between the basioccipital and the base of by a narrow anteroposterior ridge and vary the bulla is a compound foramen (UALP 1449) from elongate to circular. They are occasionally representing the conjoined posterior lacerate fo- very deeply depressed, as in P. heterodontus. ramen and posterior opening of the carotid The troughs decrease in depth posteriorly until canal. The posterior part, representing the pos- meeting the occipital condyles. In P. hetero- terior lacerate foramen, is the larger of the two. dontus the basisphenoid troughs end before In P. heterodontus these two foramina are sep- reaching the condyles, resulting in a narrow arated by a thin wall of bone. About 7.0 mm. width of undepressed basioccipital anterior to posteromesially from the posterior lacerate fora- the condyles. men is the hypoglossal foramen. The hypo- 1978 HONEY AND TAYLOR: PROTOLABIDINI 393 glossal foramen in P. coartatus is situated near width of the palate between p2 varies from 16.8 the rim of the condyloid fossa, just beneath the mm. to 22.0 mm. As was pointed out by anterior lip of the occipital condyle, and at the Henshaw (1940, p. 20), the restored width of same level as the posterior lacerate foramen. It P. coartatus specimen (UCMP 29612) is proba- is situated similarly in P. heterodontus, but in bly too great. These palatal widths are slightly some skulls of Protolabis it is deeper inside the greater than in P. heterodontus (F:AM 32880) condyloid fossa. At the bottom of the con- from northeastern Colorado. dyloid fossa and against the medial wall is the UPPER DENTITION: The rostrum of the type condyloid foramen. In both P. coartatus and P. of P. coartatus lacks the first two upper in- heterodontus the condyloid foramen is about cisors, and it was for this reason that the spe- the same size as the hypoglossal foramen. cies was originally allocated to Procamelus. The tympanic bulla of P. coartatus is The referred specimen from Fish Lake Valley smaller than that of P. heterodontus (AMNH (UCMP 29608) and most of the Milk Creek 9432), and the lateral plate is also less inflated specimens also lack the first two upper incisors. in P. coartatus. The distance between the cen- However, specimens (F:AM 73296 and 73377) ter of the external auditory meatus and the from Milk Creek have shallow alveoli for 12, bottom of the bulla measures 44.8 mm. in P. and F:AM 73293 has the root of I1 preserved. heterodontus (AMNH 9432) and 34.0 mm. in Most specimens, however, lack any indications P. coartatus (UALP 3632). The latter is a rela- of the presence of these teeth. When present, tively small individual of P. coartatus, the these teeth in the Milk Creek sample were di- larger specimens being too crushed for accurate minutive and usually lost early during the life measurements. In both species the external au- of the individual. Evidence is insufficient to ditory meatus is nearly circular without a prom- determine whether these are deciduous or per- inent outward projecting lip. The base of the manent incisors, but in F:AM 72393, one of medial plate of the bulla is slightly wider in P. these incisors was present in an old individual. heterodontus than in P. coartatus. In both P. Earlier occurring Protolabis specimens occa- heterodontus and P. coartatus the anterior edge sionally exhibit reduction and loss of the ante- of the bulla projects slightly anterior to the rior incisors, as can be seen in a referred basioccipital-basisphenoid suture. Anterior to specimen (F:AM 32880) of P. heterodontus the bulla and lateral to the basisphenoid is a from northeastern Colorado. In this specimen pair of mesial and lateral pterygoid processes, there is a diminutive 12 but no I1. separated by a deep valley. The lateral process 13 and C of P. coartatus are robust, re- is prominent and pyramidal in shape. The inter- curved, and subcircular to circular in cross sec- pterygoidal valley runs anteriorly into the fora- tion. They are indistinguishable from those of men ovale. P. heterodontus and P. barstowensis. The ca- Crushing and erosion have obscured the nine is slightly larger than 13 and separated morphological details of the orbital region in P. from it by a diastema varying in length from coartatus but those of the palate are known in 4.2 mm. to 12.0 mm. In P. heterodontus 13-C detail. The anterior border of the internal nares diastema varies in length from 8.1 mm. to 12.7 is V-shaped and lies behind the posterior border mm. In the Milk Creek sample the muzzle is of M3, with the distance varying up to 1.0 longer and the length of C-I1 diastema varies mm. Lateral to the tip of the "V" are two from 15.4 mm. to 26.0 mm. small, variably expressed tubercles. The posi- P' in P. coartatus is small and blunt with tion of the anterior palatine foramen is variable, two roots. The roots vary from widely flared in generally being between the M's, but is also the referred specimen from Fish Lake Valley sometimes between P4 and MI (UCMP 29602), (UCMP 29608), to closely appressed in the or between the anterior moieties of the M2s type, and even fused in one specimen from (UALP 9022). In the specimens measured the Milk Creek (UALP 9022). In the last specimen width of the palate between the posterior lobes P' is entirely unlike the large caniniform P' of of M3 varies from 45.2 mm. to 55.3 mm. The Procamelus. The Milk Creek sample is signifi- m0

ca 3

0 "0

0X

0

-0 e(

. ()

N .

m

394 0

* - -o

.= . CZ

L.n

* ~Q

0 00

30

ti >

o ^0 4g0 oCZ 4-0.)C

00 a)o 00s

U O,0 a)

395 NO N -00 00 - o oo <00 oIIi"ocn 'I -4 - - rIN

.. N

N00 ON~ w

11

4 .6 I I enD m en Ol

"It t Ne-Nnr N 00 _f e. N _ _ ~N_ 06. 0 W) 4 oR4Io6 oo r rNi) ON

00 (ON n 0 e lt N C14 ON It 00oo - t- 00 - - 0 - N cN n , 90N CD N cNW). . e t e N - O - _-N mN t.~~08 £64enoent -_ _; -C6-

N Lfun r- 00 N It O 00C1 - 0 )OOW000 0 4 00 r C~oR cO6o e;C; 16 1O66 4OR nj1 e - t ON O0Cr r cn n C - - 00a 0AP m 00 O f OIN N)r N - £ rl N 0 0 ol ci.0NV~ tt t~ - - 9~- l-cic ,~ ?r ~~ C 6\ 6 W e 0 en t-oc6 el; oR.doRo~6 ent )I ~trir~ 06 oo - C ON ON~tt- 't - " en cn Ci

z ,t t t oo00 oX 0COC rcu

0)~~~~~~~~~~~~~~~) 2 3~.^ E E = = c

u)0)E E E E E E _ q -3 E E T E

396 '-2 btr- r - o tn x o o m c r- m w r o 00o 0 0 0 0 N 00 00' N 0r *l<00

zo

o£ N N =; 0t0 N N tn NOr 00 <00-D--C1C1eN

0.1: -6 :~~~~~~~~~~~~~~00 0~~~~~~~~~~~0so ^ £ o m0-

00_00_ 00 _ u I ~~ ~~~~~~~~~~~~0)00

t ~>~ ~~ W^<£ 00 cr- rE oN ri ^ ...... NO' .^ . ~. 0

0)rN VI-o cr t ONc't} I \ r 000000r C NNr a0 00' I 0 om o w £mtbW w - ^ ~~~VtfFt~~ oo^110 ON ONO N Cl :oo =C

11 oN°0o-O o 0S- \C o r U00 0-ONcrm 0 m Voo_

H N > o _ coX _ ff (N L C i l _ e'4 00 WON oN O--Ot'C- 0 'C l C H Cl~~~nO .i 0 ~1 6 C7\ 16VClN

0 0

- - N £ ^ b £IC xo o. s - - ON - teNb - . s} 't ?

--Cl-r~Cl NO v \.O lz CN ,O 1. V' V s

"- -

0 ~~~~~~~~~~~~~~~0)

OoLLLLLLESESiSLES~~00O3O3O 3- - - T S U O

397 398 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161 cant in showing all graduations of PI root from None of the Milk Creek specimens possess the widely flared to fused. Individuals with com- inter-columnar tubercle on MI that is present on pletely fused roots are rare, and only one was the referred specimen from Fish Lake Valley found in which a cross section of the tooth (UCMP 29612). Because no other morphologi- failed to reveal evidence of a double root. cal differences were seen in the Fish Lake Val- The length of the premolar row and the ley specimens, we consider development of this individual lengths of the premolars are highly tubercle variable and of no taxonomic signifi- variable in P. coartatus. The sample from the cance. M2 and M3 both possess a strong para- stratigraphically highest Milk Creek Quarry has style and mesostyle; M3 also has a small relatively larger premolars than the sample metasytle. A weak rib is present on the para- from the stratigraphically intermediate Deep cone and metacone of M2 and M3, and is Spring Quarry. The premolars of P. coartatus faintly developed on the little worn M1. In P. from the stratigraphically highest level (Milk heterodontus the styles are sharper and the ribs Creek Quarry) are similar in size to those of P. slightly more prominent than in P. coartatus. heterodontus from northeastern Colorado. In P. M3 of P. coartatus from the stratigraphically coartatus however, the morphology is simpler. highest Milk Creek Quarry has a relatively P2 is double rooted with a high, central, basally more anteroposteriorly elongate second lobe swollen paracone separated by a buccal groove than that of both P. coartatus from the strat- from a buccally inflected metastyle. In P. het- igraphically lower Deep Spring Quarry and P. erodontus the metastyle is slightly more devel- heterodontus from northeastern Colorado. Thus oped. The parastyle is absent in P. coartatus, M3 in the Milk Creek Quarry sample is larger whereas P. heterodontus retains a vestigial par- than that in the other two samples. The other astyle. A weak lingual cingulum runs along the molars in the Milk Creek Quarry sample also base of P2 in P. coartatus. average larger than those of the Deep Spring Like p2, P3 of P. coartatus has a bulbous Quarry sample. paracone that is separated by a groove from a LOWER DENTITION: The lower incisors are buccally inflected metastyle. Again the para- spatulate. The canine of the type of P. coar- style is absent; in contrast P3 of P. heterodon- tatus is conical and pointed. Canines from the tus has a prominent, buccally inflected Milk Creek sample exhibit various degrees of parastyle. Consequently, P3 of P. heterodontus compression, and no morphological features has a shallow anterior groove formed by the distinguish them from the canines of P. hetero- inflection of the parastyle, which fails to extend dontus. to the base of the tooth. Mesially, P3 of P. P1 in the Milk Creek sample is double coartatus sometimes possesses a basal swelling rooted, with the roots closely appressed or even which is continuous with two crests running fused. In P. heterodontus the roots are widely from the anterior and posterior edges of the separated. P1 of P. heterodontus is anteroposte- tooth. These crests do not join to form a fos- riorly elongate with a prominent, centrally lo- sette. Although stronger in P. heterodontus, cated cusp (protoconid) from which anterior these crests also do not join to form a fossette. and posterior crests descend. In P. coartatus Pi P4 is single lobed and very similar to that of is short, and in individuals with appressed roots P. heterodontus. It possesses a parastyle and a is peglike. In a few cases P1 is suppressed weakly developed metastyle. The paracone (unerupted), or even absent. One individual forms the major buccal cusp. The internal cres- from Milk Creek (F:AM 73265) has Pi present cent consists of the protocone and weakly de- and erupted on one side and absent on the veloped hypocone. opposite side. P1 in P. coartatus is above the The molars of P. coartatus and P. mental foramen, as in P. heterodontus. heterodontus are similar. M1 has a strong As already noted in the literature, P2 in P. mesostyle that extends to the base. The para- coartatus is extremely small and simple. When style is strong in young animals and disappears present, P2 is double rooted with a single cen- with wear. The metastyle is weakly developed. tral cusp. The roots are closely appressed and 1978 HONEY AND TAYLOR: PROTOLABIDINI 399

sometimes fused. In contrast, P2 of P. hetero- The molars of P. heterodontus also have varia- dontus is low and elongate with a central, bly developed parastylids and rare protostylids swollen protoconid, a small posterior hypo- that occasionally form buttresses. conid, and a small anterolingually inflected par- MANDIBLE: In the Milk Creek sample the aconid. In P. coartatus this tooth is absent in length of the mandible is variable and ranges more than half the individuals. One specimen from 270.2 mm. to 312.7 mm. The angle (F:AM 73376) has P2 present in the right ramus formed at the junction of the symphysis with but absent in the left ramus. Absence of P2 the body of the horizontal ramus varies from cannot be correlated with absence or suppres- shallow to more upright. The contour of the sion of P,. lower border of the symphysis is generally con- P3 of P. coartatus is shorter and simpler cave; however, in one large specimen (F:AM than that of P. heterodontus. Measurements 73267) from Milk Creek Quarry, it is slightly given by Patton (1969, p. 153) for the length convex. In most specimens the symphysis flares of P3 in P. notiochorinos indicate a very small laterally in the region of the incisors and ca- tooth. This is, however, an approximate meas- nines. In P. heterodontus the angle made at the urement that was taken at the alveolus of the junction of the symphysis with the horizontal broken tooth. Because crown length decreases body of the ramus is variable, but in most toward the root, the unworn crown length of P3 specimens the symphysis turns sharply upward. of P. notiochorinos is probably close to that of Immediately behind the symphysis in P. het- the type of P. coartatus. P3 of P. coartatus erodontus, a pronounced concavity is present consists of a large, central protoconid, a lin- along the ventral surface of the horizontal gually inflected paraconid, and a labially ramus. Furthermore, the depth of the jaw at the swollen hypoconulid that is slightly narrower posterior end of the symphysis in P. heterodon- than the protoconid. A few specimens exhibit a tus is generally greater than in P. coartatus. In very small entoconid that unites with the hy- most P. heterodontus mandibles a tuberosity is poconulid to form a tiny fossetid that disap- present on the extreme anterior end of each pears quickly with wear. In P. heterodontus the ramus. These tuberosities merge anteriorly to paraconid is larger and more strongly inflected. form a rugose ridge, probably for the insertion Furthermore, P3 of P. heterodontus starts to of the digastric muscle. This rugose ridge is wear early in the life of the individual, whereas only weakly developed in P. coartatus. in P. coartatus this tooth is often little worn in The mental foramen is situated above the old individuals. posterior border of the symphysis and nearer to P4 of P. coartatus is generally smaller than the ventral border of the ramus in both P. that of P. heterodontus. It is similar to P3, with coartatus and P. heterodontus. In P. coartatus a central protoconid and a lingually inflected P, generally erupts directly above the mental paraconid. The hypoconid and entoconid unite foramen; in P. heterodontus its position is more to form a fossettid. In P. heterodontus the par- variable with P, situated behind the mental fo- aconid is larger and more strongly inflected, ramen in more than half of the jaws. As seen and the lingual groove between the paraconid in the type of P. coartatus and the referred and protoconid is relatively deeper. specimen (UTBEG 31132-332) from Lapara In P. coartatus the molars are smaller than Creek, a second, small mental foramen some- in P. heterodontus. Morphologically, there is times occurs low on the body of the ramus little to distinguish the molars of the two spe- below the posterior moiety of M,. Immediately cies. The type of P. coartatus has a pro- behind the symphysis the ventral border of the nounced parastylid on M3, which is variably ramus is concave. Beneath P4 the jaw straight- developed in the Milk Creek specimens. Some ens and becomes convex. In P. coartatus the of the larger specimens from the Milk Creek depth of the jaw between P, and P2 is less than and Deep Spring quarries also have a rather in P. heterodontus, and the post-P, diastema is well-developed M3 protostylid that unites with usually longer than in P. heterodontus. The the parastylid to form a small anterior buttress. dorsal border of the ramus anterior to the cheek X X cr F o o m ^~~~~~Mt m (Ocrt (ON -o0 M C' en m CII I I6 O.60r& & cr 0N z

cqB~~~~~~~~~~~~~~~~~~~~~~~~~~~~c- r-.0 q.- > gI _I 0~~000Nooo^Ntmxen t o o nr -

94> 00

0 0 4 6 C o Xo<002NIo I--o X _ 4 oo Qo

Cu:;;.X~~ 7£ ON tnttmf U E° 2~00NX i bi - °~i t NI *

cn 00 tn r-- tn~~~~~~~~~~~~~~~~~~~~~~0 o

r'0 o v n w - ON In IXI-m-I m tnI I I0 0

- "Xit VQ r T - - _ _ _Z - t r-X - @ ~~ ~ ~ e I7Cs ON 00- tn tn C" O Nm w

00 X 00- UXV- m 0£ £ n m O V 00

t~~X £ £ m o > t O - - - C1 C\- tn o b - o ON- C_r- C'm1so" o -£t- t - m mo - r-- a-o t1 00 000 0 0 r " t C 0~0 O0 Lm-,< tn M z-n n \)£x^£-Fo00bVhotFOt *-~~ ~ ~ ~ ~ ~ m6- en 06 C56 6 6 6 ci _: ci '- C0 W)

00 e r-- o-o of oo- o N -_ -u^ e_w5> - 0O rOC O O* 0 0 ~ *Zs gL1&°w~~~~~~~~~~~~~0N-c~tn C) 00 t ))-) 0)O C! 660-otOR

0)~~~~1 zVN b t mm £ O 00 0 ~Z --00r------0t00N- - -0d0r-1r------°-

z 00 40) 4.=00;- 0 .-- 0 7m 00) t afvycB4>7 Ea 5 E 7 e ' 2 g(L) o lt E--'7< X=. 4002Q~~~~~~~~~- 1l

3^

>

~r~

IWO1-4

401 402 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161 teeth is very thin and sharp in P. heterodontus, Protolabis specimens. Like the lateral flare, the whereas in P. coartatus it is thicker and more mesial rugosity reaches its maximum shelflike rounded. Anterior to the cheek teeth, the man- development in the Milk Creek sample of P. dible of P. coartatus narrows more sharply coartatus. than that of P. heterodontus, to coincide with On the lateral surface of the ascending the extreme rostral constriction above. ramus is a variably defined, shallow temporal The strong lateral flare of the angle (fig. 9) fossa; it extends slightly forward and ventrally is the most distinctive feature of the mandibular from the level of the condyle. The articular ramus of P. coartatus. This flare extends up- condyle is triangular and convex laterally and ward from the posteroventral border of the concave mesially. The coronoid process is low angle to the relatively small angular process. and extends only slightly above the level of the The posterior rim of the angle in P. coartatus condyle. is rougher and thicker than that of P. hetero- DECIDUOUS DENTITION: The deciduous den- dontus. See functional morphology (p. 404). A tition of P. coartatus is well represented by somewhat similar outward flaring of the angle many specimens from Milk Creek. The termi- is present in the type jaw of Nothotylopus nology employed here is that developed by camptognathus (Patton, 1969, p. 160, figs. 17 Loring and Wood (1969). and 18). A slight angular flare is seen on a In size and morphology, the deciduous up- mandibular ramus of P. heterodontus (F:AM per dentition of P. coartatus corresponds 32881), and it is also present in varying de- closely to that of ?Protolabis (UCMP 32309) grees in other samples of Protolabis in the from Big Spring Canyon described by Loring F:AM collection. The Milk Creek sample nev- and Wood (1969, p. 1204). No upper decid- ertheless exhibits the most extreme develop- uous incisors or canines are recognized in the ment of the outward flare of the angle. The Milk Creek sample. A little worn dP2 (F:AM angle of the type jaw of P. coartatus is broken, 73333) consists of a short anterior crest, a cen- and consequently lacks any indication of the tral primary cusp, and a posterior cusp. The presence of this unusual mandibular feature. primary cusp is labially swollen, and the pos- Patton (1969, p. 160) considered this angular terior cusp labially inflected. The two cusps flare a distinctive feature of P. notiochorinos, join with wear. A lingual cingulum is present but the flare of the Lapara Creek specimen is with prominent anterior and posterior shelves. slightly less than that of the Milk Creek speci- The cingulum disappears with wear. mens. This may be due in part to the age of the DP3 is bilobed with anterior and posterior individual from Lapara Creek, for the outward crescents (protocone and hypocone). The ante- flare is strongest in the older individuals of P. rior lobe is much narrower than the posterior coartatus from Milk Creek. One Milk Creek lobe and has a prominent anterior crest (para- specimen (UALP 9005) has M2 erupting and style) inflected buccally. Little worn specimens shows no lateral flare on the mandible. have a cingulum on the lingual side of the A strongly developed mesial inflection on anterior crest, which may extend posteriorly to the posteroventral border of the horizontal the lingual groove between the anterior crest ramus (fig. 9) is another characteristic feature and crescent. On specimens referred to P. of the mandible of P. coartatus. This inflection coartatus a weak posterior cingulum is some- is just anterior to the lateral flare of the angle, times present on the posterior crescent. No and is associated with the pronounced develop- specimens allocated to P. coartatus have an ment of the posteroventral shelf on the mesial anterior basal cingulum on the posterior cres- surface of the horizontal ramus. This shelf runs cent. With wear the anterior and posterior cres- anterodorsally for a short distance, resulting in cents join and their fosettes are obliterated. The a thickened, rugose surface for the attachment mesostyle is well developed and labial ribs are of the internal pterygoid muscle. The ptery- present. goidal rugosity is weakly developed in P. het- DP4 has a strong parastyle and mesostyle erodontus, and is variably developed in other and a weak metastyle. The rib on the paracone 1978 HONEY AND TAYLOR: PROTOLABIDINI 403

TABLE s Measurements (in Millimeters) and Statistics of the deciduous Dentition of Protolabis coartatus from Milk Creek, Arizona N OR X S V dP2 length 6 6.7- 8.4 7.65 7.610 0.238 dp2 width 6 3.0- 4.2 3.60 0.434 12.044 dP3 length 9 12.6-16.4 15.04 1.382 9.192 dP3 posterior lobe width 9 9.4-11.9 10.71 0.807 7.533 dP3 anterior lobe width 7 7.0- 7.8 7.34 0.282 3.841 dP4 length 12 12.5-20.0 17.20 2.469 14.359 dP4 anterior lobe width 10 10.7-12.8 11.33 0.646 5.706 dP4 posterior lobe width 11 8.8-13.4 11.24 1.275 11.342 dP2-dP4 length 6 32.9-45.5 42.83 4.909 11.462 dP2 length 5 5.0- 7.2 6.14 0.948 15.433 dP3 length 13 7.2-11.4 9.38 1.400 14.923 dP3 width 13 3.8- 5.4 4.37 0.405 9.268 dP4 length 21 18.2-22.8 20.40 1.315 6.445 dP4 width 20 7.0-10.1 8.82 0.930 10.530

is stronger than that on the metacone, and a groove. Posteriorly, the tooth is wide in all variably developed anterior cingulum is present specimens, due to the presence of both an ento- on the protocone. All specimens have a conid and a hypoconid. In early wear a small cingulum high on the posterior wall of the hy- fossettid is present between these cusps. pocone, running lingually and ventrally from DP4 is long and has a flat lingual wall. A the metastyle. DP4 is narrower in P. coartatus few P. coartatus specimens exhibit a relatively than in ?Protolabis from Big Spring Canyon well-developed metastylid which extends to (figured by Loring and Wood, 1969, fig. 4A). near the base of the tooth. Both the ribs and The lower deciduous dentition of P. coar- the entostylid are only weakly developed on the tatus also corresponds closely with that of lingual wall. As in the other deciduous teeth, ?Protolabis (UCMP 32309) from Big Spring no anterior or posterior cingula are present. Canyon (Loring and Wood, 1969, fig. 4B). LIMBS: In size and morphology, the limbs of The deciduous incisors resemble their perma- P. coartatus are indistinguishable from those of nent replacements, becoming increasingly spat- P. heterodontus. As in all Protolabis, the meta- ulate from d1 to dl3. The deciduous canine is podials are short and stout (fig. 8). No associ- very small and blunt, with a laterally com- ated fore and hind limbs of P. coartatus are pressed crown. It is much smaller than dl3. As known, but it is reasonable to assume that as in in all camelids no evidence is seen for replace- other Protolabis species, the length of the met- ment of P, or dP1. atarsus exceeds the length of the metacarpus. DP2 is diminutive with either a single root or This hypothesis is supported by the mean two very closely appressed roots. It is some- lengths of the unassociated metacarpi and meta- what shorter than the same tooth in the Big tarsi (table 6). The type of P. coartatus is a Spring Canyon specimens (UCMP 32309). A slightly larger individual than any in the Milk crest, ending in a small, lingually inflected Creek collection. stylid is anterior to a central primary cusp. DISCUSSION: The type of Protolabis coar- Another crest with only a slight swelling is tatus comes from Stewart Valley, Nevada. situated posterior to the primary cusp. Originally described as Procamelus near gra- DP3 has a prominent central cusp. Ante- cilis by Merriam (1916, p. 189), it was named riorly, a lingually inflected anterior cusp is sep- Procamelus coartatus by Stirton (1929, p. arated from the primary cusp by a lingual 295). At that time the type material was aug- 404 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161 mented by the description of a fragmentary ros- by the Frick Laboratory of the American Mu- trum and palate from Fish Lake Valley. Both seum of Natural History. A somewhat smaller authors commented upon the marked reduction collection from the same area, initially col- of the anterior lower premolars and the extreme lected under the direction of John Lance and rostral constriction; the latter was considered later under the direction of Everett Lindsay, diagnostic of the species by Stirton. In the exists at the University of Arizona. Material in same paper, Stirton also described a palate these two collections has added tremendously from Fish Lake Valley, which he allocated to to our knowledge of P. coartatus, and in fact Procamelus cf. gracilis, but which he later re- contributed significantly to the reevaluation of ferred to P. coartatus (Stirton, personal com- the genus Protolabis. The above description of mun., in Henshaw, 1940, p. 20). Stirton's P. coartatus is based on an examination of the reference of this material is here considered Milk Creek specimens, and of the type and correct. Henshaw (1940, p. 20) described and referred specimens from Nevada. allocated some material from the Avawatz Mountains to Procamelus coartatus. None of FUNCTIONAL MORPHOLOGY this material is here considered definitely re- ferable to that species, and some of it is tenta- Comparison with figures in Turnbull, 1970 tively referred to Michenia. The postcranial (figs. 14-16) reveals that the lateral flare on the material from Avawatz, originally assigned by dentary of P. coartatus occurs in the insertion Henshaw to Procamelus coartatus is similar in area of the superficial masseter in other ungul- size and morphology to limbs of that species. ates. The origin area of the superficial masse- The dental material, however, is either indeter- ter, the zygomatic ridge (=facial crest), is minate or tentatively referred to Michenia, and strongly developed in P. coartatus. The deep is discussed further on page 407. layer of the masseter originates on the zygoma- Patton (1969, p. 152) described Protolabis tic arch posterior to the origin of the superficial notiochorinos in the Lapara Creek fauna. An layer. As noted in the description of P. coar- examination of his illustrations reveals that in tatus (p. 391), the zygomatic arch is no more size and morphology, the type is practically robust than on P. heterodontus. The deep layer indistinguishable from the type of P. coartatus. inserts on the dentary dorsal to the insertion of Morphologic and morphometric differences the superficial masseter. These facts suggest from P. coartatus are considered due to preser- that the lateral flare on the dentary of P. coar- vation, and consequently P. notiochorinos is tatus is associated primarily with greater devel- synonymized with P. coartatus. The synonymy opment of the superficial masseter. Similarly, is based on comparison with the Arizona mate- the pronounced mesial flare on the dentary oc- rial discussed previously. curs at the insertion of the internal pterygoid Since 1956 a large, stratigraphically well- muscle, and appears to be due to pronounced documented collection of Protolabis coartatus development of that muscle. has been collected from Milk Creek, Arizona Because the superficial masseter and internal

TABLE 6 Measurements (in Millimeters) and Statistics of Referred Metapodials of Protolabis coartatus from Milk Creek, Arizona N OR X S V Metacarpal length 20 191.9-242.4 224.96 12.641 5.619 Metacarpal proximal width 29 35.0- 41.1 38.19 1.749 4.580 Metacarpal narrowest transverse shaft width 15 19.9- 24.3 21.60 1.090 5.060 Metatarsal length 13 215.2-247.6 232.35 10.780 4.640 Metatarsal proximal width 11 26.9- 32.6 29.85 1.655 5.540 Metatarsal narrowest transverse shaft width 14 15.3- 19.2 17.53 1.132 6.485 1978 HONEY AND TAYLOR: PROTOLABIDINI 405 pterygoid muscles are important in jaw closing cles in mastication, and decreased emphasis on and mastication, their development in P. coar- the temporalis. Tumbull (1970, p. 256) pointed tatus is in response to selection for more force out that in ungulates the masseter is the most in the mastication of food. Ostrom (1964, p. important muscle of mastication, generally fol- 22) noted that the effective force at the denti- lowed by the pterygoid and last by the tem- tion can be increased by "1) increasing the poralis. He showed (1970, fig. 36) that as the magnitude of the applied force [enlarging the size of the masseter increases, the size of the adductors] or 2) lengthening the moment arm temporalis decreases. These facts are important [of the jaw] by shifting tho point of force appli- when considering P. coartatus relative to Pro- cation [insertion of the adductors] away from camelus. In the latter, the angle of the dentary the fulcrum [glenoid condyles]". In connection is not laterally or mesially inflected, the coro- with method 2) DeMar and Barghusen (1971, noid process is longer, and the cranium is p. 634, fig. 13) show that as the length of the larger, than in P. coartatus. Thus during the moment arm (from the condyle to the point of evolution of Protolabis there was selection for insertion of the adductors) is increased in an enlargement of the superficial masseter and in- anteroventral direction, the line of force appli- ternal pterygoid muscles relative to the tem- cation (from the insertion to the origin of. the poralis. This did not occur in Procamelus. adductors) must be rotated anteriorly on the We believe that the constricted rostrum pos- skull. This anterior rotation is necessary to terior to the canine can best be explained by keep the line of force perpendicular to the mo- assuming a highly developed buccinator and ment arm. The superficial masseter has the maxillo-nasolabialis muscle. Webb (1965, p. 6) most anterior origin of the masseteric muscles, discussed the importance of the upper lips in and operates with an anterodorsal line of ac- camelids, which function both as a cropping tion, thus becoming important as the moment pad and as a prehensile organ. He concluded arm lengthens (see DeMar and Barghusen, that the lip muscles were probably highly de- 1971, fig. 13). However, we believe that its veloped in some fossil camelids. The buccina- development (and also that of the internal tor and maxillo-naso-labialis muscles are pterygoid) on P. coartatus does not reflect an important in manipulation of the lips and increase in the length of the moment arm, for cheeks. The latter muscle originates at the junc- the following reasons. First, there is no in- tion of the lacrimal, malar, and maxillary crease in the depth of the angle below the bones. There are two divisions to this muscle. glenoid condyle (corresponding to the moment According to Webb (1965, p. 6), "The ventral arm) in P. coartatus relative to P. heterodon- division reaches the angle of the mouth and the tus. Second, as discussed on page 385, in Pro- anterior end of the diastema. The dorsal divi- tolabis M3 tend to be longer relative to M12 sion inserts on the upper lip and, after joining and the premolars are more reduced, than in fibers from the other side, on the tip of the Procamelus. The reduction of the anterior nose." Thus a well-developed maxillo-naso-la- cheek teeth and the expansion of M3 in Pro- bialis muscle has an insertion in the area of the tolabis suggests that the effective force vectors rostral constriction of P. coartatus. The buc- in mastication were more posterior than in Pro- cinator originates on the alveolar border of the camelus. This is contradictory to the supposi- maxillary and, according to Sisson and Gross- tion that there was an anterior rotation of the man (1975, p. 796), retracts the angle of the effective force vectors on P. coartatus. Thus mouth and flattens the cheeks. The buccinator we suggest that the increase in masticating results in a distinct fossa on the alveolar border power in P. coartatus was accomplished only of the maxillary in certain fossil equids and the by increasing the mass of the adductors. kangaroo (Gregory, 1920, fig. 1, pl. 18). The The small size of the cranium and coronoid rostral constriction of P. coartatus is best process is correlated with the pronounced lat- termed a buccinator fossa, although a strong eral and mesial flares (fig. 9) on the mandibular maxillo-naso-labialis muscle also probably con- angle of P. coartatus. This reflects increased tributed to its development. emphasis on the masseter and pterygoid mus- Thus, we interpret the derived cranial fea- 406 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161 tures of P. coartatus as a reflection of strongly was the only described species of the genus, developed labial, buccal, and masticatory mus- the characters of Michenia were necessarily cles. The development of the jaw musculature those of M. agatensis. These authors noted, and the relatively high-crowned dentition sug- though, that Michenia persists through the Clar- gests that P. coartatus was eating tough foods, endonian. possibly grass or woody shrubs. In the Sahara, The transfer of coartatus from Procamelus Gauthier-Pilters (1974, p. 543) noted that to Protolabis and the transfer of exilis from "camels can thrive on the hardest, driest and Oxydactylus to Michenia plus the description in thorniest plants," and in fact often prefer the this paper of the derived taxon, Michenia thorny woody-acacia. Protolabis coartatus may yavapaiensis, necessitate a revision of the origi- have had a similarly rigorous diet. As discussed nal diagnosis of Michenia. The following earlier, (p. 373), Deep Springs Quarry is asso- points are pertinent: 1) Michenia was char- ciated stratigraphically with aeolian dune de- acterized by Frick and Taylor, 1971 as having posits. These deposits suggest the possibility of a narrow muzzle, but it is not always narrower a locally arid paleoenvironment in which tough, than that of Protolabis, as shown by the hardy plants thrived. Too, P. coartatus may muzzle of P. coartatus being more constricted have been eating grass; modern camels are than that of M. agatensis; 2) later occurring known to eat grass and the high-crowned equid specimens of Michenia sometimes have rela- Pliohippus is occasionally found with P. coar- tively smaller and simpler P2-P4 than those of tatus in these sediments. Protolabis; 3) the molars of later occurring Michenia specimens are taller crowned with MICHENIA FRICK AND TAYLOR, 1971 less distinct ribs and stylids and are not always easily distinguishable from those of Protolabis; Procamelus coartatus Henshaw, 1940 (in part) pp. 20-21. pl. 3, figs. 1-3, pl. 4, fig. 3. 4) the metapodials of M. exilis are unfused, in contrast with those of M. agatensis, upon TYPE SPECIES: Michenia agatensis Frick and which the original description of Michenia was Taylor, 1971. based; and 5) the metapodials of Michenia, INCLUDED SPECIES: Michenia exilis (Mat- though always slenderer than those of Pro- thew, 1960); Michenia agatensis Frick and tolabis, are more solidly fused and relatively Taylor, 1971; Michenia yavapaiensis, new spe- stouter in the Clarendonian than those of M. cies. agatensis. KNOWN DISTRIBUTION. From late Morphological changes manifested in Arikareean of Texas and early Hemingfordian Michenia parallel those of Protolabis, and the to Clarendonian in the western United States. character transformation from primitive to de- REVISED DIAGNOSIS: Michenia differs from rived in both genera is in general accord with Protolabis by smaller size; shorter braincase; the known temporal sequence from the Hem- weaker and more incisiform I1-3 in primitive ingfordian to the Clarendonian. Derived speci- species; weaker and more incisiform upper ca- mens of Michenia and Protolabis have nine; smaller lower canine; less anteroposte- moderately hypsodont molars with weak ribs riorly elongate M3; shallower symphysis, and styles. In the later occurring specimens of conspicuously unlike the deeper symphysis with Michenia M's lengthen to the point that they more acute angle in Protolabis; shallower hori- overlap the shorter M3s of Protolabis. More- zontal ramus; and relatively slender meta- over, in the most derived species of both gen- podials. era (M. yavapaiensis and P. coartatus) 11-2 are Differs from Tanymykter in more constricted absent, P2-4 are reduced in size and simplified rostrum and deeper buccinator fossa; relatively with the P2 frequently lost. Furthermore, in shorter braincase; weaker 13 and C/C; shallower both genera the metapodials become pro- symphysis; and slender metapodials. gressively shorter with the length of the meta- DISCUSSION: Frick and Taylor (1971, pp. carpus significantly less than that of the 5-6) diagnosed Michenia on the basis of the metatarsus. However, most of these mor- Hemingfordian species, M. agatensis. As this phological changes are not restricted to the Pro- 1978 HONEY AND TAYLOR: PROTOLABIDINI 407 tolabidini but are convergent to a greater or alveolus, C, roots of P12 p3-49 M13; partial lesser degree with other camelids. scapula, distal end of humerus; radius; metacar- Some of the material from the Avawatz pus; calcaneum; metatarsus; and proximal pha- Mountains (CIT 1989, 2316, 2318, and 2321) langes from near Porcupine Butte, listed by which Henshaw (1940) allocated to Procamelus Macdonald (1963, p. 23) as "AMNH 'Rosebud' coartatus, is here considered referable to 5," Rosebud Formation (Hemingfordian), Michenia. Specifically, Henshaw's measure- Shannon County, South Dakota. ments for the teeth, especially the length for KNOWN DISTRIBUTION: Late Arikareean of P2-M3 of palate (CIT 1989) lie closer to the Texas and early Hemingfordian of South Da- mean values of the same measurements of kota. Michenia yavapaiensis. The measurement for REVISED DIAGNOSIS: Differs from Michenia the narrowest rostral width is less than in agatensis in having P1 more elongate with more Michenia yavapaiensis and is similar to that of widely separated roots; molars anteroposteriorly P. coartatus. Other specimens of Michenia in shorter and brachyodont; upper molars with the F:AM collection, however, have rostral stronger buccal ribs and styles; metapodials un- widths as narrow as that of CIT 1989. One fused; and first phalanx shorter with distal artic- symphysis (CIT 2321) with a small canine is ular surface more grooved and less anteriorly very shallow dorsoventrally, and is inseparable produced. from some symphysis allocated to Michenia Differs from Michenia yavapaiensis in the yavapaiensis. It is unknown if P1 is absent or above characters plus 11-2 present; Pi2 present suppressed in CIT 2321. M3 (CIT 2318) is and erupted in adults; and molar metastylids relatively short, falling near the mean of the M3 stronger. length in Michenia yavapaiensis. Although ex- DESCRIPTION AND COMPARISON: This species treme wear has undoubtedly shortened the was previously described briefly, measured, teeth, the lengths of the extremely worn P3M2 and figured by Matthew and Macdonald (1960, of CIT 2316 measures nearest to the mean of pp. 2-6). Brief comparisons were made chiefly these teeth in Michenia yavapaiensis. The shal- with Poebrotherium. Because we have included low symphysis constitutes the strongest evi- this species within Michenia, we shall supply a dence for allocating the Avawatz Mountains supplementary description of M. exilis, and specimens to Michenia. Because rostrum (CIT make comparisons with M. agatensis, the gen- 1989) is narrower than that of F:AM 73287 otypic species. We have supplied some dental from Milk Creek, plus the fact that our sample measurements not provided by Matthew and is inadequate to determine the variation in ros- Macdonald (1960, p. 5); the reader is referred tral width in Michenia yavapaiensis, we refrain to that publication for limb measurements. from allocating the Avawatz Mountain speci- SKULL: Viewed ventrally, the muzzle of M. mens to this species. exilis is long and narrow as in M. agatensis, The description of the other samples of and does not widen noticeably anterior to the Michenia must await future study. The material canine. The narrowest palatal width posterior to of Michenia yavapaiensis described in this PI is 14.4 mm. In M. agatensis the palate is paper was collected by the American Museum slightly crushed in this area, but it was appar- of Natural History and the University of Ari- ently similar in width. The lateral edges of the zona, and is housed at these institutions. muzzle in both species form rather sharp ridges, due to the extreme lateral narrowing of Michenia exilis (Matthew, 1960) the maxillaries in the area of the buccinator Figure 10; Tables 7, 8 fossa. Because the lateral surface of the rostrum of Oxydactylus exilis Matthew and Macdonald, 1960, p. the type of M. agatensis is not preserved, a 2, figs. 1-4. ?Michenia australis Stevens, 1977, p. 48, fig. 15. comparison is not possible. The rostrum of M. exilis is comparatively deep, and the nasals are TYPE: AMNH 12997, partial skull with I3 comparatively shorter than in Poebrotherium. alveolus, C, PI, p2-4, M1-3, mandible with 13 Consequently, the premaxillaries of M. exilis 408 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 16 1 are elongate and anteroposteriorly narrow, and central paracone, separated by a deep buccal slope posteriorly at an angle of approximately sulcus from a buccally inflated prominent para- 35 degrees to the roof of the palate. The ante- style. A shallower posterior sulcus separates the rior edge of the ascending wing of the premax- metacone from the paracone. P4 lingual illa is straight, becoming slightly concave near cingulum is also more strongly developed on the palatal surface. Contact between the dorsal M. exilis, and the anterior and posterior parts border of the premaxilla and the nasals is along are separated by a narrow groove, in contrast to a suture 25.0 mm. in length. the wider valley in M. agatensis. The bone on the lateral surfaces of the max- P4 of M. exilis differs chiefly from that of illaries is largely destroyed; however, the un- M. agatensis in possessing a more angular mes- derlying matrix preserves the essential features. ial crescent, consisting chiefly of the protocone. Posteriorly, a deep, circular maxillary fossa, There is no trace of a metaconule, which on M. approximately 28.5 mm. in diameter, crowds agatensis is relatively strong. against the nasals. Anterior to the fossa and Although more highly worn, the molars of immediately beneath the nasals, the maxilla is M. exilis are obviously more brachyodont and laterally swollen. The buccinator fossa is be- less anteroposteriorly expanded than those of neath this swelling and forms an elongate arc M. agatensis. The paracone and metacone ribs extending from the canine to P2, and is deepest on M2 and M3, and apparently those also of the above P1. It is moderately developed as in M. highly worn M', are more strongly developed agatensis. Posteriorly, the maxillary root of the than in M. agatensis. The ribs, which are rela- zygomatic arc terminates above the posterior tively narrow in M. Xagatensis, are slightly lobe of M2, as in M. agatensis. Dorsally, the higher and occupy a wider area on the paracone nasals are slender and terminate midway be- and metacone of M. exilis, and the mesostyle tween C and PI. on M2 and M3 is also higher and slightly wider UPPER DENTITION: The incisors of M. exilis on M. exilis. As in M. agatensis, the parastyle are not preserved. The canine is small, blunt, is prominent and the metastyle weak. transversely flattened, and recurved postero- LOWER DENTITION: Matthew and Mac- lingually. It is approximately the same size as the canine of M. agatensis, but is situated more anteriorly. TABLE 7 P1 is laterally compressed and elongate and Measurements (in Millimeters) of the Upper Dentition of the type (AMNH 12997) of contrasts markedly with that of M. agatensis. It Michenia exilis consists of a central swollen paracone with (Matthew) elongate anterior and posterior ridges terminat- Palatal width between P2 20.7 ing in slight swelling at each end of the tooth. Palatal width narrowest part of rostrum 14.4 The roots are strong and widely flared, in con- C-P1 diastema length 18.7 trast to the closely appressed roots of M. ag- pl-2 diastema length 20.0 atensis. Upper canine length 4.7 p2-4 of M. exilis closely resemble those of PI length 8.8 M. P2 P2 length 10.2 agatensis. is elongate with a central, P2 width 4.4 basally swollen paraconid. A short anterior P3 length 10.9 ridge connects the paracone with a prominent P3 width 5.9 parastyle that is stronger than that of M. ag- P4 length 12.0 atensis. A longer posterior ridge terminates in a P4 width 9.5 slightly swollen metacone. The buccal sulci are M1 length 14.5 relatively shallow. The lingual cingulum is M1 width 14.9 more strongly developed, especially anteriorly M2 length (16.) and posteriorly, than that of M. agatensis. Me- M2 width 15.6 dially, the cingulum is very weak as in M. M3 length 17.9 agatensis. M3 width 14.7 As in M. agatensis, P3 possesses a strong M1-3 length 51.5 1978 HONEY AND TAYLOR: PROTOLABIDINI 409

FIG. 10. A-C. Michenia exilis (Matthew), type, AMNH 12997, from the Rosebud Formation (Hemingfor- dian), South Dakota. A. Metatarsus, anterior view. B. Lateral view, reversed. C. Lateral view. All X½/2.

donald (1960, pp. 2-3) noted that the canine of the molar row; due to the anteroposterior was moderately compressed and situated close expansion of the molars on M. agatensis, the behind the incisors. In these features and its premolar row length averages only 46.5 percent relatively small size and lingual curvature, the the molar row length. Extreme wear has elimi- canine compares favorably with that of M. ag- nated most of the morphological details of the atensis. molars of M. exilis. On M3 however, the lin- Only the roots of P1 are preserved; they are gual ribs are relatively stronger and wider than strong and separated, and more closely resem- on M. agatensis. ble those of the referred specimen of M. ag- MANDIBLE: The mandible of M. exilis is atensis, (F:AM 14259), than the type. The nearly identical with that of M. agatensis. As broken P2 is two rooted. P3 possesses a large, in M. agatensis the symphysis is dorsoventrally central protoconid, and a strong lingually in- shallow and forms a low angle with the hori- flected paraconid. A posterolingual valley sepa- zontal body of the ramus. The depth of the rates the buccally swollen hypoconid from the ramus between P, and P2 is slightly less than in short, posterolingually directed entoconid. The M. agatensis. The dorsal border of the ramus entoconid is slightly more strongly developed anterior to P2 forms a sharp ridge, as in M. than in M. agatensis. A minute tubercle is agatensis. Despite the more brachyodont teeth, situated on the posterobuccal margin of the hy- the depth of the ramus below M3 is the same as poconid. P4 is similar to that of M. agatensis, in M. agatensis. The condyle and angular proc- having a strongly inflected paraconid, and a ess of the mandible in M. exilis are approx- strong hypoconid and entoconid which unite imately the same height above the tooth row as during wear to form a small posterior fossettid. in M. agatensis. Like the upper molars, the lower molars of LIMBS: The metapodials are subequal in M. exilis are brachyodont and anterposteriorly length, and contrast with those of M. agatensis short. In M. exilis, the length of the premolar in being relatively stouter and unfused. Meta- row is approximately 61 percent of the length carpals II and V are less strongly coossified, 410 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161 and metacarpal V is slightly larger, in M. ex- per canine, small, laterally compressed lower ilis. Metatarsals II and V are fully coossified. canine, and slender ramus with shallow sym- In contrast to M. agatensis, metatarsal II of M. physis, all clearly ally this species with exilis presents a proportionally smaller surface Michenia. The brachyodont molars, unfused for articulation with the entocuneiform, and this metapodials and the primitive features of the surface slopes posteriorly at a higher angle than phalanges are merely primitive characters that the nearly horizontal surface of M. agatensis. were retained after the acquisition of derived The shaft of the proximal phalanx is relatively features characteristic of Michenia. The inclu- stouter than that of M. agatensis. The dorsal sion of exilis in Michenia broadens the previous surface of the shaft exhibits a slight convexity definition of this genus. It does not, however, near the distal articular surface, in contrast to warrant the creation of a new genus, because the straight dorsal surface in M. agatensis. The some specimens in the F:AM collection, al- distal articular surface possesses a deeper though differing somewhat from Michenia ex- groove on the anterodorsal surface, and this ilis and M. agatensis in rostral morphology, surface is less anterodorsally expanded, than in exhibit a morphologically intermediate stage of M. agatensis. Ventrally, the trochleae are more brachyodont molars and weakly coossified met- knoblike and asymetrical than in M. agatensis. apodials. DISCUSSION: Michenia exilis was previously Stevens (1977, p. 48, fig. 15) described a referred to Oxydactylus on the basis of new species, ?Michenia australis from the De- brachyodont molars and uncoossified meta- laho Formation in Texas and compared it with podials. In the discussion of the species, Michenia and Miotylopus but failed to compare however, Matthew and Macdonald (1960, p. 4) it with Oxydactylus exilis Matthew. The species expressed some uncertainty as to the validity of is based on the type mandible (40871-1) with an the generic assignment, stating, "It is ques- associated upper molar and limb fragments. Re- tionable whether this lack of co6ssification may ferred dentitions include two additional mandi- be considered as a generic character of Oxydac- bles and an upper cheek tooth series (40693-24, tylus in spite of the fact that it is often given fig. 15A) which she stated, "are of appropriate this distinction." size for the holotypic and referred jaws." Ste- The elongate, narrow rostrum with a moder- vens (1977, p. 49) stated that ?Michenia aus- ately developed buccinator fossa, incisiform up- tralis "Differs from Michenia agatensis by its smaller size, more prominent molar styles and TABLE 8 ribs, and unfused metacarpal." These are the Measurements (in Millimeters) of the Mandible same characters that distinguish M. exilis (Mat- and Lower-Dentition of the type (AMNH 12997) thew) from M. agatensis. In addition, the type of Michenia exilis (Matthew) (40871-1) of ?Michenia australis includes a par- tial phalanx that was figured by Stevens (1977, Depth jaw between Pi and P2 (12.6) fig. 15D). The distal end of the phalanx is also Depth jaw at P1 16.5 morphologically similar to that of M. exilis and Depth jaw below anterior moiety of M3 29.7 differs from that of M. agatensis in that the P,-2 diastema (22.7) P3 length 10.3 distal articular surface is more grooved and less P3 width 4.3 anteriorly extended. The dentition of ?M. aus- P4 length 11.8 tralis is slightly smaller than that of M. exilis P4 width 6.2 but judging by the size range in other samples Ml length 11.6 of Michenia in the F:AM collection, it is within Ml width 10.5 the range of a single taxon. The age of the M2 length 16.0 Castolon local fauna from the Delaho Forma- M2 width 12.0 tion is considered by Stevens (1977, p. 62) to M3 length 22.9 be early late Arikareean which antedates the M3 width 11.3 early Hemingfordian age of M. exilis. Because M1-3 length 52.4 the type and referred material of ?Michenia 1978 HONEY AND TAYLOR: PROTOLABIDINI 411 australis exhibits no morphological differences (root)-C, P1 unerupted and P2-M3, incomplete to distinguish it from M. exilis, we are placing metacarpus, broken calcaneum, metatarsus and it in synonymy with the latter. first and second phalanges; F:AM 73373, man- dible with 1-C, P1 unerupted and P2-M3; F:AM Michenia yavapaiensis, new species 73360, right partial ramus with P, unerupted Figures 11-12; Tables 9-12 and P3-M3; F:AM 73363, partial mandible with I, (alveolus)-13, C alveolus, P1 unerupted and TYPE: F:AM 73380, right partial ramus with P3-M3; F:AM 73254, right partial ramus with P3 symphysis, ',-13' C erupting, P, unerupted, and -M3; F:AM 73372, right partial ramus with P2- P3-M3, from Milk Creek Quarry, about 103 feet M3 (P3 and M3 broken); F:AM 73362, right above the Manzanita Quarry level, Milk Creek partial ramus with P3-M3; F:AM 73364, right Formation (Clarendonian), Milk Creek area, partial ramus with P3-M3 (P4, MI and M3 Yavapai County, Arizona. broken); F:AM 73253, left partial ramus with KNOWN DISTRIBUTION: Clarendonian of P3-M3 broken; F:AM 73361, right partial ramus Milk Creek Formation, Milk Creek area, with P3-M3; F:AM 73371, left partial ramus Yavapai County, Arizona. with P2-M3 (P3 broken); F:AM 99616, right par- REFERRED MATERIAL: From the type lo- tial ramus with P4-Mi and M2-M3 broken; F:AM cality; F:AM 73283, crushed palate with P1-M3; 73365, right partial ramus with incomplete F:AM 73287, partial palate with P2 (erupting)- symphysis, I,-C roots, P, unerupted, P3-M2 and M2 and M3 broken; F:AM 73286, crushed par- M3 broken; F:AM 99618, right partial ramus tial skull with P2(root)-M3; F:AM 73289, with P2-M3 all broken; F:AM 73259, right par- crushed partial palate with P2 (unerupted)-M2 tial ramus with P3 (root)-M2 and M3 broken; and M3 broken; F:AM 73281, right partial max- F:AM 73407, right partial ramus with M1-M3 illa with P2-M3; F:AM 73398, right partial all broken; F:AM 99614, right ramal fragment maxilla with P2-P3 roots, p4-M2 and M3 broken; with M2 (broken)-M3; F:AM 73399, left imma- F:AM 73288, right partial maxilla with P2-M1; ture partial maxilla with dP4-M2 broken germ, F:AM 73382, right ramus with symphysis, I- and possibly associated with F:AM 73399A, C, P1 unerupted and P3(root)-M3; F:AM 73381, right metacarpus and F:AM 73399B, right left ramus with symphysis, I1 -I3 broken, C, P1 tibia, astragalus, calcaneum, tarsals, right meta- unerupted, P3-P4 and M,-M3 broken; F:AM tarsus and first to third phalanges; F:AM 73261, left partial ramus with P3-M2 and M3 73338, right immature partial ramus with dP2 broken; F:AM 73256, left partial ramus with P3 root and dP3-dP4 broken; F:AM 73342, left im- (broken)-M2 and M3 broken; F:AM 73262, left mature ramal fragment with dP4 and M, broken partial ramus with P3-M2; F:AM 73271A, left germ; F:AM 73407, left metacarpus; F:AM partial ramus with P4 root, M, alveolus and M2- 73400, right metacarpus; F:AM 73364, right M3; F:AM 73407, symphysis with 11-C all metacarpus; F:AM 73408, right metacarpus; broken; F:AM 99613, symphysis with I1I2 F:AM 73406, right metacarpus; F:AM 99649, alveoli and I3-C broken; and F:AM 73262, left incomplete metacarpus; F:AM 73410, right symphysis with I, alveolus and I2-C broken. metatarsus; F:AM 73411, left broken metatar- Manzanita Quarry, 10 feet below Goldie's sus; and F:AM 73371, distal end metatarsus. Honeypot Quarry, Milk Creek Formation (Clar- Shields Ranch Quarry (=Stein Gallery, endonian), Milk Creek area, Yavapai County, UALP locality 1-7), Milk Creek Formation, Arizona: F:AM 73297, right partial maxilla same horizon as Manzanita Quarry (Clarendo- with M1-M2; F:AM 73397, right and left partial nian), Milk Creek area, Yavapai County, Ari- maxillae with P2-M2 and M3 broken; F:AM zona: F:AM 73393, right maxilla with P2-M2 73395, right partial maxilla with P3-M3; F:AM and M3 broken; F:AM 73396, right partial 73407, right detached M2 and M3; F:AM maxilla with P2-M3; F:AM 73394, left partial 73257, right partial maxilla with P2-M' and M?2 maxilla with M1 (broken) -M3; F:AM 73367, broken; F:AM 73294, left partial maxilla with left ramus with P3-M3; F:AM 73387, left partial P2-M2; F:AM 73370, partial mandible with 12 ramus with C root and P3-M,; F:AM 73260, 412 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161 left partial ramus with P3-M2 and M3 broken; M2; UALP 9028, right partial maxillary with P4 F:AM 99611, left partial ramus with incomplete (broken), M1-M2 and M3 (broken); UALP symphysis, P4 (erupting)-M, and M2 broken; 9029, right partial dentary with M2 (roots) and F:AM 99653, right partial ramus with sym- M3; and UALP 9109, right metatarsal. physis, I1-12 alveoli, 13 root, C alveolus, and P3 Deep Spring Quarry (=Johny Post, UALP (root)-M2; F:AM 99617, left partial ramus with locality 1-la), about 69 feet above Manzanita broken symphysis, P4-M2 broken; F:AM 99608, Quarry and 34 feet below Milk Creek Quarry, immature mandible with I-C erupting and P4- Milk Creek Formation (Clarendonian), Milk M2 and M3 erupting; F:AM 99609, right imma- Creek area, Yavapai County, Arizona: F:AM ture ramus with broken symphysis and dP3-M2; 99615, left premaxilla-maxilla fragment with C- F:AM 73340, left immature partial ramus with Pi; and UALP 9373, left astragalus. dP3 (alveolus)-M2; F:AM 99610, left immature The stratigraphic positions of the following partial ramus with dP4; F:AM 73340, in- University of Arizona quarries are unknown. complete symphysis with both Ils erupting; North Fossil Ridge, UALP locality 1-6: F:AM 73324, right metacarpus; F:AM 73409, UALP 9435, left proximal metacarpal; UALP right metacarpus; F:AM 73325, left metatarsus; 9436, left and right proximal radius-ulna; UALP 9389, right partial dentary with P3 UALP 9437, right distal humerus; and UALP (roots), P4 (broken), MI (broken) and M2-M3; 9390, palate with P2-P4 and M1-M2. UALP 5662, right inaxillary with P2-P4 and Creek Bottom, UALP locality 1-4: UALP M1-M3; UALP 9442, right partial ramus with 9388, left partial dentary and symphysis with P3-P4 (roots), Ml (broken), M2 and M3 C (broken), P1 (alveolus), P3 (alveolus), P4, Ml- (broken); UALP 94Q6, left metatarsal; UALP M2, M3 (broken); UALP 9441, partial palate 9407, right metatarsal; UALP 9408, left meta- with M2-M3; UALP 9420, left metatarsal, and tarsal; UALP 9409, distal metapodial; UALP UALP 9421, left metacarpal. 9410, distal metapodial; UALP 9412, right cal- Material from Milk Creek with no data: caneum, left distal humerus; UALP 9413, prox- UALP 9439, right M2-M3; UALP 9124, right imal phalanx; UALP 9414, distal metapodial; distal tibia; UALP 9116, right astragalus. UALP 9415, right astragalus; UALP 9416, left DIAGNOSIS: Differs from M. exilis in Pl less articulated metatarsal, astragalus, calcaneum, elongate with more closely appressed roots; P2 tibia, cuboid and 2 proximal phalanges; UALP relatively smaller with single prominent cusp; 3629, left astragalus; UALP 9417, left as- P3-4 more reduced with smaller parastyles; P1 tragalus; UALP 9418, right cuboid and cal- suppressed; P2 relatively smaller or absent; P34 caneum; and UALP 9419, left proximal radius- smaller; molars taller crowned with weaker ulna. ribs, styles, and stylids; M3 more elongate; Trail Prospect, (same horizon as Shields metapodials fused with metacarpus shorter than Ranch Quarry), Milk Creek Formation (Claren- metatarsus; and first phalanx longer with distal donian), Milk Creek area, Yavapai County, Ar- articular surface less grooved and more ante- izona: F:AM 73279, right partial ramus with riorly produced. P2-M3, and F:AM 73405, left metatarsus. Differs from Michenia agatensis in 11-2 Goldie's Honeypot Quarry (UALP locality frequently lost in adult; 113 higher crowned; P2 7498), Milk Creek Formation, 10 feet above relatively smaller with a single prominent cusp; Shields Ranch horizon, Milk Creek area, P3-4 more reduced with smaller parastyles; P1 Yavapai County, Arizona: UALP 9038, left M1 suppressed; P2 relatively smaller or absent; P34 and left M3; UALP 9037, right M2-M3; UALP smaller; molars higher crowned, with weaker 9039, left P3 and left MI; UALP 9035, palate ribs, styles, and stylids; M3 more elongate; with P2-P4 and M1-M2; UALP 9027, left partial ramus deeper beneath cheek teeth with deeper dentary with P3 (roots), P4 (broken), and M,- mandibular angle; and metapodials shorter, M3; UALP 9036, left P4 and left M1-M2; UALP stockier, and more strongly fused. 9026, right partial dentary with M2-M3; UALP DESCRIPTION AND COMPARISON: Upper Den- 9031, right partial maxillary with P4 and M1- tition: No complete skulls are known and only 1978 HONEY AND TAYLOR: PROTOLABIDINI 413

one rostrum of Michenia yavapaiensis is recog- and P1 (F:AM 99615) in Michenia yavapaien- nized in the Milk Creek collection. The rostrum sis. PI is single cusped and more caniniform (F:AM 73287) is long and slender, 12.4 mm. in than its double-rooted counterpart in M. ag- width at the narrowest point behind P1. In the atensis. The length of PI-2 diastema ranges from type of M. agatensis the rostrum is slightly 26.0 to 30.5 mm. and is longer than that (17.3 crushed and the narrowest point of the rostrum mm.) of the type of M. agatensis. measures approximately 12.0 mm. The rostral p2-4 are smaller than those of M. agatensis. flare between the 13s cannot be determined in P2 is double rooted, though in some specimens the Milk Creek specimen. (particularly from the Shields Ranch Quarry, An I3 alveolus is present in F:AM 73287. F:AM 73393, 73396), the roots are very Anterior to the alveolus of I3 are small depres- closely appressed or even fused, whereas in M. sions, for the healed alveoli of I1 and 12. Thus agatensis P2 roots are more widely flaring. P2 these two teeth are occasionally present in in M. yavapaiensis is shorter and simpler than adults. Other individuals of Michenia in the that of M. agatensis. It has a buccally swollen F:AM collection also exhibit the loss of I1 and paracone and differs from that of M. agatensis

13. Although small, these incisors are present in in lacking a parastyle. A crest runs posterobuc- M. agatensis. Judging by the size of alveolus, cally from the paracone and terminates in a 13 of Michenia yavapaiensis is smaller than that slight swelling (metacone). The metacone is of M. agatensis. The canine of Michenia buccally inflected and separated from the para- yavapaiensis is small, blunt, and transversely cone by a strong buccal groove. Lingually, a compressed, and near the size of that of M. weak basal cingulum extends posteriorly from agatensis. Both the mesial and lateral surfaces the paracone to the base of the metacone. are equally convex; in M. agatensis the lateral P3 is relatively shorter than in M. agatensis, surface is more convex than the mesial. and consists primarily of a central paracone A 12.0 mm. diastema separates the canine connected to a buccally inflected metacone.

TABLE 9 Measurements (in Millimeters) and Statistics of the Upper Dentition of Michenia yavapaiensis, new species from Milk Creek, Arizona N OR X S V I3-C diastema length 1 13.2 C-P1 diastema length 1 12.6 Pl1P2 diastema length 1 30.2 Upper canine length 1 4.7 P1 length 2 4.6-4.7 4.65 P2 length 10 6.5-9.2 7.68 1.064 13.850 P2 width 9 3.4-5.1 4.35 0.507 11.658 P3 length 15 8.3-12.1 10.48 1.003 9.569 P3 width 13 6.1-7.5 6.7 0.427 6.368 P4 length 16 10.0-13.9 11.38 0.908 7.977 P4 width 15 9.7-11.2 10.19 0.420 4.120 M1 length 18 13.3-20.3 16.59 2.016 12.155 M1 width 20 14.7-18.4 16.46 0.940 5.711 M! length 22 20.0-27.9 24.42 2.309 9.455 M2 width 22 13.6-20.7 16.57 1.747 10.544 M3 length 12 23.2-33.4 28.56 2.778 9.725 M3 width 10 13.0-18.5 15.62 1.835 11.750 P2-P4 length 9 26.8-32.9 29.26 7.636 2.230 M1-M3 length 9 60.5-82.8 71.07 6.920 9.743 P2-M3 length 5 84.4-109.2 95.74 9.509 9.932 414 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161

These cusps are separated by a deep buccal worn incisors are spatulate, but with wear I, groove. A crest runs anteriorly from the para- and 12 become more triangular in cross section, cone to a vestigial parastyle which may be with the transverse width equaling or exceeding obliterated in early wear. The paracone is the anteroposterior length. 13 remains spatulate greatly swollen and confluent with the para- even in late wear. style; no buccal groove separates the paracone A very short diastema separates the canine and parastyle as in M. agatensis. Lingual from 13. The small, recurved canine is near the cingula extend from the parastyle and metacone size of that of M. agatensis. It is transversely to enclose anterior and posterior fossettes in compressed with the buccal surface slightly little worn teeth. The anterolingual and post- more convex than the lingual surface. erolingual cingula do not generally meet, and P1 of M. yavapaiensis is nearly always sup- are less distinct and without the tiny cuspules pressed, in contrast, to that of M. agatensis seen on M. agatensis. which is fully erupted. At its maximum de- P4 is only slightly shorter than in M. agaten- velopment P1 approximates the size of P1 in the sis. The paracone is well developed, but the type specimen of M. agatensis. P1 of M. parastyle is much smaller than that of M. ag- yavapaiensis possesses a prominent central cusp atensis. The metastyle, when present, is only from which anterior and posterior crests extend weakly expressed. The protocone is prominent, to terminate in vestigal cuspules; P1 is never but the metaconule is represented by only a elongate as in the referred jaw of M. agatensis posterior elongation of the protocone or a weak (F:AM 14259). Many of the jaws of M. posterolingual crest. yavapaiensis have never been opened to expose The molars of M. yavapaiensis are taller the unerupted P1. However, P1s that are ex- crowned than those of M. agatensis, and M3 is posed show that the roots are fused and the especially more anteroposteriorly elongate. M1 tooth is oriented in any position from horizontal exhibits a strong mesostyle and weak parastyle; to upright. The suppressed P1 is situated either the metastyle is faintly indicated on only a few directly above, or above and slightly anterior to specimens, and a faint rib may be present on the mental foramen. the paracone. In M. agatensis M' has ribs on P2 in M. yavapaiensis is extremely variable; both paracone and metacone, and the paracone it may be erupted, suppressed, or absent. When rib is relatively stronger than in M. yavapaien- P2 is present, it is usually peglike, and single sis. Both the parastyle and the mesostyle are rooted; it is sometimes vestigial. In contrast, P2 usually stronger in M. agatensis. M2 and M3 of of M. agatensis is double rooted and elongate M. yavapaiensis possess strong parastyles and with three well-defined cusps. P2 of M. mesostyles, and M3 also has a strong metastyle. yavapaiensis consists of only a protoconid; the Faint ribs are present on the paracone and paraconid and entoconid, which are present in metacone of both molars with the paracone rib M. agatensis, are absent. Most jaws referred to being the stronger. Compared to M. yavapaien- Michenia yavapaiensis lack an erupted P2. Six sis, M. agatensis has much stronger vertical jaws without an erupted P2 were excavated to ribs and styles on M! and M3. determine if this tooth was suppressed; in only P2-4 length in M. yavapaiensis is shorter rela- one jaw (F:AM 73305) was P2 found. From tive to M'-3 length than in M. agatensis. This is this small sample it appears that P2 is more due both to the reduction in the size of the frequently absent than suppressed. Most speci- premolars and the elongation of M3. For exam- mens with erupted P2s come from strat- ple, in M. agatensis p2-3 length is equal to or igraphically lower quarries, in which the slightly greater than M3 length; in contrast, p2-3 individuals also average slightly smaller. length in M. yavapaiensis is only slightly However, within the same stratigraphic hori- greater than one-half M3 length. zon, no relationship was observed between the LOWER DENTITION: The incisors of M. size of the animal and eruption or noneruption yavapaiensis are taller crowned and in early of P2. wear are more vertically oriented and less pro- P3 in M. yavapaiensis is double rooted, but cumbent than those of M. agatensis. Little shorter and simpler than that of M. agatensis. 1978 HONEY AND TAYLOR: PROTOLABIDINI 415

TABLE 10 Measurements (in Millimeters) and Statistics of the Lower Dentition of Michenia yavapaiensis, new species from Milk Creek, Arizona N OR X S V Depth jaw between Pi and P2 23 13.3-19.1 16.64 1.142 6.863 Depth jaw at P1 13 16.1-20.7 17.77 1.420 7.993 Lower canine length 6 4.9- 6.0 5.50 0.400 7.272 P2 length 5 4.1- 5.7 4.82 0.691 14.328 P2 width 4 2.8- 3.0 2.90 0.115 3.982 P3 length 21 6.1- 9.2 8.44 0.776 9.186 P3 width 20 2.5- 5.6 4.24 0.608 14.345 P4 length 30 8.8-12.9 10.30 0.883 8.570 P4 width 27 4.6- 7.0 5.95 0.563 9.457 M1 length 29 15.6-18.9 16.13 1.429 8.866 ml width 23 9.7-11.9 10.76 0.704 6.542 M2 length 31 18.5-24.8 21.39 1.495 6.989 M2width 26 11.6-14.5 12.68 7.861 6.190 M3 length 24 28.7-37.3 32.86 2.260 6.879 M3 width 21 9.9-13.8 11.85 0.962 8.114 P3-P4 length 18 16.2-20.5 18.28 1.129 6.173 M1-M3 length 17 64.7-88.8 72.78 5.939 8.159 P2-M3 or P3-M3 length 14 82.1-97.4 89.60 3.854 4.301 Postcanine diastema length 9 45.6-83.1 57.21 11.909 20.817

In early wear P3 possesses a central protoconid greater than in the latter. For example, the that is less swollen than in M. agatensis. Ante- crown height of the second lobe of M3 (F:AM riorly, a crest extends to a diminutive, lingually 73386) in M. yavapaiensis measures 34.0 mm. inflected paraconid; this cusp is much smaller M3s of M. agatensis (AMNH 14255 and and less inflected than in M. agatensis. A sec- AMNH 14259) are less worn, yet the crown ond crest from the protoconid extends posteri- heights of the second lobes measure only 24.2 orly and divides to join with a small entoconid and 23.8 mm., respectively. The lingual walls and a slightly larger hypoconid. The entoconid of the molars in M. yavapaiensis have small and hypoconid do not unite to form a fossettid stylids and ribs, and are thus smoother than in as in P4, but are separated by a lingual groove M. agatensis. In early wear, Ml and M2 which does not quite reach the base of the possess a small entostylid. The parastylid and tooth. metastylid are also present only as small swell- P4 in M. yavapaiensis is generally shorter ings in early wear. Lingual ribs are absent or than that of M. agatensis with the paraconid extremely faint. In M. agatensis the lingual ribs strongly inflected lingually but smaller than in and stylids are usually much stronger; however M. agatensis. The protoconid is less bulbous the metastylid and entostylid are reduced in one than in M. agatensis, and P4 is widest across referred jaw (AMNH 14259). As in M. agaten- the entoconid and hypoconid, which unite to sis, the hypoconulid broadens near the base of form a fossettid. M3; thus with wear the occlusal surface changes The molars of M. yavapaiensis are higher from an angular crescent to a rounded teardrop crowned and M3 more elongate than in M. shape. Well worn M3s of M. yavapaiensis have agatensis. Furthermore, the premolars are very flat lingual walls, and the molars are gen- smaller relative to the size of the elongate mo- erally broader than those of M. agatensis. lars than in M. agatensis. Although the small- MANDIBLE: The mandible of M. yavapaien- est Milk Creek molars are no longer than those sis resembles that of M. agatensis, with the of M. agatensis, the crown height is always symphysis slender dorsoventrally and narrow 416 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161

A

PS

B

C

D

FIG. 11. A-E. Michenia yavapaiensis from the Clarendonian of Arizona. A. F:AM 73363, from Manzanita Quarry, lateral view, reversed. B. F:AM 73405, from Trail Prospect, metatarsus, anterior view. C-E. F:AM 73370, from Manzanita Quarry. C. Restored metacarpus, anterior view. D. Metatarsus, anterior view. E. Lateral view. All X1/2. transversely, and the width narrowest between smaller size, there is little to distinguish the the C and P1. As in M. agatensis, the sym- lower deciduous teeth of M. yavapaiensis from physis is only slightly inclined from the hori- those of Protolabis coartatus. The most reliable zontal. The posterior border of the symphysis means for separating the deciduous dentition of lies between the mental foramina. Due to the M. yavapaiensis from that of P. coartatus is by suppression of P1 and the small size of P24, the the smaller size of the erupting molars. postcanine diastema is much longer in M. No specimens of M. yavapaiensis have dP2, yavapaiensis. Between the suppressed P1 and but on two specimens (F:AM 73338 and P3, the ramus is exceedingly slender, with the 73340), the alveoli indicate a small tooth with shallowest depth slightly posterior to P1. Be- fused roots. A very worn dP3 consists of a cause of the taller crowned teeth, the depth of central primary cusp, a lingually inflected ante- the horizontal ramus posterior to P3 is greater in rior crest, and a posterior cusp formed from the M. yavapaiensis, and the mandibular angle is hypoconid and entoconid. DP4 is shorter than also deeper. that of P. coartatus, and has a weakly de- DECIDUOUS DENTITION: The deciduous teeth veloped parastylid and entostylid on the flat of M. agatensis are unknown. Aside from lingual wall. Lingual ribs are faint. 1978 HONEY AND TAYLOR: PROTOLABIDINI 417

A -.-- qii pi

B

C

PI

D

E

F

FIG. 12. A-F. Michenia yavapaiensis from the Clarendonian of Arizona. A-B. F:AM 73283, from Milk Creek Quarry. A. Occlusal view. B. Lateral view. C. F:AM 99615, from Deep Spring Quarry, lateral view, reversed. D-F. From Milk Creek Quarry. D-E. F:AM 73380. D. Occlusal view, reversed. E. Lateral view, reversed. F. F:AM 73382, lateral view, reversed. All X½/2.

POSTCRANIA: The metapodials of M. than those of M. agatensis. Comparative meas- yavapaiensis are shorter than those of M. ag- urements and statistics are given in table 11. atensis. However, there is not a corresponding DIscusSION: As noted above, the derived slenderness in the width of the shaft, and the features of M. yavapaiensis, coupled with the metapodials are stockier than those of M. ag- primitive features of M. exilis, have necessi- atensis. The metapodials are also more strongly tated a revision of the original diagnosis for fused in M. yavapaiensis, and the distal separa- Michenia. In view of the differences between tion does not extend so far proximally as in M. M. agatensis, the genotypic species, and M. agatensis. The other postcranial elements of M. yavapaiensis, some question may arise as to the yavapaiensis generally average slightly smaller generic assignment of the new Clarendonian 418 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161

TABLE 11 Measurements (in Millimeters) and Statistics of Referred Metapodials of Michenia yavapaiensis, new species from Milk Creek, Arizona N OR X S V Metacarpal length 9 182.1-196.7 191.07 5.380 2.815 Metacarpal proximal width 9 27.5- 32.0 29.82 1.527 5.1206 Metacarpal shaft narrowest transverse width 10 14.2- 19.2 16.39 1.631 9.952 Metatarsal length 6 192.3-202.0 198.08 1.9097 3.782 Metatarsal proximal width 12 21.7- 28.0 24.52 1.564 6.379 Metatarsal shaft narrowest transverse width 9 11.3- 15.2 13.42 1.131 8.430 taxon to Michenia. But after observing the due to greater overlap between the two groups. transformation of characters from M. agatensis This overlap is partially attributable to the fact to M. yavapaiensis in the undescribed samples that the tapering and curvature of the upper in the F:AM collection, it is evident that the molars makes it impossible to be as rigorous in new species is Michenia. As pointed out in the measuring the lengths and widths of these description and comparison the morphological teeth. This taper and curvature also contributes features that distinguish M. yavapaiensis are to the high coefficients of variation for the shared to a degree with M. agatensis and indi- upper molars in both groups. cate their relationship. The derived features of M. yavapaiensis are an exaggeration of the TABLE 12 character trends first seen in M. exilis and M. Student's t-test for Dental and Skeletal agatensis. The transformation of these charac- Measurements of Protolabis coartatus and ters from primitive to derived is in general Michenia yavapaiensis, new species from Milk accord with the temporal sequence of un- Creek, Arizona described samples in the F:AM collection. Degrees Because of the several similarities between Variate Freedom p Michenia yavapaiensis and Protolabis coar- tatus, it could be hypothesized that these dis- P2 length 17 3.2607 .001-.01 tinct species are actually sexual variants of the P3 length 29 2.6789 .01-02 PI length 33 2.5463 .01-02 same species. Thus the slender-jawed and M1 length 32 2.3329 .02-.05 slender-limbed Michenia yavapaiensis would be MN length 43 2.3618 .02-05 the females of Protolabis coartatus. This hy- M3 length 25 4.6835 <.001 pothesis is countered by the fact that based on P2-P4 length 20 1.936 .05-.1 the jaw depth the rami can be easily separated M1-M3 length 21 1.8054 .05-.1 into two groups without any overlap. Correl- P2-M3 length 14 2.0451 .05-.1 ated with this separation is the nonoverlap in Lower canine length 20 6.2712 <.001 the size of the lower canine between the two P2 length 13 2.4419 .02-05 groups. Similarly, the metapodials are easily P3 length 43 7.1562 <.001 separated into a longer, stocky group and a P4 length 56 6.108 <.001 shorter, M1 length 53 2.6085 .01-02 slender group. In our opinion, the size M2 length 56 4.7524 <.001 and proportional differences between these M3 length 47 10.113 <.001 groups are too great to be attributed to sexual P3-P4 length 34 7.608 <.001 variation. Student's t-tests were run for various M,-M3 length 35 4.4263 <.001 dental and skeletal measurements, and the re- P2-M3 or P3-M3 length 27 7.9347 <.001 sults are given in table 12. The high values of Depth jaw between Pi "t" for the majority of cases support the dis- and P2 40 18.763 <.001 tinctness of the two samples and are considered Depth jaw at P1 27 12.653 <.001 to be of generic significance. The values of "t" Metacarpal length 27 7.66 <.001 for the upper dentition are somewhat smaller, Metatarsal length 17 7.4704 <.001 SUMMARY OF RELATIONSHIPS In establishing the Protolabidini as a mono- phyletic group by sharing the following ap- phyletic group, we have relied upon cladistic omorphic characters: methodology as advocated by Henning (1966). (1) limbs elongate Simply put, we have searched for shared de- (2) metapodials longer than the basal length rived (synapomorphic) characters, which unite of the skull Tanymykter, Protolabis, and Michenia. We de- (3) cervical vertebrae elongate termine the polarity of a character (either primi- tive or derived) by outgroup comparison. This The Camelinae (point 3) are united by the fol- same comparison and the recognition of shared lowing apomorphic characters: derived features is the evidence used here to (1) weak buccinator fossa identify the primitive sister taxon to Protolabis (2) rostrum elongate and Michenia, and is the basis for our hypoth- The Protolabidini (point 4) form a monophy- esis of relationships to the other camelids. We letic group and possess the apomorphic char- choose to present these relationships by means acters of: of a cladogram (fig. 13). We believe that a (1) rostrum narrow cladogram represents the most objective ex- (2) laterally expanded anterior nares pression of our knowledge of these relation- ships, and in addition forms a testable Tanymykter brachyodontus (point 5) is derived hypothesis based upon the observed polarity of in having closely appressed P' roots. character states. Protolabis and Michenia (point 6) are derived The metapodials in the Oromerycinae and with respect to Tanymykter brachyodontus by: Poebrotheriinae (sensu Wilson, 1974) are short (1) the absence of elongate basioccipital tu- and unfused with the length of the metacarpus berosities less than that of the metatarsus. In addition, the (2) p2 without strong, continuous lingual molars are brachyodont and possess a meso- cingulum style above but lack a metastylid below. Short (3) auditory bulla less inflated with medial unfused metapodials with the metacarpus plates more compressed shorter than the metatarsus, and brachyodont (4) buccinator fossa moderate to strong molars with mesostyles but lacking metastylids, Protolabis (branching point 7) possesses the are considered primitive characters of the Cam- apomorphic characters of: elidae. (1) molars hypsodont The Aepycamelinae and the Camelinae (2) M3 anteroposteriorly elongate (point 1) share the apomorphic character of a (3) metastylid on lower molars very weak to metastylid on the lower molars. In this feature absent they differ fundamentally from the (4) ventrally produced mandibular angle oromerycines, poebrotheriines, stenomylines, with weak to strong lateral flare Miotylopus and Pseudolabis in which the meta- (5) metapodials fused stylid is absent. The stenomylines, Miotylopus (6) first phalanx elongate with distal articu- and Pseudolabis share a derived feature which lar surface anteriorly extended is absent in the Aepycamelinae and the Cam- elinae, namely, a weak to absent mesostyle. Protolabis heterodontus (point 8) is primitive in The Aepycamelinae and Camelinae retain the all known characters relative to Protolabis primitive condition of a strong mesostyle on the coartatus. upper molars. Protolabis coartatus (point 9) is distinguished by: (1) ventral rostral constriction extreme The Aepycamelinae (point 2) comprise a mono- (2) 11-2 usually absent 419 420 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161

Aepycamelinae

Other Camelinae

Tanymykter brachyodontus

| Protolabis heterodontus

Proto/abis coartatus

Michenia exilis

Michenia agatensis

(W Michenlo yavopolensis

FIG. 13. Cladogram of proposed relationships within the Protolabidini and an outgroup comparison with the Aepycamelinae and the other members of the Camelinae (Camelini, Camelopini, and Lamini). Characters indicated at the numbered nodes are discussed in the body of the text.

(3) p2-3 without parastyle (2) 13-C/weak, /C small (4) upper molars with weak ribs (3) symphysis shallow (5) P1 small and occasionally absent (4) inflection of mandibular angle sup- (6) P2 small and usually absent pressed (7) mandibular angle with extreme lateral flare and mesial shelf Michenia exilis (point 11) is primitive in all Michenia (point 10) is derived relative to Pro- known characters relative to other species of tolabis as follows: Michenia. (1) braincase short Michenia agatensis and Michenia yavapaiensis 1978 HONEY AND TAYLOR: PROTOLABIDINI 421

(point 12) share the following apomorphic char- Michenia. The mandible of Hemiauchenia and acters: Lama is relatively slender as is that of (1) molars anteroposteriorly elongate and Michenia, and in addition Lama possesses rela- hypsodont tively short metapodials, as do some Michenia. (2) P1 small with roots closely appressed There are, however, some important differ- (3) metapodials fused ences between Michenia and the lamines. The (4) first phalanx elongate with distal articu- posterior narial notch consistently extends far- lar surface less grooved and more ante- ther forward on the palatines in the lamines riorly extended than in Michenia. For example, on 16 mature (5) metapodials slender (M3 in wear) Lama skulls examined in the AMNH (M) collection, the anterior border of (point 13) is primitive the posterior narial notch is opposite the second Michenia agatensis lobe of M2, or rarely between M2 and M3. On relative to M. yavapaiensis. rel- 11 mature Hemiauchenia palates, the anterior Michenia yavapaiensis (point 14) is derived border of the posterior nares is nearly always ative to M. agatensis by: (1) 1J-2 usually absent opposite the posterior lobe of M2, except in one case, where it is opposite the valley between (2) P1 unerupted in adults the first and second lobes of M3. On three or ab- (3) P2 small and generally unerupted Paleolama specimens, the anterior border is op- sent posite the first or second lobes of M2. In con- on to ab- (4) metastylid lower molars weak trast, on 16 mature specimens of Michenia the sent anterior border of the posterior nares extends (5) upper and lower molars tall crowned and no farther forward than the second lobe of M3, anteroposteriorly elongate and often is posterior to M3. We view the (6) metapodials short and slender anterior extension of the posterior narial notch in the Lamini as the retention of a primitive Webb, (1965, p. 44) recognized the Pro- character state, for it is similarly extended in tolabidini as a tribe of the Camelinae, and sug- Poebrotherium, Gentilicamelus, Paratylopus gested that it was ancestral to the other three primaevus, Paralabis, Miotylopus, the tribes of the Camelinae. At that time Protolabis Oromeycinae and the Protoceratidae. was the only known member of the tribe. Later Upper molars of Lama and Paleolama, and (1972, p. 107) he stated that Protolabis and to a lesser degree those of Hemiauchenia, pos- Michenia "lie at the base of the radiation of the sess relatively strong ribs and styles and MI modern camelids." On the basis of our study, tend to be relatively short. P44 of Hemiauchenia we shall briefly examine this hypothesis. are large, unreduced teeth. In the above primi- The weak upper and lower canines, highly tive characters the lamines are similar to M. constricted rostrum, slender metapodials, shal- exilis and M. agatensis. Michenia yavapaiensis low symphysis, and short braincase are derived and other examples of derived Michenia pos- features which eliminate Michenia as the primi- sess weaker ribs and styles on the upper mo- tive sister taxon to all later camelids except the lars, more elongate M3, and reduced P4. These Lamini. Undescribed material in the F:AM col- same Michenia specimens possess a highly con- lection shows a remarkable similarity between stricted rostrum with a strongly developed buc- the skulls of some Michenia and some lamines, cinator fossa. Although most camelids possess in particular those of Lama and Hemiauchenia. a relatively elongate rostrum showing some Features held in common between Michenia narrowing anterior to the cheek teeth, the great and some lamines include a short skull with narrowing of the rostrum in Michenia is surely retracted nasals, a short, rounded braincase derived. Hemiauchenia possesses a moderately with a weak sagittal crest, relatively large lacri- constricted rostrum with a moderately devel- mal vacuities, and rather strong, laterally pro- oped fossa in some specimens, whereas other duced postorbital bars. Hemiauchenia and specimens have the maxillary inflated above the Lama both have greater reduction in the num- rostrum. Lama and Paleolama show little evi- ber of premolars, a derived feature relative to dence of the buccinator fossa and have more 422 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161

strongly inflated maxillaries. The present evi- the angle, provides a greatly increased surface dence suggests the derived condition for the area for the attachment of the masseteric mus- lamines is inflated maxillaries, contrasting with cle. These changes are associated with the low, the strongly excavated maxillaries of Michenia. weak, coronoid process and reflect increased The lamines also possess sharp, recurved, ca- emphasis on the masseter, and decreased em- niniform I3 and C/C, which contrast with the phasis on the temporalis. These features are weak and generally noncaniniform I3 and C/C evident, though less pronounced, in the primi- of Michenia. A relatively large caniniform I3 tive samples of Protolabis. Procamelus, on the and lower canine occur in Poebrotherium, Gen- other hand, possesses a convex lower border on tilicamelus, Paratylopus privaevus, Paralabis, the mandible and an elongate coronoid process. Pseudolabis, and Miotylopus; thus, their small The elongate coronoid process is a derived fea- size in Michenia is considered a derived fea- ture not exhibited by Protolabis, and reflects ture. The large upper canine of the lamines the increased emphasis placed on the tem- would, of course, be a derived feature of this poralis, an evolutionary trend opposite to that group, as would be the laterally compressed, of Protolabis. bladelike form of the caniniform teeth. Finally, Protolabis is also derived relative to Pro- the metapodials of Michenia are short relative camelus in possessing moderate to extreme ros- to those of Hemiauchenia, and those of M. tral constriction. This is another progressive yavapaiensis are relatively shorter than those of character which attains maximum specialization M. agatensis. Although the above differences in derived samples of Protolabis; however, in do not entirely eliminate Michenia as the sister primitive samples the width of the muzzle oc- group to the Lamini, we believe the character casionally overlaps that of the narrowest rostra polarities of Michenia do suggest that a rela- of later Procamelus. Interestingly, the relatively tionship between the Lamini and derived spe- wide rostrum of Procamelus is also derived, cies of Michenia is unlikely, because and is correlated with an increase in the size of evolutionary reversals in many characters Pi and an increase in the palatal distance be- would be necessary. tween the upper canines. Thus, the specializa- Although Protolabis most closely resembles tions in each genus are again in opposite Procamelus of the later camelines, it exhibits directions. The metapodials in primitive sam- progressive evolutionary specializations not ples of Protolabis approximate the basal present in Procamelus, which makes it highly lengths of isolated skulls from the same lo- unlikely that the two are closely related. In calities; in derived samples the metapodials are Protolabis, the subdued angular process (result- much shorter and approximate the incisive ing in a small "hook") on the posterior edge of border to orbital distance. Those metapodials the ascending ramus is a derived feature absent contrast markedly with much longer, more in Procamelus. The angular process, which is robust metapodials of Procamelus. The maxil- slightly stronger in primitive Protolabis sam- lary fossa of derived species of Protolabis is ples, is nearly absent in derived samples. The often very shallow, and the maxillary fossa of reduction of the angular process is directly cor- all Protolabis contrasts with the more posteri- related with the posterior and dorsal expansion orly extended, pocketlike fossa of Procamelus. of the mandibular angle, due, as hypothesized In Protolabis the retention of the rectus capitus above, to the increased development of the ventralis troughs separated by a narrow median masseteric muscle. The more pronounced angu- ridge is a primitive character, inasmuch as this lar process (stronger "hook") in Procamelus condition is also present on Poebrotherium. resembles the primitive condition seen in However, the derived condition in Protolabis is Poebrotherium. strongly depressed troughs that differ markedly Protolabis is also derived relative to Pro- from the weak troughs in Procamelus. camelus, in possessing a straight to ventrally The distinguishing features discussed above produced lower border on the mandibular convince us that a close relationship does not angle. This, along with the dorsal expansion of exist between Procamelus and Protolabis. The 1978 HONEY AND TAYLOR: PROTOLABIDINI 423 most important differences between these gen- tolabis and therefore qualifies as the primitive era involve the masticatory musculature, and sister taxon. Thus Oxydactylus, as previously the evidence for their divergent specializations considered, is a paraphyletic group, consisting is seen in even the most primitive samples of of the primitive sister taxa of both the Pro- each genus. In view of these differences with tolabidini and the Aepycamelinae. Procamelus, we think it unlikely that Pro- Our cladistic analysis highlights the parallel- tolabis lies "at the base of the radiation of the ism so common within the Camelidae, both modem camelids" (Webb 1972, p. 107). between closely related genera and between Tanymykter brachyodontus is more primitive more distantly related groups. For example, than Protolabis in possessing unfused meta- recognition of exilis as a species of Michenia podials, brachyodont, unelongate molars, and necessitates that fusion of metapodials and de- more inflated bullae. The primitive sister taxa velopment of hypsodonty occurred independ- to Protolabis and Procamelus would presuma- ently in Protolabis and Michenia. These genera bly possess these characters. Tanymykter, further paralleled each other in tooth reduction however, more closely resembles Protolabis in and increased rostral constriction. Procamelus possessing the derived features of strong rostral and the three "advanced" tribes of the Cam- constriction and lateral expansion of the ante- elinae (Camelopini, Lamini, and Camelini) par- rior nares. In addition, the strongly depressed alleled (and of course, sometimes surpassed) rectus capitus ventralis troughs with a narrow the Protolabidini in reduction and loss of ante- median ridge are features not seen in Pro- rior teeth, development of digitigrade stance camelus. Other similarities include relatively and fusion of metapodials. Similar changes, in small coronoid and angular processes on the various combinations, occur in more distantly mandible, weak, laterally situated upper in- related groups, such as the aepycamelines and cisors, and a dorsoventrally narrow rostrum. stenomylines. Such changes occur as con- Aside from the possession of the primitive vergence in different monophyletic groups, and characters of unfused metapodials, brachyodont therefore do not always indicate relationship, molars, and more inflated bullae, Tanymykter but are characteristic of camelid evolution as a brachyodontus is nearly identical with Pro- whole.

LITERATURE CITED Anderson, C.A., and P.M. Blacet Cope, E.D., and W.D. Matthew 1972. Geologic map of the Mount Union Quad- 1915. Hitherto unpublished plates of Tertiary rangle, Yavapai County, Arizona. U.S. mammalia and vertebrata, Amer. Mus. Geol. Survey Geol. Quad. Map GQ-997. Nat. Hist. with U.S. Geol. Surv. Monogr. Cope, E.D. ser. 2, pls. 1-154. 1874. Report on the stratigraphy and Pliocene DeMar, R., and H. Barghusen vertebrate paleontology of northern Color- 1972. Mechanics and the evolution of the synap- ado, Bull. U.S. Geol. Geogr. Surv. Terr., sid jaw. Evolution, vol. 26, pp. 622-637, serv. 1, vol. 1, no. 1, pp. 9-28. figs. 1-13. 1876. On a new genus of Camelidae, Proc. Frick, C., and B.E. Taylor Acad. Nat. Sci. Philadelphia, vol. 28, pp. 1971. Michenia, a new protolabine (Mammalia, 144-147. Camelidae) and a brief review of the early 1877. Report upon the extinct vertebrata ob- taxonomic history of the genus Protolabis. tained in New Mexico by parties of the Amer. Mus. Novitates, no. 2444, pp. expedition of 1874. In Wheeler, George 1-24, figs. 1-4, tables 1-3. M., Report upon United States geographi- Gauthier-Pilters, H. cal surveys west of the one hundredth me- 1974. The behavior and ecology of camels in the ridian. Washington, vol. 4, pt. 2, Sahara, with special reference to nomad- paleontology, pp. 1-370, pls. 22-83. ism and water management. In The behav- 424 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 161

ior of ungulates and its relation to Loring, S.H. and A.E. Wood management. vol. 2, International Union 1969. Deciduous premolars of some North for Conservation of nature and natural re- American Tertiary camels (Family Cam- sources, Publ., new series, no. 24, pp. elidae). Jour. Paleo., vol. 43, no. 5, pp. 5 12-936. 1199-1209, figs. 1-6. Gregory, W.K. McKee, E.H., and C.A. Anderson 1920. Studies in comparative myology and os- 1971. Age and chemistry of tertiary volcanic teology, no. 5, on the anatomy of the rocks in north-central Arizona and relation preorbital fossa of Equidae and other un- of the rocks to the Colorado Plateaus, gulates. Bull. Amer. Mus. Nat. Hist. vol. Geol. Soc. Amer. Bull., vol. 82, pp. 42, pp. 265-283, figs. 1-26, pl. 18. 2767-2782, figs. 1-7, tables 1-3. Henning, W. McKenna, M.C. 1966. Phylogenetic systematics, Urbana, Univ. 1965. Stratigraphic nomenclature of the Miocene Illinois, Press. 263 pp. Hemingford group, Nebraska. Amer. Henshaw, P.C. Mus. Novitates, no. 2228, pp. 1-21, fig. 1940. (I.E., 1939) A Tertiary mammalian fauna 1. from the Avawatz Mountains, San Bernar- 1966. Synopsis of Whitney and Arikareean Cam- dino County, California. Carnegie Inst. elid phylogeny. Amer. Mus. Novitates, Wash., Cont. Paleo., no. 514, pp. 2-30, no. 2253, pp. 1-11, fig. 1. figs. 1-3, pls. 1-6. McKenna, M.C., and J.D. Love Honey, J.G. 1972. High-level strata containing early Miocene [MS] The Paleontology of the Brown's Park mammals on the Bighorn Mountains, Wy- Formation in the Maybell, Colorado Area, oming. Amer. Mus. Novitates, no. 2490, and a taphonomic study of two fossil pp. 1-31, figs. 1-18. quarries in Colorado and Arizona. Un- Matthew, W.D., and J.R. Macdonald publ. Master's thesis, Univ. Arizona, Tuc- 1960. Two new species of Oxydactylus from the son, pp. 1-197, figs. 1-26, tables 1-14. Middle Miocene Rosebud Formation in Hook, Donald L. Western South Dakota. Amer. Mus. Novi- [MS] Late Cenozoic stratigraphy and structure tates, no. 2003, pp. 1-7, figs. 1-5. of a part of the Walnut Grove Basin, Merriam, J.C. Yavapai County, Arizona. Unpubl. Mas- 1916. Tertiary vertebrate fauna from the Cedar ter's thesis, Univ. Arizona, Tucson, 50 Mountain region of western Nevada. pp., 13 pls. Univ. California. Publ. Geol. Sci., vol. 9, Leidy, J. pp. 161-198, figs. 1-48. 1858. Notice of remains of extinct Vertebrata, Ostrom, J. from the valley of Niobrara River, col- 1964. A functional analysis of jaw mechanics in lected during the exploring expedition of the dinosaur Triceratops. Postilla. no. 88, 1857, in Nebraska, under the command of pp. 1-35, figs. 1-10, tables 1-2. Lieut. G.K. Warren, U.S. Top. Eng. by Patton, T.H. Dr. F.V. Hayden, Geologist to the expedi- 1969. Miocene and Pliocene artiodactyls, Texas tion. Proc. Acad. Nat. Sci. Philadelphia, Gulf Coastal Plain. Bull. Florida State vol. 10, pp. 20-29. Mus. Biol. Sci., vol. 14, no. 2, pp. 1858. [Description of Procamelus robustus and 115-226, 34 figs. Procamelus gracilis] Ibid., vol. 10, pp. Peterson, O.A. 89-90. 1904. Osteology of Oxydactylus, a new genus of Lewis, G.W. camels from the Loup Fork of Nebraska, 1968. Stratigraphic Paleontology of the Barstow with descriptions of two new species. Formation in the Alvord Mountain Area, Ann. Carnegie Mus. Nat. Hist., vol. 2, San Bernardino County, California. U.S. no. 3, pp. 434-476, figs. 1-3, pls. 4-15. Geol. Surv. Prof. Paper 600-c, pp. c75- 1906. The Miocene beds of western Nebraska c79, 1 fig. and eastern Wyoming and their vertebrate Loomis, F.B. faunae. Ibid., vol. 4, art. 3, pp. 21-72, 191 1. The camels of the Harrison beds, with figs. 1-20, pls. 9-19. three new species, Amer. Jour. Sci., vol. Plafker, Lloyd 31, pp. 65-70, figs. 1-3. [MS] Geologic reconnaissance of the Cenozoic 1978 HONEY AND TAYLOR: PROTOLABIDINI 425

Walnut Grove Basin, Yavapai County, Publ. Geol. Sci., vol. 18, no. 11, pp. Arizona. Unpubl. Master's thesis, Univer- 291-302, figs. 1-9 sity of Arizona, Tucson, 56 pp., 1 fig., 9 Turnbull, W.D. pIs. 1970. Mammalian masticatory apparatus. Fieldi- Schlaikjer, E.M. ana: Geology, vol. 18, no. 2, pp. 1935. Contributions to the stratigraphy and pal- 149-356, figs. 1-48, tables A-E. eontology of the Goshen Hole area, Wy- Webb, S.D. oming. IV. New vertebrates and the 1965. The osteology of Camelops. Bull. Los stratigraphy of the Oligocene and early Angeles Co. Mus. Sci. No. 1, pp. 1-54, Miocene. Bull. Mus. Comp. Zool., Har- figs. 1-22, tables 1-12. vard, vol. 76, no. 4, pp. 97-189, figs. 1969. The Burge and Minnechaduza Clarendo- 1-13, pls. 1-41, May. nian mammalian faunas of northcentral Schultz, C.B. Nebraska. Univ. California Publ. Geol. 1938. The Miocene of western Nebraska. Amer. Sci., vol. 78, pp. 1-191, figs. 1-46. Jour. Sci., vol. 35, pp. 441-444. 1972. Locomotor evolution in camels. forma et Sisson, S., and J.D. Grossman functio, vol. 5, pp. 99-112, 5 figs. 1975. The Anatomy of the Domestic Animals. 1974. Pleistocene Llamas of Florida with a brief vol. 1, p. 1-1211. figs. 1-1 to 36-34. review of the Lamini. In Pleistocene Stevens, M.S. mammals of Florida. A Univ. of Florida 1977. Further study of Castolon local fauna Book. Gainesville, The Univ. Presses of (early Miocene) Big Bend National Park, Florida, 1974, pp. VII-270, figs. Texas. The Pearce-Sellards Series no. 28, 2.1-10.6. Texas Mem. Mus., pp. I-V and 1-69, Wilson, J.A. figs. 1- 17. 1974. Early tertiary vertebrate faunas, Vieja Stirton, R.A. Group and Buck Hill Group, Trans-Pecos, 1929. Artiodactyla from the fossil beds of Fish Texas: Protoceratidae, Camelidae, Hyper- Lake Valley, Nevada. Univ. California tragulidae. Bull. 23, Texas Mem. Mus., pp. 1-34, figs. 1-36.

11'

,

MAE

I

I

I

n ERFAMEMINW, NV~~~%

o V , :4

a ¾

.4~~~~~~~~~~~~~~M

$4~~ ~ ~ ~ ~ ~ ~~~~~1~ ~ ~ ~~~

ImMR