OUP user menu

Deciduous dentition of Didymictis (Carnivoramorpha: Viverravidae): implications for the first appearance of “Creodonta”

Shawn P. Zack
DOI: http://dx.doi.org/10.1644/11-MAMM-A-245.1 808-817 First published online: 28 June 2012


Deciduous teeth of Carnivoramorpha can differ substantially from their permanent structural and functional counterparts, particularly dP4. As a consequence, isolated carnivoramorphan deciduous teeth have repeatedly been misidentified as permanent teeth of different, often unrelated taxa. Due in part to this potential for misidentification, the deciduous dentitions of basal carnivoramorphans remain poorly documented. This study describes and illustrates the 1st known deciduous premolars of a member of the basal carnivoramorphan family Viverravidae, with dP3–4 and dp3–4 of the early Eocene Didymictis protenus (Cope, 1874) documented by specimens associated with permanent teeth. The morphology of these specimens permits a reconsideration of the affinities of 2 mid-Paleocene taxa, Deltatherium durini Van Valen, 1978 and Prolimnocyon macfaddeni Rigby, 1980, that have been proposed to represent the earliest known members of the family Hyaenodontidae and the order “Creodonta.” Both species are founded on viverravid deciduous premolars. Reidentification of these taxa eliminates the Paleocene record of hyaenodontids from North America.

Key words
  • Carnivoramorpha
  • Creodonta
  • deciduous dentitions
  • Eocene
  • Hyaenodontidae
  • Paleocene
  • Viverravidae

Deciduous premolars of the mammalian order Carnivor-amorpha, which includes most extant carnivorous placentals and their closest fossil relatives, present a persistent challenge to identification. As in other eutherians, carnivoramor-phan posterior deciduous teeth (dP3–4/dp3–4) are most similar structurally, not to their direct permanent successors (P3–4/p3–4, respectively), but to more posterior permanent teeth (P4-Ml/p4-ml, respectively), reflecting similar functional roles. In particular, the teeth in the posteriormost pair of deciduous premolars (dP4 and dp4) function as molars early in ontogeny and, accordingly, are much more complex than their direct permanent successors (P4 and p4). Even relative to their structural equivalents, however, the posterior deciduous premolars remain distinct. They are smaller in size, reflecting their presence earlier in ontogeny, and they show consistent morphologic differences in being lower-crowned and less transverse. In particular, crowns of eutherian deciduous premolars are typically lower and narrower than their permanent equivalents (Butler 1952). These differences between the morphology of the deciduous and permanent dentitions can make it difficult to recognize when deciduous and permanent specimens belong to a single taxon.

In addition to these more widespread contrasts, carnivor-amorphan deciduous premolars, particularly dP4, show additional differences from their permanent counterparts, reflecting the distinctive functional organization of the permanent dentition in Carnivoramorpha. Members of the clade generally restrict carnassial shear to the P4/ml pair in the adult dentition, with more distal portions of the cheek dentition adapted to crushing or grinding rather than shearing (Butler 1946; Van Valkenburgh 1999). As a consequence, Ml lacks a well-developed metastylar blade to shear against m2, and the base of the postmetacrista is situated at the distobuccal corner of the crown. In contrast, dP4 may form part of a transitory carnassial pair with ml before its replacement by P4 (Butler 1946). In such taxa, dP4 has a much more prominent metastyle than Ml, giving the teeth markedly different outlines. Although dP4 and Ml of a single eutherian taxon are typically distinct due to size and proportional differences,, the prominent metastyle on dP4 in some carnivoramorphans further distinguishes it from the morphology of Ml, increasing the likelihood of misidentification. This has resulted in instances of misidentification of carnivoramorphan deciduous premolars as permanent teeth of other groups of carnivorous mammals such as creodonts and metatherians. For instance, Shotwell (1968) misidentified an isolated carnivoran dP4 as a metatherian molar, extending the record of Tertiary metatherians in North America by several million years (Korth 1994). Similarly, Russell (1972) described a new species of metatherian, Nanodelphys mcgrewi Russell, 1972, subsequently placed in its own genus, Alloeodectes Russell, 1984, on the basis of what Bryant (1991) identified as an isolated dP4 of the basal canid Hesperocyon Scott, 1890.

Despite this documented potential for misidentification, carnivoramorphan deciduous dentitions are often poorly described and illustrated. This is particularly true of basal carnivoramorphans, members of the families “Miacidae” and Viverravidae. The former family is likely paraphyletic with respect to the carnivoran crown group (Polly et al. 2006; Spaulding and Flynn 2009; Spaulding et al. 2010; Wesley-Hunt and Flynn 2005; Wesley-Hunt and Werdelin 2005), and is recorded from the late Paleocene to the late Eocene of North America and the early to late Eocene of Europe and Asia (McKenna and Bell 1997). To date, only the deciduous dentition of a few European “miacids” have been described (Beaumont 1965; Mathis 1987; Springhorn 1982, 2000). Viverravidae includes the basalmost known members of Carnivoramorpha, forming the sister taxon to Miacidae plus Carnivora (Polly et al. 2006; Wesley-Hunt and Flynn 2005; Wesley-Hunt and Werdelin 2005). Viverravids are known from the late early Paleocene to early Eocene of North America and the early Eocene of Asia (McKenna and Bell 1997; Tong and Wang 2006). Aside from the enigmatic Ravenictis Fox and Youzwyshyn, 1994, known from a single tooth, and other poorly known forms (Fox and Youzwyshyn 1994; Fox et al. 2010; Maclntyre 1966), the earliest representatives of Viverravidae are also the earliest known members of Carnivoramorpha. The viverravid deciduous dentition remains completely undocumented, although the existence of viverravid deciduous premolars has been mentioned in passing in the literature (Fox and Youzwyshyn 1994).

The primary purpose of this work is to document the deciduous dentition of a member of Viverravidae, specifically Didymictis Cope, 1875, on the basis of a number of specimens from the early Eocene (Wasatchian North American Land Mammal Age, hereafter NALMA) Willwood Formation of the central and southern Bighorn Basin of Wyoming. Included among these specimens are several in which associated deciduous and permanent teeth are preserved, allowing unambiguous referral to Didymictis. By documenting the morphology of the deciduous dentition in Viverravidae, these specimens permit a re-evaluation of some enigmatic early Paleocene material. This material, previously described (Rigby 1980;Van Valen 1978) as Deltatherium durini Van Valen, 1978 and Prolimnocyon macfaddeni Rigby, 1980, has been regarded as the earliest record of the extinct mammalian order “Creodonta,” providing an additional instance of a significant range extension on the basis of misidentification of deciduous premolars. The term Creodonta is here used in quotes to reflect the fact that the monophyly of the order is poorly established and it may prove para- or polyphyletic (Morlo et al. 2009; Polly 1994; Zack 2009). Both D. durini and P. macfaddeni have been referred specifically to the family Hyaenodontidae, which otherwise has a latest Paleocene first appearance in Africa and Asia (Meng et al. 1998; Sole et al. 2009). The other family traditionally placed in Creodonta, Oxyaenidae, makes its 1st appearance in the early late Paleocene of North America (Gingerich 1980).

Materials and Methods

Dental measurements follow Gingerich and Winkler (1985), whereas dental terminology is modified from Kay (1977; Fig. 1). Usage of NALMAs follows Lofgren et al. (2004) and Robinson et al. (2004). All measurements were taken to the nearest tenth of a millimeter using dial calipers.

Fig. 1

Schematic views of upper and lower molars showing dental terminology used in this work (modified from Kay 1977): a) right upper molar in occlusal view; b) left lower molar in occlusal view. Arrows indicate directional terms.


Carnivoramorpha Wyss and Flynn 1993, Viverravidae Wortman and Matthew, 1899

Didymictis Cope 1875, Didymictis protenus (Cope 1874), Figs. 25; Tables 1 and 2

Fig. 4

Right dentary of Didymictis protenus (USNM 541994) with dp4 and erupting p3–4 a) occlusal view; b) buccal view; and c) lingual view. Scale bar equals 5 mm.

Fig. 5

Comparisons of deciduous teeth of Didymictis protenus (left column) with their permanent functional counterparts (right column): a) USGS 1452, right dP3 in occlusal view; b) USGS 1452, right dP4 in occlusal view; c) USGS 15941, right dp3 in buccal view; d) USNM 541994, right dp4 in occlusal view; e) same in lingual view; f) USGS 199 right P4 in occlusal view; g) USGS 15491, right Ml in occlusal view; h) USGS 25277, left p4 (reversed) in buccal view; i) USNM 487894, left ml (reversed) in occlusal view; j) same in lingual view. All scale bars equal 5 mm.

View this table:
Table 1

Measurements (mm) of upper deciduous premolars and associated molars of Didymictis protenus. Abbreviations: DW, maximum distal width; L, maximum length; MW, maximum mesial width; TL, maximum trigon length; W, maximum width; USGS, United States Geological Survey.

Specimen no.dP3LdP3WdP4LdP4TLdP4MWdP4DWMILM1MWM1DW
USGS 14529.
USGS 103889.75.8
USGS 1594111.010.514.612.6
USGS 159509.8
View this table:
Table 2

Measurements (mm) of lower deciduous premolars and associated molars of Didymictis protenus. Abbreviations as in Table 1 except: TaW, maximum talonid width; TrH, maximum trigonid height; TrL, maximum trigonid length; TrW, maximum trigonid width; USGS, United States Geological Survey; USNM, United States National Museum of Natural History, Department of Paleobiology, Washington D.C.

Specimen no.dp3Ldp3Wdp4Ldp4TrLdp4TrHdp4TrWdp4TaWm2Lm2TrWm2TaW
USGS 145210.
USGS 102949.
USGS 15941>
USNM 52157410.
USNM 5216854.6
USNM 5419949.65.6>

Horizons and localities.—All specimens are from United. States Geological Survey, Denver registry, Denver, Colorado (USGS; now housed at United States National Museum of Natural History, Department of Paleobiology, Washington D.C. [USNM]) and Yale Peabody Museum, Yale University, New Haven, Connecticut (YPM) localities in the Willwood Formation, Wasatchian NALMA (early Eocene), Wyoming. For more detailed locality information, see Bown et al. (1994).

Referred specimens. —From USGS locality D-1230: USGS 1452, right maxilla with dP3–4, Ml (erupting), associated left dentary with dp4. From USGS locality D-1346: USNM 521574, right dentary with p3 (encrypted), dp4. From USGS locality D-1467: USGS 10388, right dP3. From USGS locality D-1583: USGS 10294, right dentary with p3 (encrypted), dp4, ml (encrypted); USGS 15941, right maxilla with P2–4 (erupting), dP3–4, M1, associated right dentary with p3–4 (encrypted), dp3–4, m2, associated right cl (possibly unerupted); USGS 15950, right maxilla with dP4, Ml (erupting). From USGS locality D-1943: USNM 521685, right dentary with p4 (encrypted), dp4 (talonid), and ml. From YPM locality Y-192: USNM 541994, right dentary with p3–4 (erupting), dp4. From YPM locality Y-192: USNM 487894, left ml, associated right dentary with p2, p3–4 (erupting). Of the specimens from D-1583, it is possible that USGS 10294 and 15950 derive from the same individual, as their preservation is similar, and they appear to be at a similar stage in the development of the permanent dentition.

Description.—Vom specimens (USGS 1452, 15941, USNM 521685, 541994) preserve deciduous premolars and identifiable permanent teeth in a single dentition, collectively permitting confident identification of dP3–4 and dp3–4 in D. protenus. Other deciduous teeth are referred on the basis of the comparison with these four specimens. Because the deciduous dentition of Viverravidae has not previously been reported, full descriptions are provided here, followed by comparisons with functionally equivalent permanent teeth.

The morphology of dP3 is strikingly elongate and narrow, with a very elongate ectoloph and reduced protocone (Fig. 2). The buccal cingulum is limited to a few traces beneath the metastyle. The parastyle is a large, low, conical cusp, directly mesial to the paracone, but not connected to it in unworn specimens. The transversely compressed and somewhat distally inclined paracone is by far the largest cusp. The metastyle is elongate mesiodistally and transversely compressed, but is somewhat shorter than the paracone. It is separated from the postparacrista by a well-developed carnassial notch. The combined postvallum blade is oriented primarily longitudinally, with the metastylar portion somewhat more transverse than the postparacrista. The protocone is subequal in size and shape to the parastyle and is positioned distolingual to the paracone. A very short preprotocrista terminates against the lingual wall of the paracone. The better-developed postprotocrista runs distobuccally to the base of the paracone, where it joins the distal cingulum. Conules, trigon basin, and a mesial cingulum are absent. The distal cingulum extends along the base of the metastyle, terminating at the distobuccal corner of the crown.

Fig. 2

Right maxilla of Didymictis protenus (USGS 1452) with dP3-4 and erupting Ml: a) occlusal view; b) buccal view; and c) lingual view. Scale bar equals 5 mm.

On dP4, the buccal cingulum is prominent between the stylocone and the distal margin of the crown (Fig. 2). The narrow stylar shelf is best developed between the stylocone and paracone mesially and the metacone and metastyle distally. An asymmetric ectoflexus is present, with its deepest point buccal to the distal half of the paracone. The stylocone is slightly taller than the parastyle, but both are low and transversely oriented, taking the form of crests. The stylocone is directly buccal to the mesial edge of the paracone and separated from the more mesial parastyle by a groove. Lingually, the stylocone meets the preparacrista at a right angle, whereas the parastyle joins the preparaconule crista. The buccolingually compressed paracone and metacone are small and separated to their bases. The paracone is slightly taller than the metacone, but the latter is more elongate. The distobuccally oriented metastyle is low and elongate, projecting much farther buccally than the stylocone and parastyle. The metastyle is separated from the postmetacrista by a carnassial notch. The conules are small and positioned immediately lingual to the paracone and metacone. A well-developed preparaconule crista extends buccally and slightly mesially to the parastyle. The crest is flush with the preprotocrista, and the 2 crests form a continuous shearing surface. Weak internal conular cristae are present, directed toward the bases of the paracone and metacone, but the postmetaconule crista is absent. The mesiodistally elongate protocone is larger than the paracone and metacone and subequal to them in height. Its apex is slightly distal to the apex of the paracone, and its lingual margin is nearly vertical. Narrow mesial and distal cingula are present, beginning well buccal to the margin of the crown. The mesial cingulum terminates at the base of the crown between the paraconule and paracone, whereas the distal cingulum extends to the level of carnassial notch. There is no development of a hypocone.

The morphology of dp3 is strongly elongate, narrow, and low (Fig. 3). A low mesial accessory cusp is present, aligned with the protoconid. The cusp is incomplete in USGS 15941, but it was large and likely conical. The apex of the protoconid is worn, but it was a low, elongate cusp, with low but salient crests descending mesially and distally from its apex. The distal crest enters the talonid, aligned with a crest on the apex of the principal talonid cusp. The talonid is dominated by this large, low, elongate cusp. The talonid is subequal in width to the protoconid, and the principal cusp is nearly median in position, although its lingual slope is shallower than its buccal slope. There is no talonid basin. The talonid is bounded by low buccal and lingual cingulids, which begin as folds on the distal slope of the protoconid, becoming more salient distally. The cingulids slope upward distally and meet at the distal margin of the talonid, forming a small secondary talonid cusp that is lower than the principal cusp.

Fig. 3

Right dentary of Didymictis protenus (USGS 15941) with dp3–4, encrypted p3–4, and m2: a) occlusal view; b) lingual view; and c) buccal view. Scale bar equals 5 mm.

The trigonid of dp4 is open (Figs. 3 and 4), with the paraconid and metaconid separated all the way to their bases. The mesiodistally compressed paraconid is somewhat lower than the metaconid, and its apex is slightly buccal to the lingual margin of the crown. It is topped by a flat paracristid that dips buccally toward the paracristid notch. The primary orientation of the paracristid is transverse. The protoconid is the tallest trigonid cusp, but it is not dramatically taller than the metaconid. The protoconid portions of the paracristid and the lateral protocristid are nearly vertical and end at closed carnassial notches. The metaconid is united with the protoconid for most of its height. Like the paraconid, it is mesiodistally compressed, and it is topped by a short, nearly transverse medial protocristid.

The talonid is much lower than the trigonid, but subequal in width. The largest cusp, the hypoconid, is low and triangular in occlusal view, with its apex well mesial to the distal margin of the crown. Its lingual surface slopes gently toward the center of the talonid basin. The cristid obliqua contacts the base of the trigonid below the protocristid notch. The postcristid is shorter than the cristid obliqua, running distolingually from the hypoconid apex to reach the base of the hypoconulid. The latter is the smallest and lowest talonid cusp. It has a median position, opposite the postcristid notch. Its apex is continuous with the postentocristid, but there is no buccal contribution to the postcristid. Instead, a crest runs buccally to form a weak distal cingulid that turns mesially to end against the base of the hypoconid. The entoconid is a large, triangular cusp, somewhat lower than the hypoconid, with its apex at the distolingual corner of the talonid. The postentocristid runs buccally and slightly distally from the entoconid to reach the hypoconulid, whereas the pre-entocristid runs mesially to the base of the metaconid, closing the talonid basin lingually. Near the apex of the entoconid there is a low entoconulid on the pre-entocristid. Aside from the small distal cingulid, there is a weak, horizontal mesial cingulid at the base of the trigonid, between the protoconid and paraconid.

Comparisons with permanent teeth.—As a whole, deciduous premolars of Didymictis differ from their permanent equivalents in being substantially smaller; in being more elongate, narrower, lower crowned; and in having thinner enamel. The latter feature is most pronounced on the largest cusp on each tooth (paracone or protoconid). In addition to size and proportions, dP3 differs from P4 in having a weaker parastyle, shorter metastyle, and a much smaller protocone positioned distal, rather than mesial, to the paracone (Figs. 5a and f). The relatively small size of the protocone also makes this tooth particularly narrow in comparison with its permanent counterpart.

The most notable difference between dP4 and Ml is the presence of a well-developed metastylar blade on dP4, whereas the metastyle is essentially absent on Ml (Figs. 5b and g). Additionally, the metacone, which supports the mesial portion of the postvallum blade on dP4, is much larger relative to the paracone than on Ml. The metacone of Ml is longer and only slightly lower than the paracone on dP4, but is much shorter and lower on Ml, being perhaps one-quarter the volume of the paracone of dP4. The parastyle is better developed on M1, with two cuspules present mesial to the stylocone that together form a broad, triangular shelf that projects well mesial to the remainder of the crown. This contrasts with the single cusp on dP4, which is closely appressed to the stylocone. The entire parastylar lobe has a stronger buccal projection on Ml than on dP4. Lingually, the conules are smaller on dP4 and the protocone is more mesiodistally compressed and slightly more mesially positioned. Its apex is almost at the lingual margin of the crown, whereas on Ml, the apex of the protocone is inset buccally, giving the lingual margin of the crown a sloping, rather than vertical, aspect in mesial or distal view. The cingula do not differ dramatically, although the lingual portions of the mesial and distal cingula are weaker on dP4 than on M1.

Compared with p4, dp3 is much narrower and lower crowned, particularly the protoconid, whereas the mesial accessory cusp is more prominent (Figs. 5c and h). The protoconid also differs in having its distal crest in a median position, rather than buccally as on p4. This corresponds to a difference in the position of the mesial talonid accessory cusp: centrally positioned on dp3, but on the buccal margin of the p4 talonid. The cusp is also lower and more completely separated from the protoconid on dp3. The distal accessory cusp on dp3 is taller relative to the same cusp on p4 (or cusps: in some individuals p4 has 2 distal cuspules). The talonid is broader lingually on dp3 than on p4. In the former, the lingual margin of the talonid is flush with that of the protoconid, whereas in the latter, the protoconid extends farther lingually. Finally, the buccal cingulid on p4 is more extensive, extending mesially around the base of the protoconid, rather than ascending its distal slope.

The trigonid of dp4 is more open than on ml, with its paraconid apex positioned more mesially and buccally (Figs. 5d, e, i, and j). The paraconid of dp4 is also substantially lower than the metaconid, rather than subequal as on m1. Both cusps are much taller relative to the protoconid on dp4 than on m1, and the protoconid is generally much larger relative to the other trigonid cusps on m1. Overall, the trigonid of dp4 is smaller relative to the talonid; on dp4, the trigonid and talonid are subequal in width, whereas the trigonid is substantially broader on ml. Talonid morphology is essentially identical, although the hypoconulid is less well differentiated from the entoconid on ml. Cingulids are better developed on the buccal aspect of ml than dp4. The mesial cingulid is stronger, whereas the distal cingulid extends mesially around the base of the hypoconid to reach the hypoflexid.

Comparisons with Deltatherium durini and Prolimnocyon macfaddeni.—In addition to permanent dentitions of Didymictis protenus, comparisons were also made with Deltatherium durini and Prolimnocyon macfaddeni. For D. durini, comparisons were made with the type and only known specimen, American Museum of Natural History, New York (AMNH) 102161 (Fig. 6a) described as Ml by Van Valen (1978). For P. macfaddeni, comparisons were made with the type, AMNH 100640a (Fig. 6b), and a principal referred specimen, AMNH 100639, described as M2 and ml, respectively, by Rigby (1980). Unfortunately, both the type and an additional referred specimen of P. macfaddeni, AMNH 100640b (described as Ml by Rigby), are missing from the collections of the AMNH. Although AMNH 100640a is well illustrated by Rigby (1980) and I have been able to examine a cast of this specimen, no photographs or casts of AMNH 100640b are available.

Fig. 6

Line drawings of Paleocene viverravid deciduous premolars in occlusal view: a) type of Deltatherium durini AMNH 102161, left dP4 (reversed); b) type of Prolimnocyon macfaddeni, AMNH 100640a, right dP4. Scale bars equal 1 mm.

The type specimens of both species are closely similar to dP4 of Didymictis protenus, sharing the following features: paracone and metacone low and well separated basally; 2 cusps in the parastylar region, a buccally directed stylocone, and a more mesially positioned parastyle; elongate, salient metastyle separated from the metacone by a carnassial notch; asymmetric ectoflexus that is deepest buccal to the distal half of the paracone; conules separated from the bases of the paracone and metacone; paraconule flush with the preprotocrista; mesiodistally elongate protocone; protocone apex positioned mesial to centrocrista notch; mesial and distal cingula present but not meeting lingual to protocone; hypocone absent. AMNH 100639 shares numerous features with dp4 of D. protenus: paraconid erect, lower than other trigonid cusps and well-separated from them; protoconid only modestly larger than paraconid and metaconid; talonid size and width subequal to trigonid; hypoconid large, with apex centrally positioned; distinct postcristid present; hypoconulid small but distinct and median in position; entoconid well separated from hypoconulid, at lingual margin of talonid.

Dental eruption.—Three specimens provide limited evidence of the sequence of dental eruption in Didymictis. In USGS 1452, Ml is present and almost fully erupted, whereas dP3–4 and dp4, at the least, are retained. In USGS 15941, M1 is fully erupted, as is m2 (m1 is missing), whereas dP3–4, dp3–4 remain in place, with P2–45and p3–4 just beginning to erupt beneath them. Finally, USNM 487894 preserves an erupted p2 and partially erupted p3–4, with both teeth at an equivalent stage of eruption. In sum, available evidence indicates that p2 of Didymictis erupted before p3–4. p3 and p4 erupted simultaneously, and Ml, m 1–2 erupted before the eruption of P2–4, p3–4. Partial eruption sequences for the upper and lower dentitions can be written as follows (backslashes indicate teeth that either erupt together or whose relative sequence cannot be determined):

Embedded Image

Eruption sequences in modern carnivoramorphans are summarized by Slaughter et al. (1974). Many carnivoramorphans are characterized by a delay in the eruption of p4 until after the eruption of p3. In contrast, P4 erupts before P3. In most other eutherians, both P4 and p4 erupt before their 3rd premolar counterparts. Slaughter et al. (1974) interpreted the carnivoramorphan pattern as an adaptation to maintain functionality of the deciduous carnassial pair of dP3 and dp4 until after full eruption of the permanent carnassials, P4 and m1. The eruption sequence of p4 in Didymictis is consistent with this hypothesis, as this tooth erupts with, rather than before, p3. However, P4 of Didymictis shows a similarly late eruption coincident with P3. This may reflect the particularly strong metastylar blade on dP4 of Didymictis, which would have allowed this tooth to form part of an effective transitory carnassial pair with ml.


Viverravids are widely distributed in the North American Paleocene and Eocene faunas, occurring in most well-sampled faunas between the Torrejonian (late early Paleocene) and Bridgerian (early middle Eocene) NALMAs (Flynn 1998). As such, it would not be surprising if additional viverravid deciduous teeth have been recovered, although they may not have been recognized as such. In fact, on the basis of comparisons with the material described above, 2 problematic Paleocene taxa appear to represent misidentified deciduous teeth. Van Valen (1978) named and briefly described a new species of the problematic eutherian Deltatherium Cope 1881, D. durini, on the basis of an isolated sectorial upper molariform tooth, interpreted as an Ml, from an unknown locality in the Paleocene of the Crazy Mountains Basin of Montana. Subsequently, Rigby (1980) described a new species of the hyaenodontid “creodont” Prolimnocyon Matthew 1915, P. macfaddeni, from the Torrejonian of Swain Quarry in the Washakie Basin of southern Wyoming. The type of P. macfaddeni is also an isolated sectorial upper molariform tooth that Rigby interpreted as an M2. Rigby assigned 4 other isolated teeth to the hypodigm of P. macfaddeni with varying degrees of confidence. Subsequently, Williamson (1996) argued that the types of D. durini and P. macfaddeni are nearly identical and synonymized them as Prolimnocyon durini. Lucas and Kondrashov (2004) upheld both that synonymy and the assignment of D. durini to Prolimnocyon, noting that the type specimen represents a taxon that is much more sectorial than Deltatherium.

If correctly referred to Prolimnocyon, D. durini and, particularly, P. macfaddeni would represent the earliest records of both Hyaenodontidae and of Creodonta as a whole. Otherwise, the oldest record of Creodonta is the oxyaenid Tytthaena parrisi Gingerich 1980 from the early late Paleocene of North America (Gingerich 1980), whereas the oldest records of Hyaenodontidae are from the latest Paleocene, Prolimnocyon chowi Meng et al., 1998 from Inner Mongolia, China and Lahimia selloumi Solé et al.2009 from Morocco. There is no other evidence of Hyaenodontidae in the North American Paleocene, with the family making its 1st appearance in the earliest Eocene (Gingerich 1989; Strait 2001; Zack 2011) as part of a wave of presumed immigrants that otherwise includes the earliest North American representatives of Artiodactyla, Euprimates, and Perissodactyla (Gingerich 1989; Hooker and Dashzeveg 2003; Krause and Maas 1990; Smith et al. 2006). However, at least 1 author has questioned referral of Prolimnocyon macfaddeni to Creodonta. Gunnell (1998) instead suggested potential affinities to Palaeoryctidae or Pantolestidae.

Comparisons of the type of D. durini and P. macfaddeni with deciduous dentitions of Didymictis indicate that both Paleocene species are founded on viverravid deciduous premolars, in keeping with the Paleocene age of these taxa. Unlike hyaenodontids, the presence of Paleocene viverravids in North America is well established, with numerous genera and species, most known from relatively complete, associated cheek dentitions (Gingerich and Winkler 1985; Maclntyre 1966; Meehan and Wilson 2002; Polly 1997). In the case of P. macfaddeni, 3 viverravids, Bryanictis microlestes (Simpson, 1935), Protictis haydenianus (Cope, 1882), and Simpsonictis jaynanneae Rigby, 1980, have been described from Swain Quarry (Rigby 1980). Of these, specimens of P. haydenianus are of appropriate size to represent the permanent dentition of Prolimnocyon macfaddeni, with published measurements (Rigby 1980) indicating that the size difference between P. macfaddeni and Protictis haydenianus is comparable with the size difference between deciduous and permanent teeth in associated specimens of Didymictis protenus (Tables 1 and 2). In contrast, both B. microlestes and S. jaynanneae are much too small to be conspecific with Prolimnocyon macfaddeni, as permanent teeth of both taxa would be smaller than their deciduous precursors. Accordingly, P. macfaddeni is herein considered a subjective junior synonym of Protictis haydenianus. I interpret the type specimen, an upper molariform tooth (AMNH 100640a), as a dP4, whereas a lower molariform referred to this species (AMNH 100639) as a dp4 of P. haydenianus.

The identity of Deltatherium durini is less clear. Although Williamson (1996) claims that the type is identical to that of Prolimnocyon macfaddeni, the teeth of these 2 forms differ substantially in proportions, with D. durini being less transverse, more similar to dP4 of Didymictis protenus. This makes it unlikely that P. macfaddeni and Deltatherium durini represent the same taxon. Additionally, the type of D. durini probably derives from the Melville Formation (Van Valen, 1978), indicating a Tiffanian age, which would be substantially younger than the Torrejonian faunas in which unambiguous records of Protictis haydenianus occur (Flynn 1998). Instead, D. durini is likely conspecific with either a late species of Protictis (Matthew 1937) such as P. agastor Gingerich and Winkler1985, or an early species of Didymictis such as D. dellensis Dorr, 1952. Until additional materials become available, the species is best treated as an indeterminate representative of Viverravidae.

Regardless of the specific affinities of Deltatherium durini within Viverravidae, reidentification of both this species and Prolimnocyon macfaddeni as viverravids eliminates the only Paleocene North American records of Hyaenodontidae. This supports the view that hyaenodontids originated in the Old World, most likely either Asia or Africa, where Paleocene records of hyaenodontids are clearly founded on permanent teeth, including dentigerous fragments.


Specimens studied here were collected by expeditions led by T. Bown and K. Rose over a span of 3 decades. Collecting efforts by K. Rose have been supported by numerous National Science Foundation grants (most recently IBN-9419776 and EAR-0000941, EAR-0616376) under Bureau of Land Management permit to K. Rose. T. Bown initially separated out deciduous dentitions of variety of Will wood mammals, including Didymictis. R. Fisher and K. Kusumi (Arizona State University) graciously provided access to facilities and equipment. For access to casts and other comparative materials I thank P. Gingerich and G. Gunnell (University of Michigan), K. Rose and R. Dunn (Johns Hopkins University), R. Purdy and R. Emry (USNM), and J. Flynn and R. O 'Leary (AMNH). K. Rose graciously aided in the preparation of Fig. 5. Finally, T. Penkrot and K. Rose provided many helpful discussions, and P. Houde, T. Penkrot, and an anonymous reviewer provided thoughtful comments on earlier drafts of this manuscript.


  • Associate Editor was Neal Woodman.

Literature Cited

View Abstract