SCIURAVIDAE Miller and Gidley, 1918

Kelly, Thomas S. & Murphey, Paul C., 2016, Mammals from the earliest Uintan (middle Eocene) Turtle Bluff Member, Bridger Formation, southwestern Wyoming, USA, Part 1: Primates and Rodentia, Palaeontologia Electronica 7 (8), pp. 1-55 : 17-21

Family SCIURAVIDAE Miller and Gidley, 1918

Remarks. The taxonomy of the sciuravids from the Bridger Formation has a complicated history, similar to that of the Cylindrodontidae (see below). The holotypes of many sciuravid species were originally based on fragmentary specimens, often only a partial dentition or known only from either an upper or lower partial dentition. To further complicate issues, the type localities for the holotypes of many species from the Bridger Formation are either very imprecise or unknown.

Marsh (1971) described the first sciuravid from the Bridger Formation, Sciuravus nitidus (= Sciuravus undans Marsh, 1871 and Colonomys celer Marsh, 1872 ; see Wilson, 1938a for synonymies). This is the most common and best represented sciuravid from the Bridger Formation, now known from numerous upper and lower dentitions plus cranial material (e.g., Matthew, 1910; Wilson, 1938a; Dawson, 1961). Marsh (1872) named three additional sciuravids from the Bridger Formation; Taxymys lucaris , based on a partial maxilla with P3-4 from Henry's Fork (presumably Br3), Tillomys senex based on a partial dentary with m1 from Henry's Fork (presumably Br3), and Sciuravus parvidens based on a partial dentary with m2 presumably from Grizzly Buttes (Br2). Sciuravus parvidens was later reassigned to Tillomys ( Troxell, 1923b; Wilson, 1938b). Marsh (1872) also named a second species of Tillomys , T. parvus , but this taxon was later correctly recognized as belonging to Cylindrodontidae by Wilson (1938b) and transferred to Mysops (see also remarks below on cf. Pareumys sp. ). Subsequent to Marsh (1871, 1872), investigators have questioned whether the upper dental specimens assigned to Taxymys and lower dentition assigned to Tillomys might be cogeneric because they are similar in size and compatible in occlusal morphology. Troxell (1923a) described Pauromys perditus based on a partial dentary with p4-m3 from an undetermined locality in the Dry Creek area ( Dawson, 1968; Gazin, 1976), which includes strata of both the Bridger B and C (Br2 and Br3). Wilson (1938a) described two additional sciuravids from the Bridger Formation; Sciuravus bridgeri based on a partial dentary with p4-m2 from Millersville (presumably Br2) along with referred specimens including a partial dentary with m1-3 and four isolated cheek teeth, and Sciuravus? rarus based on a partial dentary with p4- m1 from an unknown locality in the Bridger Formation. Wilson (1938c) described Taxymys? progressus based on partial maxilla with P3-4 from an unknown locality in the Bridger Formation. Wood (1959b) described Pauromys schaubi based on a partial dentary with the alveolus for p4 and m1-2 from the "red stratum at Twin Buttes" (Br3 or Ui1a?). Dawson (1962) described a partial maxilla with P3-M1 from Twin Buttes (Br3) that she left without generic taxonomic assignment (sciuravid sp.) because it might represent a sciuravid genus already known only by lower dentitions from the Bridger Formation. To summarize, eight sciuravid species have been previously recognized from the Bridger Formation ( Sciuravus nitidus , Sciuravus? rarus , Tillomys senex , Tillomys parvidens , Taxymys lucaris , Taxymys? progressus , Pauromys perditus , and Pauromys schaubi ) along with one partial upper dentition left in open nomenclature. In addition to the above sciuravids from the Bridger Formation, Dawson (1968) described a large sample of isolated rodent teeth from the early Bridgerian Powder Wash Locality of the Green River Formation, Utah, wherein she divided the sciuravid teeth into two groups left in open nomenclature as Sciuravus sp. and Pauromys sp.

Additional sciuravid taxa have also been recognized from Wasatchian and Uintan strata including the following ( Korth, 1984, 1994; Flanagan, 1986; Ivy, 1990; Walton and Porter, 2008): 1) Knightomys Gazin, 1961 , primarily known from the Wasatchian and including seven species (the type species K. senior [ Gazin, 1952, originally assigned to Tillomys ], K. depressus [ Loomis, 1907, originally assigned to Sciuravus , synonyms Microparamys lysitensis Wood, 1962 , and Microparamys cathedralis Wood, 1962 ], K. cuspidatus [ Bown, 1982, originally assigned to Taxymys ], K. reginensis [ Korth, 1984, originally assigned to Microparamys ], K. huerfanensis [ Wood, 1962], K. minor [ Wood, 1965, originally assigned to Dawsonomys ], and possibly K. cremeneus Ivy, 1990 ); 2) four additional species of Sciuravus ( S. eucristadens Burke, 1937 , S. powayensis Wilson, 1940 , S. popi Dawson, 1966 , and probably S. altidens [ Peterson, 1919]); 3) three additional species of Pauromys ( P. exallos Emry and Korth, 1989 ; P. texensis Walton, 1993 , and P. simplex Walton, 1993 ) and 4) Prolapsus Wood, 1973 , including two species (the type species P. sibilatorius Wood, 1973 , and P. junctionis Wood, 1973 ).

The type species of Pauromys , P. perditus , is characterized by the following ( Troxell, 1923a; Wood, 1959b; Dawson, 1968): 1) very small size; 2) an extremely reduced p4 relative to the molars, where the p4 ap is only 56% of the m1 ap; and 3) a transversely expanded mesoconid on m1-3 that is isolated from the protoconid. The lower molars of P. perditus also differ from those of Sciuravus and Tillomys by having the primary cusps less inflated and a more lophate occlusal pattern. Walsh (1997) described Pauromys lillegraveni based on a partial dentary with i-m3 along with a partial maxilla with P4-M1 and a number of isolated teeth from the early Uintan Stadium Conglomerate of southern California. Pauromys lillegraveni exhibits a very reduced p4 (p4 ap = 50% of m1 ap), like that of P. perditus , and its P4 is similarly reduced (P4 ap = 57% of M1 ap). Although other investigators (e.g., Walton, 1993) have suggested that the very reduced p4 of P. perditus may be a character that rises to the level of generic distinction within Sciuravidae, Walsh (1997) went further and adopted a narrower interpretation for the diagnosis of Pauromys sensu stricto, including only P. perditus and P. lillegraveni , confidently in the genus.

Walsh (1997) questioned the previous generic assignment of certain other species referred to Pauromys . He noted that although Dawson's (1968) Pauromys sp. from Powder Wash, which is known only from isolated teeth, does exhibit similarity in size and lower molar occlusal morphology to P. perditus , its p4 and P4 are not quite as reduced as in those of P. perditus or P. lillegraveni . He further noted that Pauromys exallos is known only from isolated teeth and if the referred p4 was correctly assigned by Emry and Korth (1989) to the species, then the species does not have a reduced p4 and can be excluded from the genus; however, Walsh (1997) suggested that the referred p4 may actually represent Microparamys sambucus from the same locality, so that the other molars referred to P. exallos could still possibly represent Pauromys . Lastly, he noted that some of the teeth referred by Walton (1993) to Pauromys texensis and P. simplex more closely resemble those of the eomyid Metanoiamys than those of Pauromys , and if correctly assigned, then these species are not referable to Pauromys . Walton and Porter (2008) partially followed Walsh (1997) in their diagnosis of Pauromys along with additional characters used by Porter (2001) in his analysis of the Sciuravidae , which included the following: 1) p4 reduced (p4 ap <60% of m1 ap; 2) m1 entoconid not connected to mesolophid; 3) m2 mesoconid does not contact hypolophid; and 4) m3 entoconid connects to posterolophid (= posterior cingulid). However, contrary to Walsh (1997), Walton and Porter (2008) tentatively included P. schaubi and P. exallos in the genus.

Based on the size of the p4 alveoli in the holotype of P. schaubi , its p4 is not greatly reduced with an alveolar ap that is about 80% of the m1 ap. Pauromys schaubi further differs from P. perditus by having a doubled m2 mesoconid and more robust lower molar primary cusps, and lacking a lingual extension of the posterior cingulid ( Wood, 1959b). These differences suggest that P. schaubi is probably generically distinct from Pauromys .

The isolated molars that Emry and Korth (1989) referred to Pauromys exallos are significantly larger than those of P. perditus or P. lillegraveni , but exhibit occlusal morphologies very similar to these species, including a lophate occlusal pattern and a transversely expanded mesoconid that is isolated from the protoconid. Whether the single, isolated, heavily worn p4 referred to P. exallos by Emry and Korth (1989) is correctly assigned or represents another taxon (i.e., Microparamys ), as proposed by Walsh (1997), cannot be determined without intact dentitions, but the occlusal patterns of the referred molars do suggest a close relationship with P. perditus and P. lillegraveni .

Walton and Porter (2008) supported Walsh's (1997) analysis of P. texensis and P. simplex and regarded most of the teeth previously assigned by Walton (1993) to these species to actually belong to Metanoiamys . Most recently in a geochronological and taxonomic revision of the early Uintan Whistler Squat Quarry, Campisano et al. (2014) retained P. texensis as valid, but did not mention Walsh's (1997) or Walton and Porter's (2008) referral of P. texensis , at least in part, to Metanoiamys . Walton (1993) was very careful in her analysis of P. texensis and P. simplex and only tentatively assigned premolars to these species. She recognized the difficulty of referring isolated teeth representing different positions to a single species, as have other investigators (e.g., Dawson, 1968).

Although Walton and Porter (2008) tentatively referred P. texensis and P. simplex to Metanoiamys , the assignment of all the referred teeth may be too inclusive. The lower molars of Pauromys , as typified by P. perditus and P. lillegraveni , exhibit a number of characters that differ from those of Metanoiamys . In Pauromys , the metalophid (= metalophulid II) extends lingually from the protoconid to terminate at and slightly posterior of the labial base of the metaconid, whereas in Metanoiamys , the metalophid is usually a continuous cristid with the metalophid forming a smooth arc between the protoconid and metaconid (more complete and joining the metaconid slightly more anteriorly). The ectolophid is less complete with the mesoconid either isolated or nearly isolated (only connected posteriorly by a low cristid to the hypoconid), whereas in typical Metanoiamys , the ectolophid is more complete, usually connecting the mesoconid anteriorly to the protoconid and relatively taller. The hypolophid is incomplete, whereas in Metanoiamys , the hypolophid is stronger, usually forming a complete, continuous cristid joining the hypoconid to the entoconid. The separation of the entoconid from the lingual termination of the posterior cingulid is more distinct in Pauromys . Some of the lower molars that Walton (1993) assigned to P. texensis (e.g., Walton, 1993, figures 4J-O) and P. simplex (e.g., Walton, 1993, figures 8G-I) appear more similar to those of Pauromys than to those of typical Metanoiamys . Differentiating upper molars of Pauromys from those of primitive Metanoiamys is more difficult. The upper molars of Pauromys lillegraveni are characterized by the following ( Walsh, 1997): 1) strong M1-2 anterior cingula; 2) M1-2 lacking an anterocone and protoconule; 3) a complete M1-2 protoloph that extends from the protocone to the anterolingual base of the paracone; 4) a complete M1-2 metaloph connecting the hypocone to the metacone; and 5) an incomplete M1 endoloph, only connecting the mesocone to the hypocone, but complete in M2 connecting the mesocone to the protocone and hypocone. The occlusal morphology of Pauromys sp. from Powder Wash ( Dawson, 1968) is very similar to that of P. lillegraveni . The M1-2 of primitive Metanoiamys are also similar to those of P. lillegraveni and Pauromys sp. from Powder Wash, but have slightly stronger protolophs and metalophs, a greater tendency for the endolophs to be complete with small accessory lophules that extend from the endoloph or mesocone into the central basin. These subtle differences are not surprising because Metanoiamys , the earliest known eomyid, has been postulated to have been derived from a Pauromys -like sciuravid ( Walsh, 1997; Walton and Porter, 2008). This may help explain why some of the upper molars assigned to P. texensis and P. simplex likely represent Metanoiamys instead of Pauromys . Since the lower molars of Pauromys and Metanoiamys appear to be more easily distinguished, and some of the lower molars assigned to the Texas samples of Pauromys do exhibit more similarity to Pauromys rather than primitive Metanoiamys , it is possible that the Texas samples include examples of both Pauromys (e.g., Walton, 1993, figures 6I, 6M, 7C and 8D-E) and Metanoiamys (e.g., Walton, 1993, figures 4A, 6J, 6K, and 7D). Thus, referral of all teeth from Texas assigned by Walton (1993) to P. texensis , P. simplex and Pauromys spp. (taxa left in open nomenclature) to Metanoiamys may not be justified. However, the holotype of P. texensis , a LM1 (TMM 41745-412), and the holotype of P. simplex , a LM1 (TMM 42486-515), appear more similar in occlusal morphology to those of primitive Metanoiamys than to those of P. lillegraveni and P. sp. from Powder Wash. Therefore, we accept Walton and Porter's (2008) reassignment of P. texensis and P. simplex to Metanoiamys , but question the referral of some of the lower molars in the hypodyms of these two species to Metanoiamys . Only with the discovery of intact upper and lower dentitions of both of these species will their taxonomic identities be fully clarified.

Other investigators have noted the similarity of other rodent taxa to Pauromys , which also exhibit a significantly reduced p4, including Wasatchian ' Microparamys ' scopaiodon ( Korth, 1984, 1994; Walton and Porter, 2008) and Bridgerian ' Sciuravus ' rarus ( Korth, 1994) . ' Microparamys ' scopaiodon is known only from a partial dentary with p4 and m2 plus the alveoli for m1, and its p4 and m2 occlusal morphologies are very similar to those of P. perditus and P. lillegraveni . However, the p4 of ' M.' scopaiodon is slightly less reduced than those of P. perditus and P. lillegraveni . A comparison of the relative reduction of p4 to m1-2 and P4 to M1 for Sciuravidae and ' Microparamys ' scopaiodon is presented in Table 3. If one accepts the character state of the p4 ap <60% of m1 ap ( Walsh, 1997; Walton and Porter, 2008) as a diagnostic character of Pauromys , then only P. perditus , P. lillegraveni , P. sp. from Powder Wash, and possibly ' Sciuravus ' rarus would be included in the genus. However, the m1 of ' S.' rarus is significantly larger than those of P. perditus and P. lillegraveni , but similar in size to those of Sciuravus . Moreover, the m1 of ' S. ' rarus differs from those of P. perditus and P. lillegraveni by having more well-defined, inflated primary cusps (less lophate) and a less transversely expanded mesoconid, characters more typical of the lower molars of Sciuravus . Thus until additional, more complete specimens of ' S. ' rarus are available, its phylogenetic relationships to Pauromys or Sciuravus cannot be determined with confidence.

Korth (1984) referred a sample of nine isolated teeth from the late Wasatchian (Wa7) Lost Cabin Member of the Wind River Formation to Pauromys sp. In the Lost Cabin sample, the p4 length is not greatly reduced relative to the m1-2 lengths, as in those of P. perditus and P. lillegraveni , but are within the observed ranges for species of Knightomys ( Table 3). However, the referred teeth are very similar in size and occlusal morphology to those of P. perditus and P. lillegraveni , differing primarily in having lower lophs ( Korth, 1984). Korth (1984) suggested that relatively larger p4 of the older Lost Cabin species would be predicted if there was a progressive reduction of the premolar in the genus through time. However, it could also be argued that the increased loph height and extreme reduction of the p4 seen in later species of Pauromys rises to the level of generic separation from the earlier Lost Cabin species.

The above discussion indicates that after more than 140 years of collecting fossils from the Bridger Formation, the phylogenetic relationships of many of the poorly characterized Bridgerian sciuravids is still controversial. Only with the discovery of much more complete specimens, including intact upper and lower dentitions, will their relationships be fully clarified. We follow the taxonomic scenario presented by Walton and Porter (2008) for the Sciuravidae , with the exception of ' P.' schaubi , which we regard as likely generically distinct from other species of Pauromys .

The larger sciuravid teeth (molars with ap observed range of 1.74-2.05 mm) from the TBM can be assigned with reasonable confidence to previously recognized taxa from the Bridger Formation. However, several potential hurdles arose regarding the generic and specific identification of the small to medium-small sized sciuravid teeth (molars with ap <1.47 mm). First, the sample size of sciuravid teeth from each locality is small. Second, many of the previously recognized sciuravid species are poorly documented so the amount of individual variation in their occlusal morphology and size has not been adequately determined. Lastly, many of the previously recognized sciuravid taxa are known only from either upper or lower partial dentitions. As with any analysis of isolated rodent teeth, the assignment of teeth representing different positions to a single species is difficult, but accomplished best when a very large sample of upper and lower teeth that agree in size and occlusal morphology are available from a single locality. However, this is not the case for the small to medium-small sized sciuravid teeth from the TBM. The best sample came from UCM Locality 92189, which yielded 20 medium-small sized sciuravid teeth (molars with ap observed range of 1.20-1.46 mm) that agree in occlusal morphology and can be reasonably regarded as conspecific. UCM Locality 92189 also yielded one smaller (ap = 1.13 mm) sciuravid upper molar that appears to represent a second species at the locality. Only four other localities (SDSNH Localities 5841, 5842, 5787, and UCM Locality 92189) yielded small to medium-small sized sciuravid teeth, ranging from one to five isolated teeth from each locality. As noted above, the systematic relationships of most of the Bridgerian sciuravids are unclear, so the systematic accounts of the small to medium-small sized sciuravid teeth from the TBM presented below is very conservative, wherein the majority of these teeth are only placed into informal taxonomic groups left in open nomenclature. Hopefully, confident generic and specific assignments for these teeth can be accomplished in the future with the discovery of intact associated upper and lower dentitions from the TBM.