Canis feneus, TEDFORD R. H. & WANG X. & TAYLOR B. E., 2009

TEDFORD R. H., WANG X. & TAYLOR B. E., 2009, Phylogenetic Systematics Of The North American Fossil Caninae (Carnivora: Canidae), Bulletin of the American Museum of Natural History 2009 (325), pp. 1-218 : 121-143

Canis feneus , new species Figure 45F–I; appendix 3

Type: F:AM 25515, both partial maxillae with P3 (broken)–M3 (fig. 45F–G) and parts of both mandibular rami with c, p1–p2, p3– p4, and m1 alveoli, right m2, and m3 alveolus (fig. 45H–I). ‘‘Sheridan beds’’ (late Irvingtonian), Hay Springs area, Sheridan County, Nebraska.

Etymology: Latin: fenem, hay, with reference to the type locality.

Referred Material: From the type locality: F:AM 95313, right femur missing its proximal end is tentatively referred.

Distribution: Known only from the late Irvingtonian of western Nebraska.

Diagnosis: Differs from C. thöoides and C. aureus in the following features: upper and lower premolars very slender for their length; P4 small relative to M1, slenderer especially across the metacone; M1 with weak parastyle and preparacrista; ramus slender, but deep; p4 with well-defined second posterior cusp; m2 has strong paraconid, labial cingulum surrounds protoconid and ends at hypoconid.

Description and Comparison: In addition to the features noted in the diagnosis, the maxillary fragment, which seems little distorted, rises nearly vertically above P4; the external margin of the slitlike infraorbital foramen slants anteriorly, and the anterior root of the zygoma lies opposite the junction of M1–M2 as in C. aureus . The posterior end of the palate lies at the level of the midline of M2 as in C. aureus .

The P3 has most of the crown broken away, but there is no cingular cuspule or evidence of the posterior cusp present in C. aureus . In addition to the relatively small size and slender form of the P4, the protocone is also small with a conical tip. Breakage across the anterior end of this tooth does not hide the angled anterolabial border or the fact that the protocone just protruded beyond the anterior border of the paracone as in C. thöoides but unlike in C. aureus .

The paracone of M1 is noticeably taller and larger than the metacone, and both are somewhat laterally compressed cones. There is a weak parastyle and corresponding preparacrista, but these are not as well developed as in Eucyon davisi . The hypocone is large and connected by the lingual cingulum with the metaconule. Anteriorly, the lingual cingulum is well defined across the protocone where it joins the preprotocrista at the base of the paracone. The paracone and metacone of M2 are low and of nearly equal size, surrounded labially by a shelflike cingulum without stylar cusps. A strong lingual cingulum and hypoconal shelf are prominent and, although the protocone retains a postprotocrista, the metaconule is absent. All of these features are also those of C. thöoides and C. aureus .

The horizontal ramus is deep for a small animal with such slender teeth. The premolars are not strongly graded in length backward as they are in C. aureus , but more as in C. thöoides , and the p2 is set off with slightly longer diastema as in C. thöoides and older individuals of C. aureus . A strong anterior mental foramen occurs beneath p1 and smaller foramen beneath the anterior root of p3. Only the m2 remains of the molar row and it is distinguished by retention of a paraconid and the posteriorly extended labial cingulum as indicated in the diagnosis. The metaconid was similar in size and only slightly taller than the protoconid, and there is a well-defined hypoconid and a beaded crest on the lingual part of the talonid without a clearly defined entoconid.

Discussion: Canis feneus is nearly a contemporary of C. cedazoensis yet it shows morphological traits that represent a more mesocarnivorous, or perhaps even hypocarnivorous, adaptation. If these animals, known only from the fragmentary remains of single individuals, are correctly diagnosed, their presence suggests that a previously unrevealed diversity of jackal-like canids occupied North America in the early and medial Pleistocene.

Canis cedazoensis Mooser and Dalquest, 1975 Figure 45D–E; appendix 3

Canis cedazoensis Mooser and Dalquest, 1975: 787 , fig. 2.

Canis cedazoensis: Nowak, 1979: 68 .

Type: TMM 41536-41 (originally O. Mooser collection FC 634, AMNH cast 105201), right partial maxilla with P3–M1 and M2 alveolus (fig. 45D–E) form strata referred to the Tacubaya Formation in Arroyo Cedazo (early Rancholabrean), 3 km southeast of Aquascalientes, State of Aquascalientes, Mexico.

Distribution: Only known from the Rancholabrean of Mexico. Attributed to the Sangamon interglacial (stage 5 as determined by 18 O) by Pinsof (1996).

Revised Diagnosis: Canis cedazoensis differs from C. aureus and C. thöoides in M1 anteroposteriorly shorter relative to length of P4 with reduced hypocone and labial cingula; M2 alveoli indicate smaller size tooth relative to M1.

Description and Comparison: Mooser and Dalquest (1975: 787, fig. 2) reported that this taxon is intermediate in size to the coyote and fox, and that it may have filled the niche occupied in the Old World by the jackals. They compared the type maxilla with nearly 100 Recent coyote skulls from Texas and found that some of the coyotes had the upper carnassial as short, but never as narrow, and ‘‘no coyote had the M1 as small as that of C. cedazoensis .’’ Nowak (1979: 68) followed Mooser and Dalquest in allocating this species to Canis , but because the type was fragmentary and worn, he did not evaluate the taxon.

No comparisons were previously made with the living jackals, but we have found that C. cedazoensis is close in size and aspects of morphology to C. aureus , although differing from it in significant ways. Like C. aureus , the P3 has a weak posterior cusp. The P4 is proportionally similar and has a relatively larger protocone that, as in C. aureus , barely protrudes beyond the anterior end of the tooth. The greatest difference lies in the relative size and form of M1 and M2 (as inferred from its alveoli). The M1 is small relative to P4 taking into account wear and missing enamel on the labial and lingual sides of the tooth. Nevertheless, the labial cingulum is not as well defined as in C. aureus , nor is there an anterior cingulum lingually. The hypocone and associated cingulum are reduced and confined posterior to the protocone. Although there is considerable wear on the principal cusps, the paracone is clearly larger than the metacone as in C. aureus . The alveoli for the paracone and metacone of M2 lie in a line nearly transverse to the sagittal plane, indicating that the M2, like the M1, was transversely wide for its length, and likely had a much reduced metacone and hypocone. The interdental facet for M2 on M1 is mostly lingual to the M1 metastyle, also suggesting a small tooth situated more lingually than in C. aureus .

The morphology of C. cedazoensis suggests a jackal-like canid with dental tendencies toward more hypercarnivorous adaptations than seen in any jackal or in the two other North American jackal-like species ( C. thöoides and C. feneus ) that have been recognized in the course of these studies.

Canis edwardii Gazin, 1942 Figures 39H–O, 40, 43, 44, 46A–H, 47A–F, 48A–F, 50–52; appendices 2–4

Canis edwardii Gazin, 1942: 499 , fig. 41.

Canis cf. C. lupus: Hibbard and Dalquest, 1966: 20 .

Canis priscolatrans Cope (in part): Kurtén, 1974: 7, 11.

Canis edwardii: Nowak, 1979: 82 .

Type: NMNH 12862, partial skull with I1– M2 (fig. 46E–F) and incomplete rami with i1–m2 (fig. 46G–H) from about 3.2 km northeast of Curtis Ranch homestead, Saint David Formation (earliest Irvingtonian), San Pedro Valley, Cochise County, Arizona.

Referred Material: Curtis Flats, Saint David Formation (earliest Irvingtonian), Cochise County, Arizona: F:AM 67301, right partial ramus with m1 (broken)–m2; F:AM 103409, left ramal fragment with p3–p4 (both broken); NMNH 12864, right partial maxilla with P3–M2 from 4.8 km northeast of Curtis Ranch; AMNH 21806*, left ramus with c– m3 alveoli; and AMNH 23393*, badly broken left ramus with c–m1.

Tusker fauna, 111 Ranch at Dry Mountain locality (late Blancan), near Safford, Graham County, Arizona: F:AM 63100, skull with I1–M2 (I2 alveolus) (fig. 48A–C), right partial ramus with c (root broken)–m3 (alveolus) (fig. 48D–F), and associated skeletal elements, including right partial humerus (fig. 39H), both radii (fig. 39I) and ulnae (fig. 39J–K), metacarpals II–V (metacarpal III, fig. 39L), right femur (fig. 39N), both tibae (fig. 39O), metatarsals III–V (metatarsal III, fig. 39M), calcaneum, phalanges, and vertebrae.

Whitlock Mountains ( UA locality 9806), south of Dry Mountain (late Blancan), near Safford, Graham County, Arizona: UA 23402, crushed skull (restored) and mandible showing most dentition except I1–I2, M2, i1–i3, c, m2, associated partial cervical vertebrae 3–7, fragment of left scapula, clavicle, first rib, and proximal part of right ulna.

Stagomastodon locality, Saint David Formation (latest Blancan), Curtis Ranch, Cochise County, Arizona: UA 1632, left ramal fragment with p2–p3, m1, and p4 (broken); and UA 3231, left ramal fragment with m1, m2 broken, and m3.

Cita Canyon beds (late Blancan), Randall County, Texas: WTUC 1936, fragment of right frontal, isolated right C, and right and left partial maxillae with P4–M2; CWT 2574, radius; CWT 2526, tibia; and CWT 2572, metatarsal III.

Sand Draw Quarry, Keim Formation (medial Blancan), north of Ainsworth, Brown County, Nebraska: F:AM 95189*, right fragmentary maxilla with P3.

Big Spring Local Fauna, Long Pine Formation (late Blancan), UNSM locality Ap 103, Antelope County, Nebraska: UNSM 2136-78, isolated left M1.

Haile 12B (late Blancan fide Hulbert and Morgan, 1993: table 8.1), Alachua County, Florida: UF 11516, partial skull consisting of right maxillary, jugal, anterior part of squamosal with glenoid fossa and anterior part of ectotympanic, frontal shield, anterior part of parietal including sagittal crest to interparietal, P4, M1–M2. This specimen was attributed to C. rufus by Nowak (1979: 88), who did not list the specimen number and who incorrectly listed the site as Haile 7A and the age as Rancholabrean. It subsequently was attributed to C. edwardii (his C. priscolatrans ) by Nowak (2002: 107), who then correctly listed the site as Haile 12B and the age as late Blancan, but who incorrectly listed the site ( Nowak, 2002: 127) as Haile 12A.

Reigle Gravel Pit, west of Plainview, north of Highway 20, Antelope County, Nebraska: UNSM locality Ap-133, unnamed early Irvingtonian gravel deposits of glacial origin overlying the Blancan Long Pine Formation: UNSM 2555-90, associated left lower teeth, p3–p4, m1 found together in situ.

Waneta site, near McCook, Chase County, Nebraska, UNSM locality Ch-102, late early Irvingtonian crystalline gravels incised into the Ogallala Group that appear to be continuous with similar beds along the Republican River, including the site near Orleans in Harlan County, Nebraska, type area of the Sappa Ash (equivalent to Mesa Falls Ash of Yellowstone, 1.2 Ma): UNSM 48590, right ramus with c–p4 alveoli, m1–m2, m3 alveolus including ascending ramus.

Williams site, Hitchcock County, Nebraska, UNSM locality Hk-113, stratigraphy and late early Irvingtonian age as for UNSM localities Ch-102 and Rw-105: UNSM 3011- 97, labial half of crown of right m1.

McCook Clam site, Red Willow County, Nebraska, UNSM locality Rw-105, stratigraphy and late Irvingtonian age as for UNSM localities Ch-102 and Uk-113: UNSM 88523, left ramus with partial alveolus for p3, p4 alveolus, m1 broken, m2, m3 alveolus; UNSM 88524, left maxillary fragment with P3–P4, M1–M2; and UNSM unnumbered, right m2.

Mike Jones site, UNSM locality Fn-104, Republican River near Cambridge, Furnace County, Nebraska, late early Irvingtonian gravels incised into the Ogallala Group: UNSM 88527, right ramus, p1 root, p2–p3, p4 broken, m1–m2; UNSM 5357-92, m2.

Anita Fauna (early Irvingtonian), fissure fill in Kaibab Limestone, Coconino County, Arizona: NMNH 10210B, left ramus with p1 alveolus, p2–p3 broken, p4–m1–m2, m3 alveolus (fig. 47A–B); NMNH 10210C*, left ramus with p4.

Vallecito Creek Fauna, Palm Springs Formation (early Irvingtonian), San Diego County, California: LACM 20591, isolated C, anterior part of the left P4 broken and right P4 (possibly two individuals); LACM 20591, isolated C, anterior part of left P4 broken and right P4 (possibly two individuals); LACM 6250, right and left maxillae with P3 (broken)–M2 and right and left rami with i1–p4 broken and alveoli of m1–m3; LACM 8235, right and left partial rami with c broken–p1, p1–m1 broken, and m2; LACM 6236, left partial ramus with p3–m2 broken; LACM 6770, right maxilla fragment with M1 broken–M2; LACM 3258, right and left partial rami with c broken, p1–p4 alveoli, p4–m1 broken, and m2 alveolus; and LACM 3805, metatarsals II, III, and IV.

Seventeen Palms (Irvingtonian), Palm Springs Formation, Imperial County, Cali- fornia: F:AM 31853, left partial maxilla with P4–M1.

Mecca Hills (Irvingtonian), Palm Springs Formation, Coachella Valley, Riverside County, California: LACM 1148, left partial maxilla with P3 (broken)–M2 (badly worn).

El Casco Local Fauna (early Irvingtonian, 1.3–1.4 Ma; Albright, 1999), Riverside County, California: F:AM 17863, right partial ramus with c (alveolus)–m1 (broken) (p1 and p4 roots); F:AM 17864, right partial ramus with m1 (broken)–m2 root; F:AM 17865, left partial ramus with c (broken)–p1–p3 broken, p4–m1.

USGS locality M1367, Bruneau Formation (late Irvingtonian), Owyhee County, Idaho: NMNH 184126, left partial ramus with p4 broken and m1–m3.

Rome, USGS locality M1080 (Irvingtonian), Malheur County, southeast Oregon: NMNH 23898, skull with I1–M2 (P1 missing) (fig. 46A–D).

Gilliland Local Fauna (earliest Irvingtonian), lower part of the Seymour Formation, 11 km west of Vera, Knox County, Texas: UMMP 46483*, articulated partial maxillae and premaxillae with I1–P2 (P3 broken) and both partial rami with i1–p4; and UMMP 46460, cranial fragment including the left interparietal, parietal and both temporal bones and incomplete glenoid area, broken bulla, and base of paroccipital process; MSU 8676*, right edentulous partial ramus with alveoli for c–m4; and UMMP 46479*, left upper canine.

Arkalon Gravel Pit (late Irvingtonian), Crooked Creek Formation, just above Lava Creek B (Perlette Type O) tephra (0.61 Ma), 2.4 km east of Arkalon Station of the Rock Island Railroad, Seward County, Kansas: UMMP V29068 , associated limbs of a single individual, radius, femur, tibia, calcaneum, metatarsals III and V ( Hibbard, 1953).

Inglis site 1A (earliest Irvingtonian, sensu Morgan and Hulbert, 1995), Citrus County, Florida: UF 19323, right partial ramus with p4–m1 and alveoli for p1–p3, m2–m3 (fig. 47E–F); UF 19324, right partial ramus with m1 (talonid)–m2; UF 19404, left isolat- ed m1; UF 18050, left p3; UF 18049, right P4; UF 18046, right partial maxilla with P4– M1, M2 (alveolus) (fig. 47C–D); UF 67846*, left partial maxilla with M1; UF 19405, left M1; UF 19406, left M1; UF 18051*, left premaxilla with I1–I2 alveoli and I3; UF 18052*, right lower canine; UF 18047, right partial maxilla with M2; UF 18048, left M2 and a right M2 germ (two individuals); UF 18054, atlas; UF 18056, left radius; UF 67845A, right radius (smaller individual); UF 22566, left proximal part radius; UF 22567, right distal end of radius; UF 67843, left partial ulna; UF 18055A, left partial ulna; UF 67844B, right partial ulna; UF 22568, left partial ulna; UF 22569, left partial ulna; UF 18057, left metacarpal III; UF 22570, left metacarpal IV; UF 18057A, left metacarpal IV; UF 45457, left metatarsal V; UF 45458, left metatarsal V; UF 45459, left metatarsal V; UF 22571, distal end left tibia; UF 22572, distal end left of tibia; UF 22573– 22576, four astragali; UF 18059, left calcaneum; UF 22577, right metatarsal III; UF 22578, left metatarsal IV; UF 45457, left metatarsal V; UF 45458, left metatarsal V; and UF 45459, left metatarsal V. Most of the dental remains listed above were attributed to C. rufus by Nowak (1979: 88), and later to C. edwardii by Nowak (his C. priscolatrans, 2002: 127 –128).

Rigby Shell Pit (late early Irvingtonian), Bermont Formation, Sarasota County, Florida: UF 40090, right maxillary fragment with P4 and well-worn M1–M2; UF 40091, fragment of right ramus with p1 root, p2, p3 root, p4 broken, m1–m3 alveoli.

Wolf Hill locality, Bone Valley (late Irvingtonian), Polk County, Florida: TRO 1441, skull fragments and detached teeth with I1–M1 and M2 (broken).

Leisey Shell Pits, Hillsborough County (early Irvingtonian ), Florida: UF 81665 , maxillary fragment with P4, M1–M2 (figured by Berta, 1995: fig. 1C) ; UF 81666 , right maxillary fragment with M1–M2 ; UF 81664 , right maxillary fragment with P3 ; UF 81663 , left maxillary fragment with P1–P2 ; UF 124531, right maxillary fragment with M1 ; UF 63667 , left ramus with c alveolus, p1, p2 broken, p3–p4, m1 ; UF 64399 , left ramus with m1, m2–m3 alveoli ; UF 81670 , right ramus with p3–p4, m1 ; UF 84752 , right m1, UF 87285 , right m1 ; UF 81689 , right m2 ; UF 87297 ; left m2; UF 81672 , left m1 ; UF 81662 , left P4 ; UF 81668 , right P4 ; UF 81669 , right M1 ; UF 81657 , right M1 ; UF 81657 , right M1 ; UF 81661 *, metastyle of left P4 ; UF 81658 , right p4 ; and UF 81659 , right P4 (the last four specimens listed as C. armbrusteri by Berta, 1995: 465) ; and postcranial elements listed by Berta (1995: 466).

Haile 21A (Newberry Quarry), Alachua County (late early Irvingtonian sensu Morgan and Hulbert, 1995), Florida: UF 62563 (cast), left ramus with p3–p4, m1–m2, m3 alveolus; UF 62564 (cast), left ramus with p1–p4 alveoli, m1–m2, m3 alveolus; UF 62562 (cast), right ramus with i1–i3, c, p1 alveoli, p2–p4, m1–m2, m3 alveolus; UF 62568, right ramus with m1 talonid, m2; UF 63175, left ramus with p1–p4, m1–m2; UF 63174, edentulous partial right ramus, alveoli for i1–i3, c, p1–p4; UF 124539, left p4; UF 63527, fragment of right ramus, p4 roots, m1 broken, m2 alveolus; UF 63311, right p4; UF 62567, right m1; UF 62561 (cast), fragment of right premaxillary, maxillary with I1 root, I2 alveolus, I3, C, P1 and associated left maxillary with C, P1–P2 alveoli, P3–P4, M1–M2; UF 124537, left P4; UF 63159, left radius; UF 62588, left humerus lacking distal end; UF 63162, left metacarpal II; UF 63309, left metacarpal V; UF 62585, associated left metatarsals II–IV; UF 62584, right metatarsal IV; UF 63526, left metatarsal II; UF 63098, left tibia; and UF 63100 left femur.

Crystal River Power Plant, UF locality Citrus 8 (late early Irvingtonian sensu Morgan and Hulbert, 1995), Citrus County, Florida: UF 17074, right maxillary fragment with P3 alveolus, P4, M1, M2 alveolus. Attributed to C. rufus by Nowak (1979: 88) and to C. edwardii (his C. priscolatrans ) by Nowak (2002: 128).

El Golfo de Santa Clara , northwestern Sonora, Mexico (late Blancan): IGM 10163* ( HJG 793 ), right maxillary fragment with P4, M1–M2, and associated left maxillary fragment with P4 ; IGM 10047* ( HJG 777 ), edentulous left ramal fragment with c, p1– p4, and anterior part of m1 alveoli .

Distribution: Medial Blancan of Nebraska; late Blancan of northern Mexico, Arizona, Texas, Nebraska, and Florida; and early to late Irvingtonian of Texas, Kansas, Nebraska, Oregon, Idaho, Arizona, California, and Florida.

Revised Diagnosis: Canis edwardii is distinguished from C. ferox by a suite of synapomorphies the former shares with C. latrans and the wolves: tip of p3 principal cusp lies below those of adjacent premolars and its cingulum lies entirely below that of p4; principal cusp of p4 lies below paraconid of m1, anterior face of m1 slants posteriorly; paracone of M1 is larger than metacone; maxillary-jugal suture is acute; m3 is singlecusped; and incisive foramen may extend posterior to canine alveolus.

Canis edwardii is larger than C. latrans and differs in skull and some tooth proportions as well as lacking two synapomorphies that group C. latrans with the wolves: elongation of distal part of forelimb (radius/tibia ratio,90%), and sagittal crest extends onto frontal bone.

Description and Comparison: Among the diverse materials listed above that we ascribe to C. edwardii , there are five useful skulls: two of which are from late Blancan deposits of southeastern Arizona, the type from slightly younger rocks (earliest Irvingtonian) not far away in southern Arizona, a fourth from late Blancan deposits of Florida, and the fifth from Irvingtonian deposits in southeastern Oregon. In cranial proportions these specimens show considerable diversity, but that is to be expected from their temporal (about 2 m. y.) and geographic range, as well as from sexual and ontogenetic variation (although all are adults). Compared to C. latrans , these animals are larger in all dimensions and the proportional relationships of the cranial elements are similar (fig. 43). We found no systematic changes in these proportions within the time range covered by the five skulls.

Based on all the referred specimens, C. edwardii is characterized morphologically as follows: the incisive foramen is long and extends posterior to the canine alveolus; the infraorbital foramen is slitlike and opens above the posterior root of P3; the jugalmaxillary suture is acute; the nasals extend behind the most posterior extent of the maxillary-frontal suture; the frontal sinus is expanded with modest inflation of the postorbital processes and associated parasagittal crests, extending nearly to the frontoparietal suture; the parasagittal crests join to form a low sagittal crest at the frontoparietal suture; posteriorly, the sagittal crest gains in height onto the interparietal; the broad shield formed by the nuchal crests is a dorsally truncated triangle when viewed from the rear; the mastoid shelf and process are not strongly developed; the paroccipital processes are vertically directed and have short freetips, the processes are transversely narrow and cover only a small part of the posterior surface of the bulla; the bullae are moderately inflated and may just extend anterior to the postglenoid process; the basioccipital is narrow and parallel-sided; the palate extends to just posterior to or opposite the M2; the largest posterior palatal foramen lies on the maxillary-palatine suture opposite the paracone of P4 or farther behind at the posterior end of that tooth.

The upper incisors are large, I1–I2 have lateral and medial cusps, and I3 is markedly larger and has no subsidiary cusps, only a well-developed median cingulum. The upper premolars are large, elongate, low-crowned, and closely spaced, and the only diastemata of any length separates P1 from C and P2. The P3 frequently has a small posterior cusp, but the Rome specimen ( NMNH 23898) shows a P2 with a tiny posterior cusp as well. The P4 is slender, with its protocone prominent and anterolingually directed, but it does not extend beyond the anterior border of the tooth. The M1 has a paracone that is larger than the metacone but not markedly enlarged, the protocone is situated opposite the paracone, the protoconule is barely differentiated, and the metaconule a little more so, but not prominent, so that the postprotocrista is only slightly enlarged at its position. There is a strong hypocone situated posterolingually connected to the lingual cingulum, which passes across the protocone to the parastyle. Posteriorly, the lingual cingulum ends at the position of the metaconule. A strong labial cingulum is present, terminating anteriorly in a low parastyle to which the preparacrista is attached. The M2 has the paracone larger than the metacone, as well as a shelflike labial cingulum. A postprotocrista passes to the posterior border, but the metaconule is very small or lacking. The lingual cingulum and hypocone form an internal shelf. The shape of this tooth is usually oval but may be markedly attenuated lingually by posterolingual expansion of the hypocone (e.g., NMNH 23898, fig. 46B).

The holotype is the only specimen that shows the lower incisors, all of which have lateral cusps; the canine is large and recurved. The lower premolars are large, elongate, and low-crowned, as are their upper counterparts. They are uncrowded and p2 may be separat- ed by diastemata. There is no inflection of the premolar row as present in the wolves. The p2 lacks a posterior cusp, but p3–p4 have such cusps, although other examples show that they may be absent on p3. The tip of the principal cusp of p3 lies below p2 and p4; its posterior cingulum lies below the base of the crown of p4. The p4 consistently has a tiny second posterior cusp just anterior to the posterior cingulum. The tip of the principal cusp of the p4 lies below the tip of the paraconid of m1. The anterior face of the m1 paraconid slants or curves backward and the trigonid is relatively open. The talonid is close to the same width as the trigonid; its hypoconid is prominent, laterally situated; the entoconid, markedly smaller, lies somewhat posterior to the hypoconid and is connected to it by a low cristid. The talonid may be open lingually between the entoconid and metaconid or the gap closed by a low entoconulid. A low cingulum is often found on the labial side of the talonid. A talonid shelf is consistently present, and a hypoconulid may be present behind the hypoconid. The m2 is relatively large with an anterolabial cingulum and a metaconid that is smaller than the protoconid and slightly posteriorly placed relative to the protoconid. The protoconid has a paracristid that extends to the anterior cingulum, although there is no paraconid. The talonid may be wide with a prominent hypoconid and lingual crest studded with small cuspules but no differentiated entoconid, or it may be attenuated posteriorly and consist only of a hypoconid. The m3 is preserved in only a few specimens. It is a single-cusped tooth with a lingual crest passing from this cusp to the lingual border, but no metaconid is differentiated along this crest.

The horizontal ramus is relatively shallow, but it does not markedly decrease in depth anteriorly. The largest mental foramen lies beneath p1 or between p1 and p2; a smaller foramen lies between the roots of p3. There is a shallow inflection of the lower border of the ramus behind the symphysis. The masseteric fossa is deep, especially anteroventrally and against the anterior rim, and a masseteric line is developed. The coronoid process is shorter than high, corresponding to the short temporal fossa of the skull. As in other species of Canis , the medial side of the angular process shows a large fossa for the superior ramus of the median pterygoid muscle and below it a smaller roughened area for the inferior ramus. The process is markedly attenuated posteriorly and ends in an upturned hook.

We have unassociated limbs referred to C. edwardii from four sites: Cita Canyon, Texas, and Leisey Shell Pit, Haile 21A, and Inglis 1A in Florida. Two important associated skeletons are from the late Blancan of Arizona (F:AM 63100, fig. 39H–O) and the late Irvingtonian Arkalon site in southwestern Kansas ( UMMP V 29068 , fig. 52). They represent the extremes in size and sample nearly the limits of the geological range of this species. The Kansas specimen was discussed by Kurtén (1974), and although unaccompanied by cranial or dental remains, seems properly referred to C. edwardii on the basis of similarity of its elements to those from other sites where dentitions are available. As Kurtén (1974) has pointed out, the Arkalon skeleton of C. edwardii ( C. priscolatrans in his taxonomy) is a robust form as befits its overall size, and the allometry follows the pattern of C. lepophagus , as exemplified by the relationship of the proximal width of the radius to its total length ( Kurtén, 1974: fig. 6). The mutual proportions of the forelimb elements (radius and metacarpal III) show that the forelimb of the Arkalon individual of C. edwardii was elongated as in the living coyotes. The radius lies below the range of C. latrans relative to the tibia (radius/tibia ratio,90%). In particular, the smaller Arizona specimen (F: AM 63100) shows a significantly lower radius/ tibia ratio (85%). Both specimens, however, differ little from the hindlimb proportions of C. latrans . The larger radius in the Arkalon individual hints at changes in limb proportions within C. edwardii during its long geological range.

Discussion: Cope (1899: 227–228) described Canis priscolatrans from three ‘‘superior molars of a single individual’’ (all ANSP 57), a right M1 and M 2 in early wear, and a right P4 (not a ‘‘pm1’’ as he designated it) also in early wear. These were obtained from the Port Kennedy ‘‘bone deposit’’ in upper Merion Township, Montgomery County, Pennsylvania, and are of late Irvingtonian age.

The cotypes of Canis priscolatrans have dimensions falling inside those of C. edwardii , and the ‘‘conules in the M1 are not distinct’’ as noted by Cope (1899: 227), a feature that is alluded to in his diagnosis ( Cope, 1899: 228), both of which agree with the condition in C. edwardii . However, we set aside this possible synonymy on the basis of the very incomplete knowledge of the taxon C. priscolatrans afforded by the cotypes in comparison to the type of C. edwardii and hold the former as a nomen dubium.

Canis latrans Say, 1823 Figures 40, 43, 44, 49A–H, 50–52; appendices 2–4

Canis cf. priscolatrans Cope (in part): Gidley and Gazin, 1938: 23.

Canis irvingtonensis Savage, 1951: 231 , figs. 8a–b, 9a–b, 10.

Canis priscolatrans Cope (in part): Kurtén, 1974: 6, 11.

Referred Pre-Rancholabrean Material: UCMP locality V3604, Irvington site 2 (Irvingtonian, late Matuyama Chron, older than 0.73 Ma), Bell Sand and Gravel Co. Quarry, 1.3 km southwest of Irvington, Alameda County , California: UCMP 38805 , right partial ramus with p1 (alveolus) p2–m2, m3 alveolus (type of C. irvingtonensis ) ; UCMP 38748 , left ramus with i1–p1 alveoli and p2–p3 (alveolus) ; UCMP 38804 , left radius (paratype of C. irvingtonensis ) ; UCMP 81737 *, part of left M1 ; UCMP 56090 , cranium lacking inion, condyles, bullae, and tips of postorbital processes .

Fairmead Landfill, UCMP site V93128 , Turlock Lake Formation (late Irvingtonian, Dundas et al., 1996), Madera County, California: UCMP 140413, left maxillary fragment with M1–M2, and posterior part of frontal bone.

‘‘Sheridan beds’’ (late Irvingtonian), Rushville and Gordon area, Sheridan County, Nebraska: From UNSM locality Sh-3, Rushville Quarry 1: UNSM 2913, right partial ramus with p1–p4 alveoli and m1–m2 (broken); from UNSM locality Sh-5, Gordon Quarry 1, Pit 3: UNSM 21435, right ramal fragment with i3–m1; from UNSM locality Sh-5: UNSM 21437, left partial ramus with i1–p1 alveoli, p2–p3 and p4 (alveolus)–m1 (broken); from UNSM locality Sh-4, Quarry 4B: UNSM 451-76, fragment of left ramus with p1 alveolus, p2 roots, p3–p4; from UNSM locality Sh-4, Quarry 4B: UNSM 452-76, right m1; from UNSM locality Sh-3: UNSM 6.3.8.33 NP, left m1; from UNSM locality Sh-5: UNSM 21434, left maxillary fragment P3 alveolus, P4; from UNSM locality Sh-5: UNSM 15250-38, left P4; from locality Sh-4, Quarry 4B: UNSM unnumbered, left M1.

Rock County Gravel Pit (Irvingtonian), Rock County, Nebraska: F:AM 87248, left ramal fragment with p4 (alveolus)–m1 (broken).

Mullen Pt 1 (late Irvingtonian, see Martin, 1972, for discussion of the faunal composition and age of the Mullen sites), UNSM locality Cr-10, Cherry County, Nebraska: UNSM 26115, right ramus with c–m2 and m3 alveolus (fig. 49C–D).

Conard Fissure (late Irvingtonian), 24 km south of Harrison, Newton County, Arkansas: AMNH 12394, a left unerupted m1.

Rock Creek, Tule Formation (late Irvingtonian ), 14.5 km southwest of Silverton , Briscoe County, Texas: JWT 2303, right partial maxilla with M1 (paracone broken away)–M2 ; F:AM 95314, left humerus; F:AM 95315, right metatarsal IV; and YPM 10080, right M1 .

Slaton Quarry Local Fauna (late Irvingtonian, 0.5–0.6 Ma, Dalquest and Schultz, 1992), west side of Yellowhouse Canyon, 7.6 km northeast of Slaton, Lubbock County, Texas: MSU 6522, maxillary fragment with posterior half P3, P4–M1; MSU 5043, right ramal fragment with canine alveolus, p2–p3.

Cumberland Cave (late Irvingtonian), 6.5 km northwest of Cumberland, Allegany County, Maryland: NMNH 7660, partial skull (fig. 49A) with I1–M2 represented by alveoli, roots, or broken teeth; NMNH 494384, right ramus, c–p2 alveoli, p3–p4, m1–m2, m3 alveolus; NMNH 494385, right M1.

Hamilton Cave (late Irvingtonian, 0.74– 0.85 Ma), Pendleton County, West Virginia: NMNH 336123, left M1 (between Cheeta Room and Smilodon 2 site); NMNH 494386, left metacarpal III; NMNH 494387, right astragalus; NMNH 494388, left m1 (near Smilodon site); NMNH 494389, right m1 (Smilodon 2 site); NMNH 494390, right m1 (Smilodon 2 site). Porcupine Cave (late Irvingtonian, 0.35–0.48 Ma), Park County, Colorado: Anderson (1996) listed material that resided in the CM, DMNH, and UCMP collections at that time. We examined CM 49121, fragment of left maxillary and M1 broken and M2 from the Badger Room. Recently obtained material in the DMNH collection (a skull, 30676) from the Badger Room confirms the identification of C. latrans there ( Bever, 2005).

Distribution: Late Irvingtonian of California, Colorado, Nebraska, Arkansas, Texas, West Virginia, and Maryland. Rancholabrean to present of North and Central American (see Nowak, 1979).

Discussion: Nowak (1979: 73–82) gave a much more comprehensive survey of the later Pleistocene coyotes than we attempt here. Our purpose in further describing and figuring some of the specimens listed above is to compare the earliest records of C. latrans -like forms with their contemporaries and immediate antecedents. We do not deal with the much larger Rancholabrean and Holocene records reviewed by Nowak (1979). All of our referred specimens fall within the size range of living C. latrans . They are smaller than the isolated teeth (ANSP 57) selected and figured by Cope (1899: fig. 3, 3d, and 3e) as the cotypes of C. priscolatrans .

Savage (1951: 231) based C. irvingtonensis on a mandible with a deep horizontal ramus and closely spaced, broad, and massive premolars. A radius was also chosen as a paratype, and from this evidence Savage ‘‘visualized that this shorter-jawed, shorterlimbed, stockier animal from Irvington [was] a coyote with habitus tending to be somewhat different from known coyotes.’’ Nowak (1979: 73), however, found that the tooth and mandibular measurements of the type fell within the range of variation of late Pleistocene and Recent specimens of C. latrans , and concluded that it warranted no more than subspecific distinction. The radius, however, appears proximally wide for its length, unlike living coyotes ( Kurtén, 1974: fig. 6).

A coyotelike skull (NMNH 7660, fig. 49A) from the late Irvingtonian deposits of Cumberland Cave in Maryland, referred to C. priscolatrans by Gidley and Gazin (1938: 23) and C. latrans by Nowak (1979: 80), has a smaller dentition, as judged by the roots, than either C. priscolatrans or C. edwardii . Gidley and Gazin pointed out that this skull is close in size to that of C. latrans but slightly more robust, particularly in the frontal region. The nasals are short, as in most C. latrans , not reaching beyond the maxillaryfrontal suture. The zygomatic arch is very robust and measures 15.5 mm at the shallowest point anterior to the postorbital process. A broad scar for the masseter muscle is present, and the jugal-maxillary suture forms an acute angle at the base of the arch. The height of the maxilla between the toothrow and the orbit is 29.3 mm. The frontal shield is about 50.4 mm wide and moderately inflated. Especially noticeable is the expand- ed braincase (59.7 mm wide at the parietosquamosal suture) that is markedly wide anterior to the frontoparietal suture as in coyotes. At the point of the greatest postorbital constriction the skull measures 38.7 mm. The short frontal crests join at the frontoparietal suture to form a strong sagittal crest rather than in front of this suture as in most C. latrans . The supraoccipital shield is broad and the inion is less narrowed than in C. latrans . Some of the above features of the Cumberland Cave skull that differ from C. latrans , including the strong sagittal crest and the broad supraoc- cipital shield, are primitive characters. Otherwise, the Cumberland Cave individual, although estimated to be a young adult by Nowak (1979: 80), has a large skull with an expanded braincase that closely resembles that of C. latrans .

Recently obtained material from Cumberland Cave, Maryland, includes a ramus with p3–p4, m1–m2 (NMNH 494384), and an isolated M1 (NMNH 494385) whose dimensions clearly fall into the range of C. latrans . The p3 is low with a posterior cusp. The p4 has a second posterior cusp independent of the posterior cingulum. The first posterior cusp lies in line with the principal cusp and the second cusp, rather than labially as in many C. latrans . The m1 entoconid is situated at the posterolingual corner of the talonid, oblique to the hypoconid. These cusps are united by cristids. The talonid basin is closed lingually by the entoconulid crest. The m2 has a well-developed anterolabial cingulum that passes posteriorly across the protocone on to the talonid. The metaconid of m2 is only slightly oblique to the protoconid and smaller than the latter. Depth of the ramus below m1–m2 is 20.0 mm. The isolated M1 is about the right size for the skull (NMNH 7660). Morphologically it closely resembles that of C. latrans , including the anteriorly directed preparacrista that lies medial to the parastyle.

Somewhat older (early late Irvingtonian) materials from Hamilton Cave, West Virginia, include four m1s more robust than that in the ramus just described from Cumberland Cave and rather like the type of C. irvingtonensis and the sample from late Irvingtonian ‘‘Sheridan beds’’ exposed along the Niobrara River in Sheridan County, northwest Nebraska. These lower carnassials have talonid basins that open lingually in front of the entoconid.

A maxilla fragment (JWT 2303), humerus, and metatarsal IV are listed here as C. latrans from the later Irvingtonian part of the Tule Formation at Rock Creek, Texas. The length and width of the broken M1 are estimated to be 11.0 and 13.5 mm, respectively; the M2 measures 6.0 and 9.3 mm. These measurements are slightly smaller than those of an M1 (YP 10080) from Rock Creek, which was figured as Canis ? priscolatrans by Troxell (1915: fig. 19). Troxell recorded the length and greatest transverse diameter of this M1 as 12.0 and 17.0 mm, respectively. Judging from Troxell’s figure, the greatest transverse diameter was probably taken obliquely from the paracone to the hypocone. The paracone in JWT 2303 is broken, but the transverse diameter would still be less than that of the M1 figured by Troxell. However, the measurements of the teeth and the limbs from Rock Creek, Texas, are within the range of those of C. latrans and smaller than those of the other two larger taxa (‘‘ C. dirus ’’ and ‘‘ C. ’’ texanus ) that Troxell recognized at this locality.

The referred material includes a late Irvingtonian ramus (UNSM 26115, fig. 49C–D) from near Mullen, Nebraska, which was referred by Kurtén (1974: 7) to C. priscolatrans . Kurtén also referred the types of both C. irvingtonensis and C. edwardii to C. priscolatrans . These types are both larger and the premolars are more closely spaced and more robust than those in the Mullen jaw. The Mullen ramus shows a few derived characters typical of coyotes: premolars separated by diastemata; p2 isolated by longer diastemata than other lower premolars; p4 crown lies below level of tip of m1 paraconid; anterior face of m1 paraconid slants backward; m1 hypoconid and entoconid connected by cristids; and m2 metaconid not significantly enlarged over the size of the protoconid. The lengths of the m1 and m2 (20.1 and 8.9 mm, respectively) of the Mullen jaw are similar to the Cumberland Cave specimen and within the lower part of the range of C. latrans . Additional measurements of the lower molars of Irvingtonian coyotes include an unerupted m1 (AMNH 12394, length, 20.02 mm; trigonid width, 7.2 mm; talonid width, 6.7 mm) from Conard Fissure, Arkansas, that indicates a small form, but within the range of samples of living coyote populations ( Nowak, 1979: appendix C).

Discussion: Although the evidence is fragmentary, there are convincing indications of the presence of a taxon lying within the morphological and metrical limits of living C. latrans as early as late Matuyama time (Irvington, California; Porcupine Cave, Colorado; Cumberland Cave, Pennsylvania; and Hamilton Cave, West Virginia). These early examples include robust individuals at the upper limits of the living coyote populations. Together they possess a number of cranial features typical of coyotes: the incisive foramina do not reach beyond the canine alveoli; the short nasals reach only to the most posterior limit of the frontal-maxillary suture; the short palate reaches only the m2 and does not extend posterior to it; the postorbital processes are not markedly inflated; the parasagittal crests are little inflated and join close to the frontoparietal suture; the cranium is inflated anteriorly (enlargement of the prorean gyrus of the neocortex), forming a prominent ‘‘shoulder’’ to the anterior part of the braincase in dorsal view, and consequently the postorbital constriction is short and broad; the frontal sinus does not reach the frontoparietal suture; the paroccipital process is nearly vertical, not markedly extended posteriorly, and has a short and laterally directed free-tip.

Canis armbrusteri Gidley, 1913 Figures 40, 43, 44, 52, 53A–D, 54A–F, 55A–D; appendices 2–4

Canis priscolatrans (in part): Cope, 1899: 227, fig. 3a, 3b, 3f.

Canis occidentalis?: Brown, 1908: 182 .

Canis armbrusteri: Patterson, 1932: 334 .

Canis priscolatrans (in part): Kurtén, 1974: 7.

Canis lupus: Martin, 1974: 71–77 , figs. 3.11, 3.12.

Canis armbrusteri: Nowak, 1979: 90 .

Type: NMNH 7662, left partial ramus with p4–m2, figured by Gidley, 1913: fig. 2, 2a (our fig. 53C–D) from Cumberland Cave (late Irvingtonian), Allegany County, Maryland.

Referred Material: From the type locality, Cumberland Cave (late Irvingtonian ), Allegany County, Maryland: NMNH 11886 , skull with P3–M2 (fig. 53A–B) ; NMNH 12888 , crushed skull with P2–M2 broken ; NMNH 7994 , posterior part of skull with P4–M2 ; NMNH 11887 , posterior part of skull with P3 (broken)–M2 and partial ramus with p1 (broken)–m2 (p3 and m1 broken) ; NMNH 11881 , anterior part of skull with C (alveolus)–M2 and left partial ramus with c (broken)–m2 (p2 alveolus). Gidley and Gazin (1938: 16–17) listed and measured: NMNH 11883 , partial skull with C–M2 (P2 alveolus) ; NMNH 11885 , partial skull with P1 and P4– M2 ; NMNH 8168 , ramus with c–m2 ; NMNH 8169 , ramus with p3–m3 ; NMNH 8172 , ramus with c–m2 (p1 missing) ; NMNH 10210 A*, left partial ramus with c, p2, p4, m2–m3 (both broken), and alveoli ; NMNH 10210 C*, left partial ramus with c–m1 (all broken or represented by alveoli) ; NMNH 4889 , left ulna ; NMNH 7995 , associated right humerus, radius, ulna, and metacarpal III ; NMNH 8167 , associated right tibia, left and right metatarsal III ; and unnumbered four humeri, an ulna, a radius, four metacarpal IIIs, three femora, and a tibia are present in the NMNH collection from Cumber- land Cave in 2002 (F. Grady, personal commun., NMNH). FMNH 14790 *, right M1–M2 (figured and described by Patterson, 1932) .

Leisey Shell Pit 1A, upper part of Bermont Formation (late early Irvingtonian sensu Morgan and Hulbert, 1995), Hillsborough County, Florida: UF 81654, left maxillary fragment with P3 alveolus, P4–M1, M2 alveolus (figured by Berta, 1995: fig. 1A); UF 81655, right maxillary fragment with M1–M2; UF 81656, right P4; UF 67091, right C; UF 87283, left ramal fragment with p3 alveolus, p4; UF 95647, right ramal fragment with p1–p3 alveoli, p4, m1 broken, m2; and UF 81660, left m1 broken.

Haile 21A, Newberry Quarry (late early Irvingtonian), Alachua County, Florida: UF 63623, left M1.

McLeod Lime Rock Mine, Smith Pit, pocket ‘‘A’’ (medial Irvingtonian sensu Morgan and Hulbert, 1995), Levy County, Florida: F:AM 67286, restored partial skull with C–P1 alveoli, P2–M2, and M3 alveolus (fig. 54A–B) and associated mandible with right c–p1 (alveoli), p2–m1, m2–m3 alveoli, left c alveolus, p1–p3 roots, p4, m1 broken, m2, m3 alveolus, both coronoid processes broken away (fig. 54C–D); F:AM 68017, left distal half of humerus; F:AM 68017D, right ulna; F:AM 68017E, proximal half of right ulna; F:AM 68017A, left radius; F:AM 68107B, proximal and distal ends of right radius; F:AM 68108B, left partial femur; F:AM 68108A, right femur; F:AM 68108C, proximal part of right tibia; F:AM 68108D, proximal and distal ends of left tibiae; F:AM 68020Q, right astragalus; F:AM 68020N, left calcaneum; F:AM 68020O, right calcaneum; F:AM 68020C, right metacarpal V; F:AM 68020D, metacarpal V: F:AM 68020, left metacarpal IV; F:AM 68020B, right proximal part of metacarpal IV; F:AM 68020F, right metatarsal II; F:AM 68020I, left metatarsal IV; F:AM 68020K, right partial metatarsal IV; F:AM 68020L, left metatarsal IV; F:AM 68020R–S, two first phalanges (most of the foregoing limbs may represent the same individual as the skull and mandible); F:AM 67291, right ramus with c alveolus, p1, and p2 alveolus–m3 (fig. 54E–F); F:AM 68018, right femur; and F:AM 68020P, right calcaneum.

Coleman site 2A (latest Irvingtonian sensu Morgan and Hulbert, 1995), Sumter County, Florida: UF 11519, partial skull with C alveolus–M2 (P1 alveolus), left premaxilla– partial maxilla with I1 broken alveolus–C broken alveolus and two detached upper canines (fig. 55A–B); UF 12120, fragment of right ramus, p2–p4 alveoli, m1, m2 alveolus; UF 11518, fragment of right ramus, m1–m2, m3 alveolus; UF 12114, two right p4; UF 12121, two right m1, right m2; UF 11520, right and left partial rami with c–p2 and p4–m3 (fig. 55C–D); UF 12125, right humerus; UF 12122, right tibia; and UF 12120, right metacarpals IV–V, left metatarsal IV. Described by Martin (1974) as Canis lupus .

Anita Fauna (early Irvingtonian), fissure fill in Kaibab Limestone, Coconino County, Arizona: NMNH 10201A*, left partial ramus with p4, m1 lost in life; AMNH 14360, left partial ramus with partial alveolus of m1 and m2–m3.

W.C. Stouts Ranch, Crooked Creek Formation (late Irvingtonian), 2 km southeast of Arkalon station on the Rock Island Railroad, Seward County, Kansas: F:AM 95181, crushed partial skull with M1–M2, and associated partial skeleton including both humeri, partial radius, both partial ulnae, partial femur, both partial tibiae, one with incomplete fibula, astragalus, calcaneum, incomplete metatarsals II–IV, phalanges, and vertebrae.

Rock Creek, Tule Formation (late Irvingtonian ), 14.5 km southwest of Silverton , Briscoe County, Texas: YPM 10079*, tibia, phalanx, rib, and P3 ( Troxell, 1915: 626– 634) ; TMM fragment of right premaxillary, right upper canine and associated right P3, P4 broken, and M1* described by Cope (1895: 453–454) from the W.F. Cummins collection (now in the Texas Memorial Museum) from the ‘‘ Equus horizon of the Tule Canyon on the Staked Plains of Texas,’’ and presumably from the Tule Formation at the above locality. JWT 678, right partial ramus with p1–p2, p3 broken–p4 alveolus, m1, and m2–m3 alveoli ; and JWT 688, right partial maxilla with P4 root, M1 broken, and M2 partial alveolus.

Conard Fissure (late Irvingtonian), Newton County, Arkansas: AMNH 11761, fragment of left ramus, alveoli for p3–m3; AMNH 11762, fragments of left and right astragali and right m2; AMNH 11762a, fragment of right P4 lacking protocone and anterior border; AMNH 90981, posterior cranial fragment, gnawed by rodents; AMNH 96633, distal end right humerus. Identified as Canis occidentalis ? by Brown (1908: 182).

Port Kennedy Cave (late Irvingtonian), Montgomery County, Pennsylvania: The following isolated teeth ( ANSP) including a broken right P4, a canine, and phalanx were a part of the syntypic series of Canis priscolatrans figured by Cope (1899: pl. 1B, fig. 3a, 3b, and 3g, respectively). In the course of his description of C. priscolatrans, Cope also mentioned unnumbered limb bone fragments ‘‘very large, exceeding those of the largest wolf known to me.’’

Haile 7A (early Rancholabrean sensu Morgan and Hulbert, 1995: table 2), Alachua County, Florida: UF 11845, associated right and left rami with p1–m2, left maxillary fragment with P2–M1, fragments of the cranium and left ascending ramus. Identified as C. rufus by Nowak (1979: 88).

Distribution: Early Irvingtonian of Arizona; late early to late Irvingtonian and early Rancholabrean of Florida; late Irvingtonian of Kansas, Texas, Arkansas, Maryland, and Pennsylvania.

Revised Diagnosis: Canis armbrusteri shares with C. lupus and C. dirus the following synapomorphies: incisive foramina extend posterior to limit of canine alveoli, strong posterior expansion of paroccipital process, posterior expansion of frontal sinus to frontoparietal suture, and very reduced M1 parastyle lacking union with preparacrista. It differs from C. lupus in that it retains premolars in which P3 and p2–p3 usually have posterior cusps, and the second posterior cusp of p4 is separate from the posterior cingulum. Like C. dirus , it also differs from C. lupus in having short-crowned and relatively straight canines, the metaconid of m1–m2 reduced, and a reduced P4 protocone. Although a sister taxon to C. dirus , C. armbrusteri lacks the other hypercarnivorous dental features characteristic of C. dirus including a primitively retained labial cingulum on M1.

Description and Comparison: C. armbrusteri was originally described by Gidley (1913: 98) based on three incomplete jaws from Cumberland Cave, which he thought differed from those of C. lupus in the greater depth of the jaw, a smaller canine, simpler p2 and p3 usually without posterior cusps, the presence of an additional posterior cusp on p4, and a relatively larger m1 talonid. Gidley also wrote that, compared to C. lupus , the m1 paraconid is less expanded at the base, and the metaconid is larger and situated higher on the protoconid and is narrower. Additional material obtained later shows that some of the characters listed in Gidley’s original diagnosis of C. armbrusteri are variable within the Cumberland Cave population of that species. The depth of the horizontal ramus is no greater in some C. armbrusteri (F: AM 67286) than in C. lupus . Based on all of the specimens that we have examined, the p4 of C. armbrusteri always has an additional posterior cusplet on the cingular shelf anterior to the dorsally produced posterior cingulum. In mature individuals diastemata are present between the anterior premolars of C. armbrusteri , especially between p2 and p3.

Patterson (1932) described two upper molars from Cumberland Cave and pointed out the relative large size of M2 and the variability in the size of the M1 hypocone. He wrote that the reduction in the size of the M1 hypocone does not reach the stage common to C. dirus from Rancho La Brea.

Gidley and Gazin (1938: 15) referred additional materials, including skulls, to C. armbrusteri from the type locality. They figured four skulls ( NMNH 7994, 11886, 11883, and 11885) and discussed additional morphological features that strengthened the diagnosis of C. armbrusteri . Compared to similar-sized skulls of C. lupus occidentalis, Gidley and Gazin found that the skull of C. armbrusteri has a slender muzzle, broad heavy frontals, a prominent sagittal crest, and the inion in the largest skull ( NMNH 11886) ‘‘projects backward to a marked degree, but not nearly so much as in the Rancho La Brea wolves’’ (i.e., C. dirus ). Our comparison of the proportional relationships of 14 cranial measurements taken on the Cumberland Cave samples of C. armbrusteri by Nowak (1979: appendix B) with the mean values for C. lupus occidentalis from Alberta and Northwest Territory of Canada and Alaska in the Department of Mammalogy collection ( AMNH) shows (fig. 43) rather close proportional resemblance between these taxa except for the narrow muzzle (width of palate at P1) and palate, short temporal opening (length M2 to bulla), shallow maxillary below the orbit, and wider M 2 in C. armbrusteri . Although C. dirus has a relatively reduced M2 and greater width across the frontals and muzzle, its relatively shallow maxillary and short temporal fossa resemble C. armbrusteri rather than C. lupus .

Comparison of proportions of the dentition of C. armbrusteri and C. dirus and the living populations of C. lupus occidentalis referred to above (fig. 44) reveals that, except for the small P1–P2, the dentition agrees in proportional relationships with C. l. occidentalis and lacks the degree of hypertrophy of P4 and p4 shown in C. dirus compared to other teeth.

Cope (1895) described a fragmentary upper dentition from the Tule Formation of Rock Creek, Texas, under the name C. indianensis Leidy, 1869 (a synonym of C. dirus ), and later Merriam (1912) compared this material favorably with C. dirus from Rancho La Brea except for the alveolar evidence of a large protocone root on P4. Additional materials of later Irvingtonian age at Rock Creek have been described by Troxell (1915), including a P3 and limbs that he also referred to C. dirus . The collections of the Panhandle Plains Museum from Rock Creek include fragments of a maxillary and ramus that corroborates the occurrence in the Tule Formation of a large canid whose morphology and dental dimensions fall within those for C. armbrusteri and below the range reported for C. dirus by Merriam (1912). Morphologically the M1 of JWT 688 (length X width, approximately 16.5 X 20.5 mm) resembles Cope’s specimen in having a small hypocone and very reduced anterolingual cingulum that is discontinuous across the protocone. In these ways it resembles C. dirus , as discussed below. The M2, judging from its alveoli, is relatively large however. Likewise, the lower carnassial of JWT 678 (length, 31.5 mm; width trigonid, 12.0 mm; width talonid, 11.0 mm) lies below the range of C. dirus , but is similar in size to those of late Irvingtonian C. dirus from Nebraska. The premolars (length and width p2, 13.7, 5.5 mm; p3, 14.7, 5.8 mm) are shorter and narrower than in C. dirus , and they lack posterior cusplets and are not separated by diastemata. They lie within the range of C. armbrusteri . Thus, this material, although approaching C. dirus in aspects of the morphology of M1, retains more primitive proportions and morphology in the rest of the available dentition. It is possible that this material samples an early population of C. dirus like that more clearly seen in the Nebraskan late Irvingtonian. Better preserved specimens are needed from Rock Creek, and so we follow the conservative course of referring this material to C. armbrusteri .

Based on his study of the material from the late Irvingtonian Coleman 2A site in Florida, Martin (1974: 77) concluded that C. armbrusteri is synonymous with C. lupus , and thus C. dirus and C. lupus were the only wolves in the middle and late Pleistocene deposits of North America. However, the morphology of this sample agrees better with C. armbrusteri , as Nowak (1979: 92) concluded. The Coleman 2A sample is chronologically not far removed from the Cumberland Cave type series. It serves to reinforce certain morphological features characteristic of late examples of this species. In the skull (fig. 55A–B) the muzzle is narrow, the postorbital processes are moderately inflated but with a nearly flat frontal shield, the nasals reach just beyond the posterior limit of the maxillary-frontal suture, the inflated parasagittal crests join at or just ahead of the frontoparietal suture, and the inion overhangs the condyles but does not achieve the hooklike form seen in C. dirus . The paroccipital process is slender, posteriorly expanded but not strongly produced ventrally; the jugal-maxillary suture at the anterior base of the zygoma is acute; the bullae are moderately inflated and reach forward just to the postglenoid processes; and the palate ends just behind M2. The rami are notably slender anteriorly, even in older adults ( UF 11520 young adult, fig. 55C–D, depth at p1/2 of 19.6 mm, depth at m1/2 of 25.5 mm; UF 11518 old adult, depth at m1/2 of 28.0 mm). The premolar row is nearly straight and the premolars show little imbrication. In UF 11520 the P2 and p2 possess posterior cusps, a rare variant in C. armbrusteri . The lower dentition shows little evidence of hypercarnivory beyond the small size of m1–m2 metaconids and entoconids, and the m3 is unicuspid but retains a large lingual cingular shelf. The P4 is wider, the M1 labial cingulum less prominent, and the lingual cingulum across the M1 protocone is absent in comparison with the early Irvingtonian Leisey Shell Pit sample, which otherwise resembles the Coleman 2A individuals in size.

The Haile 7A skull and jaw fragments ( UF 11845) of early Rancholabrean age are the latest well-preserved material of C. armbrusteri available from North America. The young adult skull fragments show the parasagittal crests meet at the frontoparietal suture so that there is no frontal contribution at this stage in ontogeny. The mastoid process is tiny, smaller than in the Coleman 2A sample. The upper dentition shows modifications transitional to C. dirus in M1: loss of anterolingual cingulum and much of the labial cingulum, tiny parastyle, but the hypocone is still large and essentially continuous with the posterior cingulum. The upper premolars are small and separated by diastemata; the P2–P3 have posterior cusps. The mandible is shallow anteriorly as in other C. armbrusteri (depth at p1/2 of 23.6 mm, depth at m1/2 of 33.8 mm). Lower premolars 2–3 are low-crowned and lack posterior cusps as in the McLeod sample, and p4 has a second posterior cusp anterior to the cingulum. The metaconid is reduced on m1–m2, but the m1 talonid is bicuspid. This individual is clearly an example of C. armbrusteri .

The best sample of limb bones of C. armbrusteri was obtained from Cumberland Cave, the type locality. Measurements of the lengths of these elements were kindly forwarded by Fred Grady of NMNH in 2002. None of these was described by Gidley and Gazin (1938) beyond the remark that the limb bones ‘‘are equally as long but slenderer than corresponding elements of various individuals of the Brea wolf in the National Museum collections.’’

Discussion: Nowak (1979: 90) concluded that C. armbrusteri was distinct from C. lupus . He thought that C. armbrusteri evolved from the basal stock of primitive wolves represented by C. edwardii and that the presence of these large wolves in Irvingtonian sites suggested that divergence of C. armbrusteri occurred early in the Pleistocene. Canis edwardii does show a number of synapomorphies with the larger wolves and coyote, including reduction of the m2 metaconid; p3 principal cusp positioned lower than those of adjacent premolars, crown base of p3 lies below p4; p3 with posterior cusp (reversal); and paracone of M1–M2 markedly enlarged over metacone. However, the skull proportions of C. edwardii , as shown on the log-ratio diagram (fig. 43), diverge widely from those of C. armbrusteri , which, in turn, closely parallel those of C. dirus and C. lupus . Only the proportionally narrower muzzle and simple P2–P3 and p2–p3 of C. armbrusteri deviate from both C. dirus and C. lupus and agree with C. edwardii . Additionally, the skull of C. armbrusteri is narrower across the frontals and the cheek teeth than seen in C. dirus . In general, however, the measurements of C. armbrusteri , C. dirus , and C. lupus form a general grouping on the logratio diagrams (figs. 43 and 44) and diverge from those of C. edwardii , C. latrans , and C. lepophagus .

Canis armbrusteri represents the first lineage of wolf-sized Canina to appear in the North American record, with its earliest occurrence near the beginning of the Pleistocene (earliest Irvingtonian, Anita fissure fills, Arizona, correlated tentatively with the Old- uvai Subchron, Lundelius et al., 1987). This slightly predates the first occurrence of large wolves in Europe, that is, C. falconeri Forsythe-Major, 1877 (Tasso Fauna of Italy, Azzaroli et al., 1988; Masini and Torre, 1989; see appendix 1), in just post-Olduvai, early Pleistocene (late Villafranchian) time. However, in eastern Asia, large Canis is known from the later Pliocene in China ( C. chihliensis , appendix 1). The latter taxon appears to represent a sister lineage to the Holarctic and North American forms represented by Canis lupus , C. armbrusteri , and its derivative C. dirus (fig. 65). The abrupt appearance in the early Pleistocene of North America of large C. armbrusteri suggests that this is an immigrant taxon from an Asian source, much as C. lupus was later in the Pleistocene.