Holotype: AMNH 33810, left dentary with p2-m3 from Scarritt Quarry, Fort Union Formation, Montana.

Referred Specimen: RAM 9099, right dentary fragment with m2-m3 from RAM locality V200510, member 4a (fig. 2A–B).

Description and Discussion: “Propalaeosinopa” was the generic name employed for a common North American Torrejonian-Tiffanian pantolestid, but Scott et al. (2002) demonstrated that the type of “Propalaeosinopa” P. albertensis, is a nomen dubium. Van Valen (1967) originally proposed the synonymy of Bessoecetor thomsoni, B. diliculi, and “Propaleosinopa” albertensis. Scott et al. (2002) argued that B. diliculi is identical to the type of “Diacodon” septentrionalis described by Russell (1929), and thus has priority. Scott et al. (2002) then resurrected Bessoecetor and transferred the four species of “Propaleosinopa” (P. thomsoni, P. dilucli, P. albertensis, P. septentrionalis) to the new combination, Bessoecetor septentrionalis. This synonymy was provisionally followed by Secord (2008), but he noted that the type specimen of the genus should be the holotype of B. thomsoni (AMNH 33810) from Scarritt Quarry, not the type of “Diacodon” septentrionalis (UALVP 126).

The m2-m3 of RAM 9099 have tall trigonids with large protoconids and metaconids and small paraconids that are shifted labially and low talonids with large hypoconids and slightly smaller entoconids, with small and medially placed hypoconulids (fig. 2A–B); features present in both known species of Bessoecetor; B. septentrionalis and B. pilodontus. However, B. septentrionalis can be differentiated from B. pilodontus by size, although teeth of B. pilodontus and B. septentrionalis may overlap slightly in length (Secord, 2008). Tooth dimensions of RAM 9099 are very similar to the holotype of B. septentrionalis and are significantly smaller than the holotype of B. pilodontus (table 1). Therefore, RAM 9099 is referred to B. septentrionalis.

Holotype: SMM P77.6.64, right dentary with canine and p4-m2 from the Judson Locality, Tongue River Formation, North Dakota.

Referred Specimen: RAM 6927, right dentary fragment with c, p1-2, p4, and partial ml from RAM locality V200001, member 4a (fig. 2C–D).

Description: The canine of RAM 6927 is large. Its apex extends anteriorly and its root curves posteriorly underneath p1. A wide diastema separates the canine and p1, which has a tall and narrow anterior cusp, and a small and low posterior accessory cusp (fig. 2C–D). The diastema separating p1 and p2 is short, and p1 and p2 are similar in morphology, with p2 a larger version of p1, but with a relatively tall anterior cusp. Two alveoli filled with sediment mark the position of the missing p3, which indicates the p2-p3 diastema was short and the p3-p4 diastema was absent. A mental foramen is positioned below the anterior root of p3.

A well-developed anterior cuspule and large main cusp and two small posterior cuspules are present on p4. The apex of the anterior cuspule is conical and separated from the main cusp by a broad carnassiform notch. The first posterior cuspule is positioned on the posterior slope of the main cusp, with a small carnassiform notch separating them. A third carnassiform notch separates the first posterior cuspule from the slightly lower second posterior cuspule. A small talonid basin is developed between the posterior cuspules that opens lingually and cusps and cuspules of p4 are aligned in a slight arc that opens lingually in occlusal view.

The m1 of RAM 6927 has a tall and well-preserved trigonid and a low and damaged talonid. The protoconid is the tallest cusp, with a distinctly smaller metaconid that is taller and slightly larger than the paraconid. Part of the talonid is missing and only a small hypoconid is present whose position is significantly lingual to the protoconid, suggesting that the talonid was small. Depth of the mandible below p2 is 1.09 mm, and 2.01 mm below m1.

Discussion: The presence of a sectorial main cusp, a smaller but well-developed anterior cuspule, small posterior cuspules developed on the posterior flank of the main cusp, and a small talonid basin, are features of the p4 of RAM 6927 that indicate it represents Protictis. Torrejonian and Tiffanian species of Protictis from the Bighorn and Clark's Fork basins, including P. schaffi, P. agastor, P. dellensis, P. haydenianus, P. laytoni, and P. paralus, were noted by Gingerich and Winkler (1985) as usually differentiated by size. Later, Polly (1997) removed P. schaffi, P. dellensis, P. laytoni from Protictis by synonymizing P. schaffi with Viverravus politus, Protictis dellensis with Didymictus proteus, and Protictis laytoni with Viverravus bowni. More recently, two Torrejonian species of Protictis, P. simpsoni and P. minor from the San Juan Basin of New Mexico, were described by Meehan and Wilson (2002). Protictis agastor, P. minor, P. simpsoni, and P. haydenianus are larger than P. paralus (Gingerich and Winkler, 1985: tables 4–5, 8; Meehan and Wilson, 2002: table 2; Holtzman, 1978: table 31). Direct comparison of the holotype of P. paralus (SMM P77.6.64) and RAM 6927 indicates that the p4 of RAM 6927 is nearly identical in all respects to the p4 of the holotype, except the anterior accessory cusp of RAM 6927 is slightly larger. The p4 of both the holotype and RAM 6927 have a small talonid basin and are similar in size (table 2).

The ml of RAM 6927 has a metaconid that is slightly taller and larger than the paraconid. In the ml of the holotype, the paraconid is slightly larger than the metaconid, but a second ml (SMM P77.8.190) of P. paralus from the type locality has a paraconid and metaconid equal in size and height, indicating that some variation can be expected in ml trigonid cusp morphology in P. paralus. These three m1s (RAM 6927, SMM P77.6.64, SMM P77.8.190) have identical trigonid widths (table 2). Based on nearly identical p4s, little difference in ml trigonid cusp development, and close similarity in size between RAM 6927 and the holotype of P. paralus, RAM 6927 is confidently referred to this rarely reported small species of Protictis.

Holotype: PU 21311, left dentary with ml-2 from Cedar Point Quarry, Polecat Bench Formation, Wyoming.

Referred Specimen: RAM 7246, left dentary fragment with partial ml and alveoli of m2 from RAM locality V200001, member 4a (fig. 2E–G).

Description: RAM 7246 is a dentary fragment found in a broken concretion whose ml is missing the anterior face of the trigonid below the apex of the paraconid (fig. 2E–F). The trigonid is much taller than the talonid. RAM 7246 has a large protoconid and smaller paraconid and metaconid that are equal in height. The talonid basin is deepest adjacent to the trigonid, and a large hypoconid, smaller hypoconulid and entoconid, and low and very small entocristid are present (fig. 2G). The hypoconulid is medial, with the entoconid positioned directly adjacent, so they appear twinned. The alveoli of m2 are sediment filled, with the anterior alveolus wider than long, and the posterior alveolus longer than wide. RAM 7246 has a length of 5.0 mm (minimum, because of breakage) with a trigonid width of 4.39 mm and talonid width of 3.37 mm. Mandible depth beneath the talonid of ml is 8.75 mm.

Discussion: RAM 7246 represents a second species of Protictis from RAM locality V200001, which is larger than P. paralus (RAM 6927). Species of Protictis can usually be separated by size (Gingerich and Winkler, 1985; Meehan and Wilson, 2002) and RAM 7246 is most similar to P. haydenianus and P. agastor in this respect. P. haydenianus differs from P. agastor in having a more sectorial m2, with a shorter and less massive trigonid (Gingerich and Winkler, 1985), but m2 is not present in RAM 7246. However, two lines of evidence suggest RAM 7246 is probably P. agastor. Protictis haydenianus is well known from Torrejonian strata and questionably known from Tiffanian strata (Gingerich and Winkler, 1985), and Protictis agastor and RAM 7246 are both known from mid-Tiffanian strata (this report). Also, P. paralus (RAM 6927) is found at the same locality as RAM 7246, which is important because P. agastor and P. paralus are rarely found in Tiffanian strata but occur together at the Chappo Type locality and Cedar Point Quarry in Wyoming (Gingerich and Winkler, 1985; Gunnell, 1994), both mid-Tiffanian sites (Archibald et al., 1987; Lofgren at al., 2004b). Based on size and its Tiffanian occurrence, RAM 7246 is tentatively referred to Protictis cf. P. agastor.

Referred Specimen: RAM 15328, lower canine fragment from RAM locality V200612, member 4a (fig. 2H–I).

Description and Discussion: RAM 15328 is the distal end of a tooth about 2 cm in length and width, with enamel coating half of the tooth's longitudinal axis, a strong indicator that it represents a taeniodont (fig. 2H–I). RAM 15328 was compared directly to canines and incisors of Psittacotherium and Ectoganus housed at the AMNH and it was most similar in size and shape to AMNH 3392 and AMNH 3390, lower canines identified as Psittacotherium (Schoch, 1986). Although RAM 15328 too is fragmentary to be assigned to Psittacotherium with confidence, it does represent the only reported record of a taeniodont from the West Coast of North America.

Holotype: USNM 9546, part of maxilla with P4-M3 from Gidley Quarry, Fort Union Formation, Montana.

Referred Specimen: RAM 6426, right P4 from RAM locality V94014, member 4a.

Discussion: This isolated tooth was described and figured by Lofgren et al. (2008: fig. 4). RAM 6426 closely resembles P4s of Paromomys from Swain Quarry that are inseparable from those of P. depressidens from Gidley Quarry (Rigby, 1980), the type locality.

Holotype: AMNH 17368, left maxilla with P2, P4-M2 Mason Pocket, Animas Formation, Colorado.

Referred Specimen: RAM 6433, left m3 from RAM locality V98012, member 4b.

Discussion: RAM 6433 was described and figured by Lofgren et al. (2008: fig. 5). The morphology of this isolated m3 is nearly identical to m3s referred to Ignacius frugivorus from the Roche Percee Local Fauna in Saskatchewan, Canada, the Brisbane and Judson localities in North Dakota, and Mason Pocket, the type locality in Colorado (Lofgren at al., 2008).

Holotype: RAM 9044, right dentary fragment with damaged il, p2 alveolus, and p3-m3 from RAM locality V200510, member 4a, Goler Formation, California (fig. 3A).

Referred Specimens: RAM 9668, left maxilla fragment with C, alveoli of P2 and P3, P4-M1 (fig. 5C), and RAM 9669, crushed skull fragment with broken left I1, right P4, M1-2 and associated right p3 (fig. 5E–F) from RAM locality V200510, member 4a; RAM 9429, right dentary fragment with il root and p3-m3 from RAM locality V200706, member 4a (fig. 3B); RAM 10029, right dentary fragment with damaged p4-m3 from RAM locality V200804, member 4a (fig. 3C); five m3s RAM 6961 (fig. 4A), RAM 6963 (fig. 4B), RAM 6960 (fig. 4C), RAM 6700 (fig. 4D), RAM 6934, m2 RAM 6446, mx fragment RAM 6935, mx trigonid RAM 6946, M3 RAM 6957 (fig. 5H), M2 RAM 6962 (fig. 5G), P3 RAM 6931 (fig. 5D), and 2 I1s RAM 6431 (fig. 5A), RAM 6929 (fig. 5B) from RAM locality V98012, member 4b; RAM 7193, partial right dentary with talonid fragment of p4, m1 and broken m2 from RAM locality V200202, member 4b; RAM 6925, partial left m3 from RAM locality V94014, member 4a.

Etymology: Named in honor of Richard “Dick” Lynas, for his dedicated efforts in support of collection and study of Goler Formation vertebrate fossils.

Diagnosis: Midsized plesiadapid with a narrow p4 and m2 that align it with Nannodectes and a unique combination of other features that separates it from all known plesiadapid species, which include, I1 with medio cone small or absent and centroconule absent, upper molars with no significant development of crenulated enamel and lacking mesostyles, p3-4 with inflated protoconids and distinct talonid basins, m 1-2 with squared entoconids, and m3 with slight to moderate talonid enamel crenulations and broad and unfissured hypoconulid that is usually rounded.

Description: The holotype of Nannodectes lynasi (RAM 9044) has a damaged il, p3, m2, and m3 (fig. 3A). Only the triangular-shaped posterior part of il is preserved. A single round depression, posterior to il and directly anterior to p3, represents the p2 alveolus. The trigonid of p3 is missing due to breakage. The talonid of p3 has two small medial cusps on its posterior margin (labially positioned cusp largest), and a very small, but distinct basin is present. The trigonid of p4 is dominated by a massive and worn protoconid (if a paraconid and metaconid were present, all trace has been removed by wear), and a distinct cristid obliqua connects the trigonid to a large hypoconid. The posterior margin of the talonid is defined by a postcristid developed between the hypoconid and smaller entoconid. The talonid basin of p4 is large, but relatively small compared to m1-m2. Molar trigonids are simple, with large metaconids and protoconids, and smaller labially shifted paraconids (breakage removed metaconid and paraconid of m3). Talonids of m 1–2 have large hypoconids and the entoconid of ml is squared (entoconid of m2 broken). Crenulated enamel is not developed on m1–2. Notable features of the m3 talonid are faint to moderately developed crenulations, a labially projecting hypoconid lobe, and a broad, rounded, and unfissured hypoconulid.

RAM 9429 represents another mostly complete dentary of N. lynasi (fig. 3B), but one with significant enamel corrosion. RAM 9429 has a small, laterally crushed depression directly anterior to p3 that may represent the p2 alveolus, and a large, anteroposteriorly elongate protoconid dominate the trigonid of p3. The entoconid, hypoconid, cristid obliqua, and talonid basin are more developed in the p3 of RAM 9429 than the p3 of the holotype (RAM 9044). The trigonid of p4 has a very large protoconid, but unlike the holotype, a small metaconid and labially shifted paraconid are also present. Similar to p3, the hypoconid and entoconid are larger, and cristid obliqua and talonid basin more developed in the p4 of RAM 9429 than the holotype. The ml of RAM 9429 has a large protoconid and metaconid and a smaller, lingually positioned paraconid (less lingually positioned than in p4). In m2, the paraconid is more lingually positioned and more closely appressed to the metaconid. Entoconids of ml and m2 are squared in RAM 9429. The talonid of m3 has a broad, rounded, and unfissured hypoconulid like the holoype, but differs in having a narrower hypoconulid lobe and a small entoconid that forms a distinct lingual bulge.

RAM 10029 is a partial dentary with heavily damaged p4-m3 as the crowns are partially dissolved (fig. 3C). Only the most robust crown features of RAM 10029 are visible, but in general shape and size, RAM 10029 closely resembles the holotype (RAM 9044) and RAM 9429. Also, the p4 talonid basin of RAM 10029 was large, like the p4 of RAM 9429. RAM 10029 also has a broad and rounded m3 hypoconulid and lacks an entoconid lobe, features more like the holotype than RAM 9429.

The partial dentary (RAM 7193) and isolated m3s previously referred to Plesiadapis cf. P. churchilli or P. cf. P. anceps (Lofgren et al., 2008), compare favorably to the holotype of Nannodectes lynasi and RAM 9429, except for RAM 6963. Isolated m3s of N. lynasi from the Goler Formation are uniform in morphology (fig. 4A, C–D), but RAM 6963 is smaller than all other m3s (table 3) and its hypoconulid is distinctly squared (fig. 4B). Because RAM 6963 comes from the same site as four other m3s of N. lynasi (fig. 4A, C–D), it most likely represents an extreme morphological variant of the species.

Upper incisors of N. lynasi are broken, but RAM 6929 and RAM 6431 preserve a significant part of the crown (fig. 5A–B), and RAM 9669 just the alveolar rim. RAM 6929 and RAM 6431 were described by Lofgren et al. (2008), and their important features are the minute mediocone and lack of a centroconule in RAM 6929 and the absence of both a mediocone and centroconule in RAM 6431 (fig. 5A–B). The alveolar rim of RAM 9669 is 2.54 mm in width and 4.34 mm in length.

The upper canine and premolars of N. lynasi are represented by RAM 9668 (C and P4, with alveoli of P2-3), RAM 6931 (P3), and RAM 9669 (P4). In RAM 9668 (fig. 5C), a single-rooted tooth is positioned adjacent to the anterior alveolus of P2, indicating the lack of a diastema between the teeth. This rarely preserved tooth has been identified as the canine in plesiadapids (Gingerich, 1976; Boyer, 2009), although at first glance it would appear to be P1 based on its position adjacent to P2. The canine of RAM 9668 is a single upright cusp elongated anteroposteriorly, with a small shelf positioned low on the anterior face of the tooth. The close placement of the canine and P2 in RAM 9668 is reminiscent of the condition seen in Wa 282, a partial maxilla of Plesiadapis walbeckensis. In contrast, a canine to P2 diastema is present in PU13393, a partial maxilla of P. fodinatus. Alveoli present anterior to the P4 in RAM 9668 indicate that P2 had two circular roots orientated anteroposteriorly, with the anterior one approximately half the diameter of the posterior root and P3 had two roots of equal size, oriented transversely. Diastemas apparently were not developed between premolars in RAM 9668.

The only known P3 of Nannodectes lynasi (RAM 6931), described by Lofgren et al. (2008) as a P4, lacks a metacone and metaconule, but has a prominent protocone, and a paracone with a paraconule developed on its lingual face (fig. 5D). Both P4s (RAM 9669 and RAM 9668) of N. lynasi have a closely appressed paracone and metacone, small parastyle, prominent paraconule with a small enamel bulge positioned posteriorly (that could represent an incipient metaconule), large protocone, well-developed anterior and posterior cingulum, and weakly developed stylar shelf and labial cingulum (fig. 5C, E). Also, the lingual cingulum is not continuous in P4 and the mesostyle and metastyle are absent. In contrast to P3, whose paraconule is positioned on the lingual face of the paracone, the paraconule and paracone of P4 are separated by a distinct valley.

Features of the Ml and M2 in N. lynasi are similar and include: widely spaced paracone and metacone subequal in size and connected by centrocrista, metacone positioned posteriorly, parastyle small and connected to paracone by weak paracrista, ectoflexus weak, labial cingulum present, protocone larger than metacone and paracone and connected to paraconule and metaconule by preprotocrista and postprotocrista, small hypo cone present on crest extending posteriorly from protocone that merges with postcingulum, well-developed precingulum, lingual cingulum not continuous across protocone, and mesostyle and metastyle absent (fig. 5C, F, G). The only differences between M1 and M2 are a more developed centrocrista and parastyle on M2, and a stronger M2 ectoflexus in RAM 6962 (fig. 5G), compared to both the M1 of RAM 9668 and the M1-2 of RAM 9669 (fig. 5C, F). RAM 6957, the only known M3 of N. lynasi (fig. 5H) is damaged and has slightly crenulated enamel and lacks a distinct hypocone (Lofgren et al., 2008). Measurements of the dentition of N. lynasi are provided in tables 3–4.

Discussion and Comparisons: The initial sample of this Goler plesiadapid consisted of 14 isolated teeth and a dentary fragment found before 2003 that were referred to Plesiadapis cf. P. churchilli (dentary fragment) and P. cf. P. anceps (isolated teeth) because the specimens could not be confidently assigned to a known species (Lofgren at al., 2008). Between 2007 and 2009, three partial dentaries and two partial maxillae were recovered from a 300 m thick sequence of member 4a strata in a large canyon informally termed Grand Canyon (no. 2 in fig. 1). RAM locality V200510 in Grand Canyon yielded a dentary with p3-m3 (RAM 9044) and two partial upper dentitions (RAM 9668, RAM 9669), specimens reasonably certain to represent the upper and lower dentition of a single species.

During the early and middle Tiffanian in North America, lineages of similar plesiadapid genera, Nannodectes and Plesiadapis, evolved in parallel, with Nannodectes distinctly smaller than Plesiadapis when both genera are found at the same site (Gingerich, 1976). Also, Nannodectes was distinguished from contemporaneous species of Plesiadapis by its narrower cheek teeth (especially in the lower premolars), more triangular P3 (usually lacking a paraconule), and lack of a centroconule on I1 (Gingerich, 1976).

In the Goler Formation, a single plesiadapid species is present, so size comparison between morphotypes is not possible. Also, the lack of a centroconule on I1 and a more triangular P3 are characters that do not reliably distinguish the two genera as they are found in Nannodectes and some species of Plesiadapis. However, narrower cheek teeth appear to distinguish Nannodectes and Plesiadapis as a cladogram based on dental characters of plesiadapid species indicates that Nannodectes is monophyletic, based on proportional differences between length and width of p4, M1-3, and m1-3 (Boyer, 2009: fig. 5.1, table 5.1, characters 18, 23, and 31). Thus, the lower cheek teeth of Nannodectes should be narrower than those of Plesiadapis and upper molars of Nannodectes more squared compared to more rectangular upper molars of Plesiadapis. To test this, ratios of length versus width of cheek teeth of Nannodectes and North American species of Plesiadapis were calculated using length and width means provided by Gingerich (1976: table 2, tables A3–A13) for each species. These ratios were compared to similarly derived ratios for the cheek teeth of Nannodectes lynasi from the Goler Formation (table 5). For the lower dentition, a number above 1.00 for a tooth site indicates that the teeth are longer than wide (narrower). For the upper dentition, all teeth are wider than long, but a number close to 1.00 for a tooth site indicates that these teeth are more squared (less rectangular).

Length versus width ratios of cheek teeth provide limited support for distinguishing Nannodectes from Plesiadapis. Ratios of the p4 of N. intermedius, N. gazini, N. simpsoni, and N. gidleyi are all above 1.00, indicating that the p4 of Nannodectes is consistently narrow. Early and middle Tiffanian species of Plesiadapis (P. praecursor, P. anceps, P. rex, P. churchilli) have a p4 that is wider than long (ratio less than 1.00), but later Tiffanian species (P. fodinatus, P. dubius, P.cookei) have narrow p4s like Nannodectes (table 5). The m2 of Nannodectes is slightly narrower than the m2 of Plesiadapis praecursor, P. anceps, and P. rex, but not narrower than the m2 of P. churchilli, P. fodinatus, P. dubius, and P. cookei (table 5). For the p3, ml or m3, there is no significant difference in length versus width ratios between genera. For the upper dentition, the P3 of Nannodectes gazini, N. simpsoni, and N. gidleyi is more squared than in any North American species of Plesiadapis, while the P4 of Nannodectes gazini, N. simpsoni, and N. gidleyi is significantly more squared than only Plesiadapis praecursor, P. rex, and P. churchilli (table 5). There are no significant differences in ratios of length and width in the upper molars of Nannodectes and North American species of Plesiadapis (table 5). Length versus width ratios indicates that only a narrow p4 and more squared P3 and P4 are measurable characters that separate Nannodectes from early and middle Tiffanian species of Plesiadapis.

Goler dentaries (RAM 9044, RAM 9429, and RAM 10029) have narrow p4s and m2s with ratios above 1.00 (table 5), which supports their referral to Nannodectes, but Goler P4s and the single P3 have ratios not clearly aligned with either Nannodectes or Plesiadapis (table 5). Thus, proportional differences in certain cheek teeth of Nannodectes and Plesiadapis that distinguish the two genera are valid only for the lower dentition of the Goler sample. However, comparison of the dentition of this Goler plesiadapid to all species of Nannodectes, Plesiadapis, and Chiromyoides indicates a closer morphological affinity to Nannodectes gidleyi and N. aff. gidleyi, which also supports referral of the Goler sample to Nannodectes and the diagnosis of the new species, N. lynasi.

When compared to middle and late Tiffanian species of Plesiadapis, Nannodectes lynasi is much smaller than Plesiadapis simonsi, P. gingerichi, P. cookei, P. tricuspidens, and P. remensis, and lacks the more derived dental features seen in some combination in P. fodinatus, P. dubius, and P. churchilli. These features include highly crenulated upper molar enamel, m3 hypoconulid squared and fissured, m3 with highly crenulated talonid enamel, rounded m1-2 entoconids, centroconule on I1, and upper molars with mesostyles (Gingerich, 1976).

Compared to the early and middle Tiffanian species of Plesiadapis, Nannodectes lynasi is much larger than Plesiadapis praecursor, P. insignis, and P. anceps. Previously, isolated plesiadapid teeth from the Goler Formation were referred to Plesiadapis cf. P. anceps (Lofgren et al., 2008), but molars and premolars of the three recently collected dentaries (RAM 9429, RAM 9044, and RAM 10029) indicate the size of the lower dentition of Nannodectes lynasi significantly exceeds Plesiadapis anceps (table 6). Also, the lower premolars from these dentaries show the size of the p3 and p4 protoconid and talonid basin of p4 in Nannodectes lynasi far surpass development of these features in Plesiadapis anceps.

The European species, P. walbeckensis, is similar to Nannodectes lynasi in size and some other features, but it has poorly developed p3–4 talonids and a fissured m3 hypoconulid with more talonid crenulations than the m3 of N. lynasi. Plesiadapis rex, is also generally similar to Nannodectes lynasi in size, but differs in having an m3 hypoconulid that is usually fissured and squared, a centroconule on I1, mesostyles on M2–3, and a mesostyle sometimes on Ml (Gingerich, 1976). Also, Plesiadapis rex has weakly developed p3–4 talonids, compared to the relatively well-developed p3–4 talonids of Nannodectes lynasi.

Chiromyoides (particularly C. minor) shares some features with Nannodectes lynasi as it has a rounded, broad, and unfissured m3 hypoconulid with uncrenulated talonid enamel, but its p4 lacks a talonid basin. Also, the central upper incisors of Chiromyoides are extraordinarily robust (Gingerich, 1976), unlike Goler I1s.

In comparison to species of Nannodectes, N. lynasi is much larger than N. intermedius, N. gazini, and N. simpsoni. The lengths of the m3s and p4s in these species do not exceed 4.3 mm and 2.4 mm respectively (Gingerich, 1976: tables A3–A5), while in N. lynasi the lengths of m3s range from 4.32–5.43 mm and p4s from 2.52–3.05 mm (table 6).

Nannodectes gidleyi has many similarities to N. lynasi as it has a rounded, unfissured m3 hypoconulid, lacks a centroconule on I1, and has a small talonid basin on p3-4, usually with a small but distinct cristid obliqua (Gingerich, 1976). Also, a hypoconid and entoconid are developed on the labial and lingual edges of the posterior talonid margin of the p4 in AMNH 17389 of N. gidleyi, a feature shared with the holotype of N. lynasi (RAM 9044). Finally, some m3s of N. gidleyi (such as AMNH 17365 and AMNH 17174) have talonid enamel crenulations that are slightly or moderately developed, indicating some overlap in this feature with N. lynasi. However, development of crenulated talonid enamel in most m3s of N. gidleyi (for example, AMNH 17171 and AMNH 17389) exceeds anything seen in N. lynasi. Also, N. gidleyi usually has mesostyles on Ml and mesostyles always on M2–3 (Gingerich, 1976) and relatively small protoconids on p3–4. Nannodectes lynasi lacks upper molar mesostyles and has large p3–4 protoconids. Also, lower premolars and molars of N. lynasi are distinctly larger than those of N. gidleyi, although this distinction is less evident when comparing m3s (table 6).

Nannodectes aff. N. gidleyi from Big Bend, Texas, has a p4 that is closer in size to N. lynasi, but its lower molars are smaller, about the same size as N. gidleyi (table 6). Nannodectes aff. N. gidleyi apparently lacks a centroconule on I1 (see TMM 41365-274 in Schiebout, 1974: fig. 16b) and has an unfissured and often rounded m3 hypoconulid lobe (Gingerich, 1976), features shared with N. lynasi. However, Nannodectes aff. N. gidleyi has distinct upper molar mesostyles, squared hypoconulid lobes on some m3s, and highly crenulated m3 talonid enamel (Gingerich, 1976), although the two m3s illustrated by Schiebout (1974: TMM 41365-541: fig. 16k and TMM 41366-77: fig. 17) show slightly crenulated m3 talonid enamel. In any case, N. aff. N. gidleyi and N. gidleyi are closer in morphology to N. lynasi than any other plesiadapid species.

Referred Specimen: RAM 9041, right ml from RAM locality V200510, member 4a (fig. 6A).

Description: RAM 9041 is missing the metaconid due to breakage, but the protoconid, hypoconid, hypoconulid, and entoconid are all well preserved (fig. 6A). A small paraconid is connected to the protoconid by a long and arcuate paracristid. Broadly and deeply basined, the talonid is significantly longer and slightly wider than the trigonid. The hypoconid is large and the entoconid and hypoconulid are much smaller, with the hypoconulid slightly smaller than the entoconid. The hypoconulid is markedly shifted lingually, positioned immediately adjacent to the entoconid, so that the two cusps appear twinned, and a small entoconulid is present. The talonid basin is deepest against the back wall of the trigonid and a talonid notch is not present. A prominent shelflike labial cingulid extends continuously from the anterior face of the trigonid below the paraconid, across the protoconid, to the most labial margin of the hypoconid. On the posterior face of the talonid a shelflike postcingulid descends labially to the posterolabial corner of the hypoconid, where only a short gap separates it from the posterior terminus of the labial cingulid.

Discussion: RAM 9041 is most closely aligned with oxyclaenids and is most similar to Thryptacodon, based upon crown height, and occurrence and relative position of primary and secondary cusps. Based on size, RAM 9041 is much smaller than Thryptacodon antiquus, T. australis, and T. pseudarctos and is most similar to specimens included by Gazin (1956) in T. demari and T. belli, which were synonymized by Van Valen (1978) as T. demari. Similarities of RAM 9041 with T. demari include paraconid lingual and positioned anteriorly, talonid basin much wider than trigonid, distinct entoconulid, talonid notch closed, and hypoconulid lingually displaced and close to the entoconid. Only its smaller size and presence of a strong labial cingulid keep RAM 9041 from being identified as the Bison Basin species, now synonymized with T. orthogonius (Scott et al., 2002). RAM 9041 measures 4.95 mm in length, with a trigonid width of 2.64 mm and talonid width of 3.05 mm, compared to the ml of the holotype of T. “belli” (UW 1045), whose length and width are 5.6 mm and 3.8 mm respectively (Gazin, 1956). Thus, RAM 9041 is assigned to Thryptacodon species indeterminate until more specimens are recovered of this oxyclaenid.

Holotype: RAM 6908, right maxilla with P4-M3 from RAM locality V94133, member 4a, Goler Formation, California.

Referred Specimens: RAM 15622, right dentary fragment with talonid of p4, and m1-2 from RAM locality V201014, member 4a (fig. 6B–C); RAM 9670, left dentary fragment with the base of p4 and damaged ml from RAM locality V200510, member 4a (fig. 6D); RAM 6928, right dentary fragment with m3 talonid and base of trigonid (fig. 6E), and RAM 15333, left M2 from RAM locality V200001, member 4a.

Discussion: The holotype of M. tedfordi, a maxillary fragment with P4-M3 (RAM 6908), from V94133, high in member 4a, was described by McKenna and Lofgren (2003), who noted that M. tedfordi was significantly smaller than M. subtrigonus and all other described species of Mimotricentes and Lambertocyon. The holotype was the only known specimen of Mimotricentes tedfordi, so the lower dentition was not described.

In 2010, a dentary fragment (RAM 15622) was recovered from V201014, a site low in member 4a, about 3 km south of the type locality of M. tedfordi. RAM 15622 represents the lower dentition of a small arctocyonid that almost certainly is M. tedfordi. Two other fragmentary arctocyonid dentaries (RAM 6928 and RAM 9670), collected prior to 2008, probably also represent M. tedfordi. Also, an isolated M2 (RAM 15333), found in 2012, is nearly identical to the M2 of the holotype. These four specimens are described below.

Description: The M2 of the holotype of M. tedfordi (RAM 6908) and RAM 15333 are similar in all features, their unusually large crestlike parastyles being especially noteworthy. Also, the length of RAM 15333 is 4.54 mm and the width 5.19 mm, dimensions similar to that of the M2 of the holotype (length 4.62 mm and width 5.50 mm).

RAM 15622 is a dentary fragment with a broken p4, and m 1–2, with only the talonid and posterior wall of the trigonid of the p4 preserved (fig. 6B–C). The p4 has a low, 1.2 mm long, posterior shelf, which has a weakly developed posteromedial cusp connected to the trigonid wall by a small crest, features often seen in arctocyonids. The metaconid and protoconid of M1 are subequal in size and height, both larger than the low, labially shifted paraconid, and a protocristid and paracrista are present. The precingulid and a partial labial cingulid are present (labial cingulid terminates posteriorly at ectoflexid). The cristid obliqua is well developed and terminates at a large hypoconid, and a small but deep talonid basin, with steep posterior and labial slopes is present. A small posterolingually positioned entoconid and hypoconulid are closely appressed, with the hypoconulid slightly posterior to the transversely aligned hypoconid and entoconid. The m2 is similar to M1 in most respects, but m2 is more quadrate in outline and has a more labially shifted paraconid. Depth of the mandible below m1 is 6.15 mm.

The p4 of RAM 9670 is sheared off just above the gum line and the anterior part of the tooth is missing (fig. 6D), but its length was at least 3.5 mm, and width about 2.5 mm. The talonid of M1 is well preserved but the trigonid crown is missing due to breakage. The dimensions of the M1 of RAM 9670 are similar to those of the M1 of RAM 15622, but the talonid of RAM 9670 has crenulated enamel, a broader and shallower basin, and greater spacing of the entoconid and hypoconulid, differences suggesting they could represent different taxa. However, significant variation in lower molar morphology is well documented in Mimotricentes subtrigonus, a larger species than M. tedfordi (table 7), from Swain Quarry in Wyoming, where sample sizes of isolated teeth can exceed 60 specimens (Rigby, 1980: table 38). Teeth of M. subtrigonus exhibit considerable variation in talonid development, as the proximity of the hypoconuild and entoconid can vary from nearly fused to well separated with the hypoconulid medially located (Rigby, 1980). Also, some mis from Swain Quarry in the AMNH collections lack crenulated talonid enamel, while in others it is well developed. If this amount of variation is also present in M. tedfordi, RAM 9670 likely represents the same species as RAM 15622.

RAM 6928 is a dentary fragment with a damaged m3 that may also represent M. tedfordi. RAM 6928 has an intact m3 talonid, but its trigonid is broken at the gum line (fig. 6E). The talonid has a low hypoconulid, an entoconid crest with three cuspules, and a large hypoconid. The talonid basin of m3 is broad and shallow and opens lingually through a small talonid notch. The morphology of the m3 talonid in Mimotricentes subtrigonus from Swain Quarry, varies even more than the M1 talonid, so much so that the m3 is “posteriorly too variable taxonomic use” (Rigby, 1980: 104). Specimens of M. subtrigonus observed at the AMNH from Swain Quarry have m3 talonids of similar construction as RAM 6928, but are much larger (table 7). Thus, RAM 6928 probably also represents M. tedfordi.

Holotype: UM 81147, left M1 from Chappo Type Locality, Wasatch Formation, Wyoming.

Referred Specimens: RAM 9040, right M1 (fig. 6F), and RAM 9043, left p3 fragment (fig. 6G) from RAM locality V200001, member 4a.

Description: RAM 9040 is an unworn M1 that is in pristine condition except for the metacone, which is missing due to breakage (fig. 6F). Based on the size of its base, the metacone was about as tall and broad as the large paracone. The protocone is massive, taller and larger than the paracone. A small and low hypocone is positioned slightly lingual to the protocone on the posterolingual corner of the postcingulum. The paraconule is larger than the metaconule, and a postparaconule wing connects the paraconule with the lingual base of the paracone (a premetaconule wing is not developed). The parastyle is large and crescent shaped, but a metastyle is not present. The lingual cingulum is complete, but weakly developed across the base of the protocone. The labial cingulum is also complete, but is weakest labial to the base of the paracone. The mesostyle is large and elongate, with its apex forming a large part of the labial cingulum. Also, a series of cuspules are present on the anteroposterior axis of the mesostyle and a small ridge projects lingually from its apex to the base of the cusp.

RAM 9043 is broken and only the anterolabial third of the tooth is present (fig. 6G), but what remains is nearly identical in morphology to p3s of L. gingerichi from the Chappo Type Locality. Distinct enamel crenulations on the anterolabial face of p3, its size, and the wear facet on its broken crown all mirror features observed in Lambertocyon.

Discussion: Three species of Lambertocyon have been described, L. ischyrus (Gingerich, 1978, 1979), L. eximius (Gingerich, 1979), and L. gingerichi (Gunnell, 1994), with the following reported occurrences: (1) L. ischyrus, Debeque Formation, Colorado (Gingerich, 1978, 1979; Kihm, 1984); (2) L. eximius, Blacks Peak Formation, Texas, (Schiebout, 1974; listed as Arctocyonidae, undescribed genus and species), Fort Union Formation, Wyoming (Winterfeld, 1982), and Polecat Bench Formation, Wyoming (Gingerich, 1979; Secord, 2008); and (3) L. gingerichi, Polecat Bench and Wasatch formations, Wyoming (Gunnell, 1994; Secord, 2008). The Goler specimens of Lambertocyon are both from the Land of Oz site (V200001), so they almost certainly represent the same species.

Species of Lambertocyon are differentiated by mesostyle development and size, with L. eximius significantly larger than L. ischyrus (Gingerich, 1979), and L. gingerichi 8%–10% larger than L. eximius (on average) and also having a well-developed mesostyle (Gunnell, 1994). Based on size, RAM 9040 is more like the holotype of L. eximus than the holotype of L. gingerichi, but measurements of the two species overlap (table 8), demonstrating that distinguishing species oí Lambertocyon by size is problematic. For example, the sample of Lambertocyon from Divide Quarry was referred to L. gingerichi (Secord, 2008), but one of the two M1s from Divide Quarry (UM 87040), has a length of 7.7 mm and a width of 8.1 mm (table 8), dimensions closer to the holotype of L. eximus, than that of L. gingerichi (table 8). The only reported M1 of L. ischyrus (FMNH PI5545) is 6.86 mm in length and 7.87 mm in width, which is significantly smaller than M1s of L. eximius and L. gingerichi (table 8).

The very large mesostyle of RAM 9040 is unique for Lambertocyon. Secord (2008) noted that M1 mesostyle development varies from strong to weak in L. gingerichi from Divide Quarry and larger samples are needed to conclusively demonstrate L. eximius and L. gingerichi are distinct species. RAM 9040 was compared directly to Mis of L. gingerichi (UM 81147 holotype and UM 85163 from Chappo Type locality; UM 87040 and UM 110259 from Divide Quarry), and these specimens have small mesostyles, in comparison to the distinctly larger mesostyle of RAM 9040. Also, the complete lingual cingulum of RAM 9040 is not reported for any M1 of Lambertocyon or observed in M1s of L. gingerichi from Chappo Type Locality or Divide Quarry.

The large mesostyle and complete lingual cingulum of RAM 9040 demonstrate that morphological variation within Lambertocyon is greater than previously documented and the Goler Formation occurrence may represent a new species. RAM 9043, the fragmentary p3, adds little to this discussion. An unusually large mesostyle and a weak but complete lingual cingulum on a single tooth is an inadequate basis from which to diagnose a new species, especially considering L. eximius and L. gingerichi require larger samples to confirm they represent distinct species. Thus, RAM 9040 and RAM 9043 are referred to Lambertocyon cf. L. gingerichi until more specimens of Lambertocyon are recovered from the Goler Formation, as the mesostyle development of RAM 9040 is more aligned with that species than L. eximus.

Holotype: RAM 9047, left dentary fragment with a partially erupted p4 and M1 from RAM locality V200702, member 4a, Goler Formation, California (fig. 7A–C).

Etymology: Named in honor of R. Larry Ashton, Jr., for his support of the research programs of the Raymond M. Alf Museum of Paleontology.

Diagnosis: Protoconid of p4 highly inflated, significantly more than other known p4s of Protoselene, p4 lacking metaconid and anterior basal cusp in contrast to P. opisthacus, paraconid of M1 more lingual than P. opisthacus or P. novissimus, relative length and width of M1 talonid shorter and wider than P. novissimus, and m1 larger than P. griphus.

Description: The distinctive features of the p4 of RAM 9047 are its greatly swollen protoconid and the lack of both an anterior basal cusp and metaconid (fig. 7A–C). On the anterior face of the protoconid is a small, but distinct ridge, that defects posterolabially as it traces the midline of the crown and extends to the apex. A well-developed cristid obliqua extends to the posterior edge of the p4, where it defects lingually and lessens in height. Thus, a small talonid basin is present that opens lingually. Talonid enamel is rugose, in contrast to the faintly crenulated protoconid enamel.

The m1 of RAM 9047 has a low, lingually positioned paraconid connected by a welldeveloped paracristid with a large protoconid. The metaconid and protoconid are subequal in size and height. The talonid basin of M1 opens lingually and is deep, and broad. The cristid obliqua extends posteriorly to the hypoconid, which is larger than the entoconid and the slightly smaller hypoconulid, which are closely appressed. A well-developed precingulid is present anterior to the trigonid, as is a small postcingulid posterior to the hypoconid. The p4 of RAM 9047 is 5.99 mm in length and 4.08 mm in width, and the M1, 5.63 mm in length, with trigonid and talonid widths of 4.13 mm and 4.33 mm respectively.

Discussion and Comparisions: Protoselene is a relatively rare component in Torrejonian and Tiffanian mammalian faunas from western North America. Found at many Torrejonian localities in the San Juan Basin of New Mexico (Matthew, 1937; Williamson, 1996), P. opisthacus is the best-known species. Another Torrejonian species is Protoselene (“Dracoclaenus”) griphus known from a few specimens collected at the early Torrejonian Dragon Canyon site in central Utah (Gazin 1939, 1941; Robison, 1986) and a single P4 from the lowest Torrejonian part of Kimbeto Wash in the San Juan Basin (Williamson, 1996). The third species is P. novissimus from early Tiffanian strata in the Bison Basin of southwest Wyoming (Gazin, 1956). A few other specimens of Protoselene have been identified from Tiffanian strata, including fragmentary teeth of P. cf. P. opisthacus from the Torrejonian—Tiffanian overlap zone and early Tiffanian of the Hanna Formation of southern Wyoming (Higgins, 2003), and a dentary with m1–3 of P. opisthacus from Ray's Bonebed (middle Tiffanian) from the Big Bend area of Texas (Schiebout, 1974).

The massive and swollen p4 protoconid of P. ashtoni is its most distinctive feature and separates it from P. opisthacus whose protoconid is only slightly inflated. Also, the p4 of P. opisthacus usually has an anterior basal cusp and a small metaconid, features absent in RAM 9047. The p4 is not known for either P. griphus or P. novissimus, so comparisons of these species to P. ashtoni are limited to the M1. The only known M1 of P. novissimus (USNM 21023) is an isolated tooth (Gazin, 1956) with a relatively longer and narrower talonid than the M1 of P. ashtoni. Also, the m1 paraconid of P. ashtoni is positioned more lingually than in either P. novissimus or P. opisthacus. The only known M1 of P. griphus (USNM 15773) has a lingually positioned paraconid like the M1 of RAM 9047, but the M1 of P. ashtoni is larger than the m1 of P. griphus as the length and width of USNM 15773 are 5.3 mm and 4.0 mm, compared to 5.63 mm and 4.33 mm for RAM 9047. Also, a slight accessory cusp is present on the anterior crest of the entoconid in P. griphus (Gazin, 1941), a cusp absent in the only known M1 of P. ashtoni.

Species of Protoselene are poorly known except for P. opisthacus, which can be easily distinguished from P. ashtoni by its lack of a massive p4 protoconid. Based on the single mis known for P. ashtoni, P. griphus, and P. novissimus, differences in size or the position of the paraconid indicate they are distinct species, but the evidence is less robust. The recovery of more complete material of P. ashtoni, as well as, of P. griphus, and P. novissimus, is needed to fully diagnose the three species.

Holotype: RAM 9098, left m1–3 with lingual half of M1 and anterior part of trigonid of m2 missing, and right dentary fragment with m2–3 with m2 missing all of the trigonid except the basal portion of the posterior wall from RAM locality V200510, member 4a, Goler Formation, California (fig. 8B–E).

Referred Specimens: RAM 6724, dentary fragment with m2 and talonid of M1 (fig. 8A), and RAM 6926, right maxilla fragment with M2–3 (referral questionable) (fig. 8F–G) from RAM locality V200001, member 4a.

Etymology: Named in honor of the late Steven Walsh, whose enthusiastic leadership of Goler Formation screenwashing efforts significantly increased the sample of Tiffanian vertebrates from southern California.

Diagnosis: Significantly larger than all other described species of Promioclaenus.

Description: Lower premolars of Promioclaenus walshi are unknown. The M1 of the holotype (RAM 9098) is broken, with only the labial half present, which preserves part of the talonid basin and a large hypoconid and protoconid. A complete but worn M1 talonid is present in RAM 6724 (fig. 8A) and the hypoconid is large and the talonid basin is well developed and opens lingually through a deep talonid notch. A distinct entoconid and hypoconulid are present, with the hypoconulid slightly smaller than the entoconid and placed medially, while the entoconid is positioned on the posterolingual tooth margin. The hypoconulid is positioned closer to the entoconid than the hypoconid and a small postcingulid is developed between the hypoconid and hypoconulid.

RAM 6724 has a slightly worn, but well preserved, quadrate shaped m2 (fig. 8A). The trigonid is anteroposteriorly short because the large metaconid and small paraconid are closely appressed and the paraconid is slightly shifted labially. The paracristid is faintly visible while the protocristid is distinct. The talonid of m2 is very similar to M1, with a large hypoconid and smaller hypoconulid and entoconid, with the entoconid the larger of the two. The partial m2s of the holotype (fig. 8B–E) have the same features as the complete m2 of RAM 6724.

Left and right m3s of the holotype are identical in size and morphology (fig. 8B–E). The protoconid is the largest trigonid cusp and the paraconid and metaconid are subequal in size, with a slight labial shift of the paraconid. The paracristid is more distinct in m3 than in m2. The m3 talonid has a large hypoconid, broad hypoconulid and a smaller ridgelike entoconid positioned directly adjacent to the hypoconulid, separated by a slight constriction in the talonid wall. The talonid basin is narrow, but deep, and opens lingually through a small talonid notch.

A maxilla fragment with right M2–3 (RAM 6926) was found in a concretion at the same site as RAM 6724 and may represent the upper dentition of Promioclaenus walshi. The maxilla fragment is crushed making it difficult to photograph M2–M3 in occlusal view (fig. 8F), so a drawing is provided (fig. 8G). The M2 is quadrate in outline and has moderate to heavy wear on all major cusps. The protocone is the largest cusp and the paracone is slightly larger than the metacone. A centracrista wear facet connects the metacone and paracone, as do preparacrista and postprotocrista wear facets connect the protocone with the paraconule and metaconule. The precingulum and postcingulum are both present, and a well-developed hypocone is positioned on the postcingulum. Before wear, it appears that the apex of the hypocone was positioned directly posterior to the apex of the protocone or slightly labial to it. The lingual cingulum is not developed across the lingual face of the protocone and the labial cingulum is developed only adjacent to the ectoflexus and paracone. A minute parastyle and weak ectoflexus are present, but the metastyle is absent.

The M3 of RAM 6926 has a rectangular shape and is significantly smaller than M2. The protocone is the largest cusp, with the metacone and paracone subequal in size. The metacone of M3 is shifted lingually in comparison to its position in M2. The paraconule and metaconule are also relatively smaller in M3 compared to M2. A distinct hypocone is not evident on the postcingulum of M3.

Discussion and Comparisions: The lower dentition of Promioclaenus walshi is essentially a larger version of the morphology seen in P. lemuroides from the Nacimiento Formation of New Mexico described by Matthew (1897, 1937). Compared to the holotype of P. lemuroides, lower molars of P. walshi are consistently larger (table 9). In addition to the holotype, 12 dentary and three maxillary fragments of P. lemuroides at the AMNH, were also measured to document the range in dimensions of premolars and molars of P. lemuroides. These results also indicate that P. walshi is significantly larger than P. lemuroides, particularly in lower molar width (table 9). Promioclaenus walshi is also larger than P. pipiringosi, a poorly known species described by Gazin (1956) from early to middle Tiffanian strata in the Bison Basin of Wyoming. The holotype of P. pipiringosi (USNM 20571) is a partial dentary whose molars are similar in morphology and size to those of P. lemuroides (table 9). However, Gazin (1956) also referred a dentary fragment with p2–4 (USNM 21021) from the type locality to P. pipiringosi. If this referral is correct, P. pipiringosi has smaller and less inflated premolars than P. lemuroides (table 9). The lower premolars of P. walshi are unknown, so no comparison can be made with P. pipiringosi or P. lemuroides. Promioclaenus acolytus is the only other species of Promioclaenus. The size of its lower dentition, based on the large sample from Swain Quarry (Rigby, 1980: table 46), is much smaller than P. walshi (table 9).

A second species of Promioclaenus, that is larger than P. pipiringosi, may occur in the Bison Basin. A fragmentary dentary with the posterior portion of m2 and part of the trigonid and roots of m3 (USNM 21020) was noted as possibly representing Litaletes or another species of Promioclaenus (Gazin, 1956). The talonid of the m2 in USMN 21020 has an entoconid, hypoconulid, and hypoconid that have about the same relative size and positioning as those cusps in Promioclaenus. The size of the m2 in USNM 21020 (based on a photograph with scale) shows a length and width of approximately 4.75 mm and 4.00 mm, respectively, 0.3–0.4 mm larger than P. pipiringosi and 0.4–0.5 mm smaller than P. walshi (table 9). Gazin (1956) surmised that if USMN 21020 represented Promioclaenus, it was too large to be P. pipiringosi. USNM 21020 is too fragmentary for detailed comparison to P. walshi, but its smaller size suggests it is not the Goler Formation species.

A partial maxilla with M2–3 (RAM 6926) from RAM locality V200001, the same site that yielded a partial dentary of P. walshi (RAM 6724) could represent the upper dentition of P. walshi. RAM 6926 was compared to the M2–3 of P. lemuroides and similarities include the relative size and position of the protocone, metacone, paracone, and conules, and cingulum development. However, there are significant differences, as RAM 6926 has a more quadrate shaped M2 (like the M1 of P. lemuroides), a relatively larger M2 hypocone, an M3 closer in size to M2 compared to the significantly smaller M3 relative to M2 in P. lemuroides, and an anteroposteriorly broad, rectangular M3, compared to the narrow M3 in P. lemuroides. Thus, M2–M3 of P. lemuroides and RAM 6926 have the same cusp construction and arrangement, but proportional differences in certain cusps and tooth shape cast doubt on whether RAM 6926 represents P. walshi, although these differences might represent characters that distinguish P. walshi from other species of Promioclaenus.

A maxilla with heavily worn M2–3 (USNM 21022) from the Bison Basin that was tentatively referred to P. pipiringosi (Gazin, 1956) has some bearing on this issue. The M2 of USNM 21022 is complete and has a distinct rectangular shape, measuring about 3.2 mm in length and 5.0 mm in width. Although heavily worn, the wear facets indicate that the paracone of USNM 21022 was slightly larger than the metacone and a distinct hypocone was present on the posterior cingulum. Thus, two discernible features of the M2 of USNM 21022 resemble features present in the M2 of RAM 6926, but the shape of the M2 of USNM 21022 is rectangular like P. lemuroides, unlike the quadrate-shaped M2 of RAM 6926. Also, the M3 of USNM 21022 is nearly as large as the M2, more so than in either P. lemuroides or P. walshi. It is uncertain whether USNM 21022 even represents P. pipiringosi, but if it does, then upper molar morphology of Promioclaenus would exhibit significant variation. The two molars preserved in RAM 6926 are about the correct size to represent the upper dentition of P. walshi. But based on differences in upper molar morphology evident in RAM 6926, P. lemuroides, and USNM 21022 (P. pipiringosi?), referral of RAM 6926 to P. walshi is tentative.

Holotype: USNM 20564, right maxilla with P4–M2, and a left maxilla with M1 and M2 thought to be from the same individual from the Bison Basin, Fort Union Formation, Wyoming.

Referred Specimens: UCMP 69122, parts of both dentaries with right m2 and left p4–m2 from UCMP locality V65710, member 4b(?); RAM 9659, RM1 from RAM locality V98012, member 4b (fig. 11C); RAM 6722, right M1 (fig. 11D), RAM 7205, right dP4 (fig. 11I), RAM 7245, maxilla fragment with left dP3–4 (fig. 11H), RAM 7208, left dP4, RAM 9725, right dP4, RAM 6721, damaged right m2, RAM 6723, right dentary fragment with m2–3 (fig. 9E) from RAM locality V200001, member 4a; RAM 7248, left maxilla fragment with P4–M1 and damaged M2 from RAM locality V200304, member 4a (fig. 11B); RAM 9019, right m3 from RAM locality V200508, member 4a; RAM 9022, left dentary with damaged dp4 and erupting m1, RAM 9672, left maxillary with P3–4 (fig. 11A) from RAM locality V200510, member 4a; RAM 9025, right M1, RAM 9021, right M2 (fig. 11F), RAM 9046, left dentary fragment m 1–3 (fig. 9D) from RAM locality V200603, member 4a; RAM 9020, associated left M2 and M3 (fig. 11G); RAM 9024, left dentary with m2–3 (fig. 9F), RAM 9023, right dentary with p3–m2 and anterior portion of m3 trigonid (fig. 9A–C) from RAM locality V200612, member 4a; RAM 9045, left dp4 from RAM locality V200704, member 4b (fig. 9G); RAM 10290, right dentary fragment with M1 and m2 trigonid and partial talonid, RAM 10291, left m3, RAM 10292 left dentary with c-m3 and right dentary with il—2, and c-m1 from RAM locality V200802, member 4a (fig. 10A-C); RAM 15000 right maxilla fragment with dP4-M2 from RAM locality V201122, member 4a (fig. 11E).

Description and Discussion: This medium-sized phenacodontid, referred to Phenacodus cf. P. bisonensis, is the most common mammalian species found by surface collecting in the Goler Formation. Included in P. cf. P. bisonensis is UCMP 69122, a concretion with parts of both dentaries previously referred to Tetraclaenodon puercanensis (West, 1970, 1976: pl. 120, 1–2). Found by a hiker in 1954, exact provenance of UCMP 69122 is uncertain, but based on general location information (West, 1970), it likely came from member 4b. In the past two decades, multiple specimens of the upper dentition of Phenacodus were recovered from member 4a and member 4b and these specimens have well-developed molar mesostyles and large P4 metacones, features that distinguish Phenacodus from Tetraclaenodon (Thewissen, 1990). Lower dentitions of Phenacodus and Tetraclaenodon are often difficult to distinguish (Thewissen, 1990), but if all upper dentition specimens from member 4a and member 4b are Phenacodus, UCMP 69122 almost certainly represents Phenacodus as well.

Features of the lower dentition of Phenacodus cf. P. bisonensis are shown in RAM 9023 (fig. 9A–C), RAM 9046 (fig. 9D), RAM 6723 (fig. 9E), RAM 9024 (fig. 9F) and RAM 10292, an extraordinary complete specimen for the sparsely fossiliferous Goler Formation, with the left and right dentaries of a single individual within a concretion (fig. 10A–C). Based on RAM 10292, crowns of the canines and incisors are missing due to breakage, but these teeth were single rooted. Smaller than i3, i2 probably had a circular-shaped crown based on its root, while the crown of i3 was anteroposteriorly elongate, based on the same line of evidence. Small diastemas separate i2 and i3, and i3 from the much larger canine, which has a circular root in cross section. The p1 is single rooted and its crown is elongate anteroposteriorly, exhibiting a single tall cusp and a miniscule posterior basal cusp on a weakly developed posterior cingulid. In contrast, the posterior basal cusp and posterior cingulid are well developed on the tworooted p2. Diastemas are small between cl and p1, and p1 and p2, and minute between p3 and p4. The p3 lacks a metaconid, but has a large protoconid and small paraconid positioned low and directly anterior to the protoconid, and a very small talonid basin. The p4 metaconid is smaller than the protoconid, and a paralophid connects the small and low paraconid to the protoconid. The talonid of p4 has a relatively narrow basin and a large hypoconid and small entoconid. The m1–m3 of P. cf. P. bisonensis are of basic phenacodontid morphology, as they lack paraconids, but have well-developed metaconids, protoconids, and hypoconids, and smaller entoconids and hypoconulids, and wide talonid basins. The m3 of P. cf. P. bisonensis is significantly smaller than the m2, but it has a larger hypoconulid.

The upper dentition of Phenacodus cf. P. bisonensis consists of two M1s (RAM 9659, RAM 6722), an M2 (RAM 9021), maxillary fragments with P4–M2 (RAM 7248) and dP4-M2 (RAM 15000), an associated M2–3 (RAM 9020), and a maxillary fragment with heavily worn P3–4 (RAM 6972). The P3 of P. cf. P. bisonensis has a large paracone and protocone and smaller metacone and the P4 has a large paracone, metacone, and protocone, and a small parastyle (fig. 11A–B). All upper molars of P. cf. P. bisonensis have well-developed mesostyles and parastyles (fig. 11B–G). The single M3 of P. cf. P. bisonensis (RAM 9020) has a small hypocone (fig. 11G).

The dentition of this midsized species of Phenacodus from the Goler Formation is too small to be P. grangeri, P. magnus, or P. intermedius, and too large to be P. matthewi or Ectocion (tables 10–11) (see Thewissen, 1990: tables A4–A14, A18–A20, A26–A29), and Ectocion lacks a hypocone on M3 (Thewissen 1990). Direct comparison of specimens from member 4a and member 4b to the Bison Basin sample of Phenacodus bisonensis at the USNM, indicates they are more similar to P. bisonensis than any other phenacodontid. Phenacodus vortmani and P. bisonensis are both similar in size to the Goler specimens (table 10) (Thewissen, 1990: tables A15–A17, A21–A25), but the p4 of the Goler sample is more aligned with P. bisonensis. The entoconid is present in 59% of p4s of P. bisonensis, but is absent in all Tiffanian and early Clarforkian p4s of P. vortmani, and P. bisonensis has a relatively narrower talonid basin than P. vortmani (Thewissen, 1990). Goler p4s from member 4a and member 4b have entoconids and relatively narrow talonid basins (figs. 9A–C, 10A–C), which indicates they are not P. vortmani. Goler lower molars are also very similar to those of P. bisonensis from the Bison Basin.

One notable difference between Goler p4s and those from the Bison Basin is paraconids of P. bisonensis are slightly larger and more anteriorly positioned relative to the protoconid in comparison to Goler p4s. The only consistent difference between Goler upper molars and those of P. bisonensis is the mesostyle is usually shifted posteriorly, so part of it is positioned labial to the metacone in P. cf. P. bisonensis (fig. 11B–G), and in P. bisonensis the mesostyle is positioned close to the metacone, but not partly labial to it. Also, the postparacrista and premetacrista usually intersect with the mesostyle at its lingual base in P. cf. P. bisonensis (fig. 11C–G), while the postparacrista and premetacrista intersect with the mesostyle at its apex in P. bisonensis. Because of these differences, the Goler specimens are referred to Phenacodus cf. P. bisonensis.

Seven specimens of the deciduous dentition of Phenacodus were also recovered from the Goler Formation and RAM 15000, a maxilla fragment with dP4–M2 (fig. 11E), is particularly informative as it preserves part of the permanent and deciduous dentition. The M1–M2 of RAM 15000 are identical to M1s and M2s referred to P. cf. P. bisonensis, indicating that the dP4 of RAM 15000 must also be this taxon, which facilitated identification of four isolated dP4s (RAM 7245; fig. 11H, RAM 7205; fig. 11I, RAM 7208, and RAM 9725) as representing P. cf. P. bisonensis. Two dp4s (RAM 9022 and RAM 9045) also represent P. cf. P. bisonensis as the specimens are closely comparable to dp4s of P. bisonensis described by West (1971) and RAM 9022 is from a site that yielded a specimen of the permanent dentition (RAM 9672).

West (1971) described two specimens of the deciduous dentition of P. bisonensis, a dp4 from Bison Basin (USNM 26325; Fort Union Formation, Wyoming) and a dP4 from Douglas Quarry (PU 14634b; Lebo Formation, Montana). Estimates of their dimensions based on graph plots by West (1971: figs. 14, 15) are comparable in size to deciduous teeth of P. cf. P. bisonensis (tables 10–11). RAM 9022 is a partial dentary with dp4 that is missing its crown, so only its approximate size is known (table 10). RAM 9045 is an unworn dp4 that is anteroposteriorly elongate, with a metaconid and protoconid of equal size, a lingually positioned and smaller paraconid, a miniscule cusp labial to the paraconid, and a well-developed and cuspidate cristid obliqua (fig. 9G). The talonid basin of RAM 9045 is large and the talonid has a large entoconid and hypoconid and smaller hypoconulid. RAM 9045 is most similar in morphology and size to USNM 26325 (table 10), the single dp4 referred to P. bisonensis by West (1971).

RAM 7245, a maxillary fragment with dP3–dP4, represents the only known dP3 of P. cf. P. bisonensis (fig. 11H). Triangular in shape, dP3 has a large paracone and protocone, smaller metacone, and a parastyle, paraconule and metastyle, but lacks a metaconule. A dP3 of P. bisonensis was not identified (West, 1971), so the dP3 of RAM 7245 serves as a close proxy. Five dP4s of P. cf. P. bisonensis all have well-developed paraconules and hypocones (fig. 11E, H–I). The single dP4 (PU 14634b) referred to P. bisonensis does not have a paraconule and hypocone, but these cusps are present on deciduous dP4s of the similar sized phenacodontid, P. vortmani (West, 1971). Goler dP4s are similar in approximate dimensions to dP4s of P. bisonensis and P. vortmani, but Goler dP4s are wider than long, unlike both P. bisonensis and P. vortmani (table 11). Also, the mesostyle, and postmetacrista and premetacrista in Goler dP4s are positioned as they are in molars of P. cf. P. bisonensis, while these features in the only known dP4 of P. bisonensis (PU 14634b) have the same orientation as molars of P. bisonensis. These differences are further evidence that this mid-sized Goler phenacodontid referred to Phenacodus cf. P. bisonensis is not conspecific with P. bisonensis, although it resembles it in many ways.

Holotype: AMNH 17191, right dentary with m2–3 from Mason Pocket, Animas Formation, Colorado; see Thewissen (1990) for alternative interpretation of type locality.

Referred Specimen: RAM 7210, right dentary fragment with p4 and outline of base of p3 from RAM locality V200001, member 4a (fig. 12A).

Description: RAM 7210 represents a small phenacodontid with a p3 broken off at the gum line and a well-preserved p4. The p3 was approximately 6.37 mm in length and 3.52 mm in width, and the p4 is 7.62 mm in length and 5.30 mm in width. The paraconid of p4 is low and small and is shifted labially to the midline of the tooth (fig. 12A) and the metaconid and protoconid are well developed with the protoconid positioned more anterior than the metaconid. Trigonid enamel is crenulated only on the labial side of p4 and is slight. The talonid basin of p4 is well developed and a large hypoconid and slightly smaller entoconid are present.

Discussion: The paraconid and size of the p3 and p4 of RAM 7210 are significantly smaller than Goler Formation p3s and p4s referred to Phenacodus cf. P. bisonensis (table 10), indicating that RAM 7210 may represent Phenacodus matthewi, the smallest species of Phenacodus. The holotype of P. matthewi is a dentary with m2–3 (AMNH 17191) which can' be compared directly with RAM 7210, but two other specimens from the type locality (AMNH 56284, AMNH 17193) each have a heavily worn p4. The p4 of AMNH 56284 is 8.02 mm in length and 5.17 in width, similar in dimensions to RAM 7210, but the paraconid of RAM 7210 is significantly smaller than the p4 paraconid of AMNH 56284, which is equal in size to the metaconid or protoconid. Also, trigonid enamel is distinctly crenulated on the labial side and slightly crenulated on the lingual side of the p4 in AMNH 56284, while trigonid enamel is crenulated only on the labial side of the p4 in RAM 7210. AMNH 17193 (right p4, M1, and m3), the holotype of “Phenacodus gidleyi” (Simpson, 1935) now referred to P. matthewi (Delson, 1971; West, 1976; Thewissen, 1990), has a p4 that is 6.91 mm in length and 4.76 mm in width, smaller than either AMNH 56284 or RAM 7210. The paraconid is well developed in AMNH 17193, indicating a closer affinity to AMNH 56284 than RAM 7210. The only other reported p4 of P. matthewi is CM 34292 from the Fort Union Formation in the Wind River Basin of Wyoming (Thewissen, 1990). CM 34292 is about the same size as RAM 7210, but has a large paraconid, like AMNH 17193 and AMNH 56284. Thus, RAM 7210 is similar in size to p4s referred to P. matthewi, but its small paraconid makes referral to the species uncertain.

Holotype: AMNH 17185, right maxilla with M1–2 and labial fragment of M3 from Mason Pocket, Animas Formation, Colorado.

Referred Specimen: RAM 7172, left m3 from RAM locality V94014, member 4a (fig. 12B).

Description and Discussion: RAM 7172 is a slightly worn m3 whose paraconid is absent, but whose protoconid, metaconid, hypoconid, entoconid, and hypoconulid are large (fig. 12B). RAM 7172 is 12.87 mm in length and has trigonid and talonid widths of 8.72 mm and 7.99 mm, respectively.

RAM 7172 is almost identical to the m3 of USNM 21025, a dentary fragment with m3 from the Bison Basin. When describing P. bisonensis, Gazin (1956) noted the presence of a larger phenacodontid and described it as Phenacodus? sp. (large), and included USNM 21025 in it. Later, Thewissen (1990) referred USNM 21025 to P. grangeri. RAM 7172 appears to represent P. grangeri, but m3s of P. grangeri cannot be reliably distinguished from P. intermedius based on size or morphology, or from P. magnus based on size (m3 of P. magnus is unknown) (Thewissen, 1990). Thus, RAM 7172 is referred to Phenacodus cf. P. grangeri until more complete material of this taxon is recovered.

Holotype: AMNH 4824, left dentary with p4 and fragment of m1, roots of p3 and m3, and alveoli for p1–2 from Alkali Creek, Wind River Basin, Wyoming.

Referred Specimen: RAM 7253, rp4 from RAM locality V200307, member 4d (fig. 12C).

Description: RAM 7253 is a slightly worn p4 (fig. 12C), that is 10.2 mm in length, with trigonid and talonid widths of 6.5 mm and 6.4 mm respectively. RAM 7253 has a large protoconid and metaconid that are the same size and height, but has two smaller cusps in the position of the paraconid. The labially positioned paraconid cusp is twice the size of the lingually positioned cusp. The talonid of RAM 7253 is wide, and a large hypoconid and small entoconid are present.

Discussion: An isolated p4 of Phenacodus is difficult to identify to species, but size is helpful as RAM 7253 is too small to be P. intermedius, P. magnus, or P. grangeri, and too large to be P. matthewi. The length and width of RAM 7253 approach the maximum length (10.3 mm) and width (6.8 mm) of p4s of P. bisonensis (Thewissen, 1990: tables A15–A17). RAM 7253 is also similar to P. vortmani, as p4s of P. vortmani from late Paleocene strata range from 9.1– 9.8 mm in length and 5.7–7.1 mm in width (Thewissen, 1990: table A–21). Phenacodus bisonensis differs from P. vortmani in having a narrower talonid basin on p4 with an entoconid commonly present, a cusp commonly absent in P. vortmani (Thewissen, 1990). The p4 of RAM 7253 has a wide talonid and a small entoconid (fig. 12C). Presence or absence of the p4 entoconid does not distinguish P. vortmani and P. bisonensis, but the broad talonid basin of RAM 7253 shows a greater affinity to P. vortmani. Thus, RAM 7253 is tentatively referred to Phenacodus cf P. vortmani based on its wide p4 talonid.

Holotype: RAM 7171, right dentary with p4–m2 and part of trigonid of m3 from RAM locality V200120, member 4b, Goler Formation, California.

Referred Specimens: RAM 6506, rm2, RAM 6417, 1p3, UCMP 131790, 1m2, UCMP 44761, left dentary fragment with m3, UCMP 49487, rm3 fragment from RAM locality V94014, member 4a.

Discussion: UCMP 44761 was the first mammalian specimen described from the Goler Formation and was referred to cf. Conacodon entoconus (McKenna, 1955). Study of additional isolated teeth of this taxon recovered from the same site as UCMP 44761 resulted in revised referrals: “unnamed genus and species of anisonchine periptychid” (McKenna, 1960: 12) and “unnamed genus and species of conacodontine periptychid” (McKenna et al., 1987: 36). In 2001, J. Honey collected a partial dentary with multiple teeth (RAM 7171) that served as an adequate holotype for description of the new genus and species, Goleroconus alfi (McKenna et al., 2008: figs. 3–6), which is only known from the Goler Formation.

Referred Specimen: RAM 6507, talonid of M1 (possibly m2) from RAM locality V94014, member 4a (fig. 12D).

Description and Discussion: Mesonychids are among the rarest mammals in North American Paleogene faunas (O'Leary and Rose, 1995) and the only specimen recovered from the Goler Formation (RAM 6507) is a well-preserved talonid of a lower molar of Dissacus, probably an m1 (fig. 12D), which is 5.50 mm in width. Based on size, RAM 6507 cannot be referred to a specific species with confidence as Dissacus navajovius, D. argenteus, D. willwoodensis, and D. serior are about the same size, and D. praenuntius is only about 15% larger than the other four (O'Leary and Rose, 1995). The only morphological feature present in RAM 6507 that may be taxonomically significant is the nearly symmetrical shape of the talonid caused by equal sloping of its labial and lingual surfaces (as seen in distal view). This condition is similar to mis of D. navajovius (O'Leary and Rose, 1995: fig. 3).