Sphenocoelus, Osborn 1895

Mader, Bryn J., 2008, A species level revision of Bridgerian and Uintan brontotheres (Mammalia, Perissodactyla) exclusive of Palaeosyops, Zootaxa 1837 (1), pp. 1-85 : 30-44

publication ID	https://doi.org/ 10.11646/zootaxa.1837.1.1
persistent identifier	https://treatment.plazi.org/id/03EB87C9-FFE0-DA08-EAFE-FEA8FB3F6D69
treatment provided by	Felipe
scientific name	Sphenocoelus
status

Treatment

Genus SPHENOCOELUS Osborn 1895

= Dolichorhinus Hatcher 1895

= Tanyorhinus Cook 1926

= Dolichorhinoides Granger and Gregory 1943

Type species. S. uintensis Osborn 1895

Included species. S. intermedius ( Osborn 1908) , S. hyognathus ( Osborn 1890) , S. angustidens ( Granger & Gregory 1943, an Asian taxon not discussed in detail in the present paper).

Age. early Uintan

Diagnosis. Medium to large sized (length P1 to M3 approximately 173–287 mm, length M1 to M3 approximately 101–173 mm) dolichorhinine brontothere with extremely long (hyperdolichocephalic) skull. The genus is also distinguished from other dolichorhinine brontotheres by a small, angular horn at the frontonasal boundary over the orbit and by a double suborbital protuberance in which there is a small anterior flange and a large posterior flange. The horn is weak or absent in at least some specimens of Sphenocoelus uintensis , however, and the anterior flange of the suborbital protuberance is weak or absent in certain specimens of S. uintensis and S. intermedius . The condition of the suborbital protuberance is not known with certainty in S. angustidens .

Discussion. The single unifying synapomorphy that consistently unites all specimens of Sphenocoelus within the family Dolichorhininae is the highly unusual hyperdolichocephalic skull ( Mader 1998). Many specimens of Sphenocoelus (spanning all recognized species) also have a small horn over the orbit, but this may be weakly developed or absent in some specimens of the most plesiomorphic species, S. uintensis . Although the presence of a horn is clearly a synapomorphy, its absence in certain specimens limits the use of this character for diagnostic purposes. Furthermore, there are some specimens of the plesiomorphic dolichorhinine brontothere Mesatirhinus that may exhibit an incipient horn ( Osborn 1929; McCarroll et al. 1996).

Similarly, in many specimens of Sphenocoelus the suborbital protuberance is divided into a small anterior flange and a large posterior flange, a character that is clearly lacking in the other two North American dolichorhinine brontothere genera: Mesatirhinus and Metarhinus . The anterior flange is weak or absent, however, in some specimens of S. uintensis and S. intermedius , (and the condition is unknown in S. angustidens ) limiting the usefulness of the character in making identifications.

Finally, in most specimens of Sphenocoelus , the orbit is vertically elongated and angled backwards, which is also a derived condition (compare to other brontotheres and outgroup perissodactyls). However, some specimens of Mesatirhinus have a similar orbital morphology (e.g., UFH V-258), which prevents use of this character as a distinctive synapomorphy for Sphenocoelus .

In my previously published diagnoses of brontothere genera ( Mader 1989; 1998) I have used several descriptive terms to indicate brontothere sizes. These descriptive terms were loosely related to measurements appearing in Mader (1989) of the length of the upper cheek tooth series or the length of the upper cheek tooth series exclusive of the first premolar. The second measurement (length P2 to M3) was the preferred measurement since the first premolar is often missing in brontothere skulls.

Sphenocoelus (= Dolichorhinus ) was described in both of my papers ( Mader 1989; 1998) as being “moderately large-sized” a description that I used for brontotheres in which the length of P2 to M3 was approximately 164–193 mm. It will be noted that, in the present paper, the size is described as “medium to large size,” which reflects a slightly different size range. Based on my previous usage, the length of P2 to M 3 in “medium” sized brontotheres would be approximately 128–165 mm, and in “large” sized brontotheres the length of P2 to M3 would be approximately 195–255 mm. The change in description is necessitated by the relatively diminutive size of S. intermedius (P2 to M3 approximately 154–175 mm) and the substantial size of S. angustidens , which were not taken into account in my previous papers.

It should be noted that I have not personally taken measurements from the type and only known specimen of S. angustidens , although Granger and Gregory (1943) reported that the length of the cheek tooth series (P1 to M3) is approximately 287 mm. Although this is not my preferred measurement, comparison of Granger and Gregory’s measurement against measurements of the cheek tooth series from my own data places the type of S. angustidens in the “large” to “very large” category. Since I had intended the term “very large” for the largest brontotheres (e.g., Brontops and Megacerops , minimal length of P1 to M3 approximately 300 mm), and the type of S. angustidens is slightly smaller than these, the descriptive term “large” seems to be the most appropriate one to apply to S. angustidens .

Among North American specimens of Sphenocoelus there are two general skull morphologies ( Fig. 10 View FIGURE 10 ): a plesiomorphic form with a prominent sagittal crest (similar to that of Eotitanops , the earliest and most primitive known brontothere, and outgroup perissodactyls) and a more derived form in which the cranial vertex has been widened and the sagittal crest lost. In the past, these more derived specimens have been referred to the genus Dolichorhinus (e.g., Riggs 1912; Peterson 1924; Osborn 1929; Granger & Gregory 1943; Simpson 1945; Mader 1989; McCarroll et al. 1996). More recently, however, Mader (1998) recognized Dolichorhinus as a junior synonym of Sphenocoelus ( Osborn 1895) , as did McKenna and Bell (1997), apparently based upon Mader’s conclusions (which were unpublished, but available to them at that time).

Mihlbachler (2005), however, has continued to regard Sphenocoelus as generically distinct from Dolichorhinus , although characters that he regarded as distinctive at the generic level are regarded here as distinctive only at the species level. Of greater consequence, however, is that in Mihlbachler’s phylogenetic analysis (2005), Sphenocoelus did not group with Dolichorhinus as would be necessary if they are both members of a single monophyletic taxon. In my opinion there are several important problems with the characters used in the Mihlbachler study that will need to be addressed. For the present I continue to regard Sphenocoelus and Dolichorhinus as synonyms.

Several specimens exhibiting the plesiomorphic morphology were referred to a new genus, Tanyorhinus , by Cook (1926). West and Dawson (1975), however, noted a similarity between Cook’s specimens of Tanyorhinus and specimens of Dolichorhinus , and stated that the two names could be regarded as synonyms. Mader (1998) accepted this synonymy and further synonymized Tanyorhinus and Dolichorhinus with Sphenocoelus (see above).

Specimens exhibiting the plesiomorphic morphology are all very similar to each other and any minor dissimilarities can be attributed to intraspecific variation and to differences in ontogeny, dental wear, and preservation. Because there is insufficient material to perform a statistical analysis, it cannot be determined whether or not there is more than one size group among these specimens, which might be accorded species status. I conclude, therefore, that all of this material should be referred to a single species, Sphenocoelus uintensis .

Specimens exhibiting the derived morphology, previously referred to Dolichorhinus ( Fig. 11 View FIGURE 11 ), are known in sufficient numbers to analyze quantitatively. Mihlbachler has placed all of these under a single species, Dolichorhinus hyognathus , but the analysis that follows, suggests that two species are actually present.

Most of these derived specimens were collected from the Uinta Formation (Wagonhound Member) of the Uinta Basin of Utah (indeed, most Sphenocoelus specimens in museum collections, including most of the type specimens, are from this formation and locality). Because the sample size was relatively small (fewer than twenty individuals for any given variable studied) the sample was supplemented by three specimens from the Washakie Basin of Wyoming (Washakie Formation, Adobe Town Member ) .

Table 7 presents summary statistics for all specimens of derived Sphenocoelus examined for this study, including specimens from both the Uinta and Washakie Basins. For most variables, the coefficient of variation falls within the range of 4 to 10 although, in many cases, the values of V are at the higher end of this range. The average value of V for the sample (excluding diastema length, which I have found to be extremely variable in all brontotheres) is also rather high (8.7), but this is due in part to the high values of V for canine size. Canine size is often sexually dimorphic in perissodactyls and sexually dimorphic characters frequently have high values of V associated with them. Even if canine dimensions are excluded from the calculation of the average, however, the average value of V for the sample remains high (7.9). The high average value of V suggests that there may be some heterogeneity in the overall derived Sphenocoelus sample.

A cluster analysis of all of the variables used to generate the statistics appearing in Table 7 (34 variables in all) does not indicate that more than a single group is present ( Fig. 12 View FIGURE 12 ). Interestingly, however, a cluster analysis using only the length of the skull, length of the cheek tooth series (exclusive of P1), and length of the molar series, results in two major size groups ( Fig. 13 View FIGURE 13 ). These were the same variables that delineated two size groups among specimens of Mesatirhinus (see above).

Tables 8 and 9 present the t -test results for the two Sphenocoelus size groups, which demonstrates that, for most variables, the means of the two size groups are significantly different. Out of thirty-four variables, well over three quarters (82%) have probabilities of.05 or less and over half (65%) have probabilities of.01 or less. Of all the cheek tooth and cheek tooth series measurements, only one (width of left P4) has a probability greater than.05 and this is so close (.061) that it can probably be discounted. Even though there is an 83% chance (see Methods section, above) that at least one significant result in this analysis will be invalid (i.e., a false rejection of the null hypothesis), it seems highly unlikely that well over three quarters of the results would be attributable to Type I errors.

a Based, whenever possible, on an average of left and right measurements.

b Excluding Diastema Length.

The only variables that are not significantly different in the two size groups are canine size and diastema length. As indicated above, both canine size and diastema length are highly variable in brontotheres. Furthermore, because canine size is probably sexually dimorphic, it may have a bi-modal distribution. It would not be surprising, therefore, if t -test results for these measurements could not establish significant differences between the groups, since t -tests are sensitive to outlying data points and non-normal distributions.

Thus, two size categories appear to exist in the data and it must be determined whether they represent two distinct taxa or are two groups within a single species (such as males and females). Tables 10 and 11 present the summary statistics for each of the two size groups suggested by the cluster analysis in Figure 13 View FIGURE 13 , and it will be noted from these tables that the average value of V for each group is more typical of extant mammalian species than the average value when the two groups were combined ( Table 7). Once again, the individual values of V for the canine measurements are rather high and this has some influence on the average value of V for both groups. If canine dimensions are eliminated from the calculation of the average, however, the average value of V in both groups remains quite acceptable. In the smaller-size group ( Table 10) the average value of V would become 5.8 and in the larger-size group ( Table 11) it would become 4.8.

Although the average value of V for both groups is within the acceptable range of extant mammalian species, a relatively large fraction (one fifth to one seventh) of the individual values of V (rounded to the nearest whole number) fall below 4. These low values are of consequence because the sample sizes are probably large enough to show most of the variation that is actually present. Thus, while the range and average value of V in the combined sample suggests that more than a single taxon may be present, analysis of V in the two groups within that sample is not conclusive. The average value of V for each group suggests that each might represent a distinct species, but some of the individual values of V tend to support the possibility that the groups represent size categories within a single species.

a Based, whenever possible, on an average of left and right measurements.

b Separate t -test.

The high values of V for the canine dimensions in both size groups are particularly informative because they suggest that canine size may be dimorphic in each. Although there are relatively few individuals within each group that had measurable canines (a problem with many brontothere specimens), cluster analyses ( Fig. 14 View FIGURE 14 ) support the hypothesis that canine size is bimodal in each of the groups. In both groups the canines form two distinct clusters strongly suggestive of males and females.

In my opinion, the weight of the evidence suggests that there are two species present. This conclusion is suggested by the high values of V for the combined derived Sphenocoelus sample and the more typical average value of V that results when the two groups within that sample are analyzed separately. Although some of the individual values of V in the two size groups are rather low, most are within the acceptable range and cluster analysis of canine size demonstrates that there are probably males and females within each group.

The large-size group of derived Sphenocoelus includes the holotype of the type species of Dolichorhinus , D. cornutum . According to Osborn (1929) Dolichorhinus cornutum is a junior synonym of Palaeosyops hyognathus Osborn. Previously I had not decided whether Osborn's conclusion was correct ( Mader 1989) but now agree that both species are synonymous. For a discussion of the reasons for my acceptance of this synonymy, see the Discussion section below for the species S. hyognathus .

The types of several species fall into the smaller-size group and, of these, Sphenocoelus intermedius (Osborn) has nomenclatural priority. Sphenocoelus intermedius and its junior synonyms are discussed in detail in the Discussion section for that species below.

In 1929, Osborn provisionally referred the lectotype lower jaw (AMNH 5098) of Palaeosyops vallidens Cope (1872) to the genus “ Dolichorhinus ” and, if he was correct, then the species name Sphenocoelus vallidens might have nomenclatural priority over one of the two species of derived Sphenocoelus recognized here. In 1989 Mader provisionally referred the same specimen to Telmatherium , however, while noting that the stratigraphic occurrence suggests that it might belong to “ Dolichorhinus ” (= Sphenocoelus in the present paper).

According to Osborn (1929) the type of Palaeosyops vallidens is from “Washakie B” (Uintan-aged deposits in the Washakie Basin), which would suggest that it is a specimen of Sphenocoelus . The locality recorded by Cope, however, may allow for a Bridgerian age, which is inconsistent with assignment to Sphenocoelus , but is consistent with Telmatherium as suggested by Mader. The specimen is described by Cope as being from Mammoth Buttes (= Haystack Mountain, according to Osborn 1929, p. 85) near the headwaters of Bitter Creek. This seems to place the locality on the north side of Haystack Mountain where both Uintan and Bridgerian (“Washakie A”) deposits are found (see Osborn 1929, Figs. 60 and 61). Interestingly, a partial skull referred to the Bridgerian genus Palaeosyops (AMNH 5105) was also described by Cope (1873) as being from Bitter Creek, but was provisionally identified as being from Washakie B (Uintan) by Osborn (1929, p. 163). It seems plausible, therefore, that the type of Palaeosyops vallidens is also from Bridgerian deposits. I continue to regard the specimen as being referable to Telmatherium , although, perhaps, the name Palaeosyops vallidens should be regarded as a nomen dubium.

Despite the fact that two derived species of Sphenocoelus are recognized in the present paper, there does not appear to be any consistent morphological difference between them (at least with regard to cranial anatomy). The suborbital protuberance may lack an anterior flange in some specimens of the small species, S. intermedius (e.g., UCMP 31845) but other specimens of this same species have a slight indication of this flange (UCMP 31846). It is quite possible that a study of the lower jaw or postcranial skeleton will reveal some consistent anatomical differences, but until then, size remains the only ready means of identification (see Fig. 15 View FIGURE 15 ). Skull length, length of the upper cheek tooth series (with or without first upper premolar), and length of the upper molar series are the most effective parameters to use in identification because there is a relatively large size interval between the two species. Length of the premolar series and individual cheek tooth measurements are less useful for diagnostic purposes because there is at least some size overlap between the taxa (indeed, this statistical “noise” may explain why the cluster dendrogram appearing in Figure 12 View FIGURE 12 did not clearly delineate more than one size group). If these less effective measurements must be used, identification will only be possible if the specimen is a small member of the smaller species ( S. intermedius ) or a large member of the larger species ( S. hyognathus ). Canine size and diastema length probably should be avoided in making a diagnosis because these dimensions are highly variable and overlap considerably.

a Based, whenever possible, on an average of left and right measurements.

b Excluding Diastema Length.

a Based, whenever possible, on an average of left and right measurements.

b Excluding Diastema Length.

Although the stratigraphic data is limited for the specimens examined in this study, it appears that Sphenocoelus intermedius and Sphenocoelus hyognathus were contemporaneous. Both occur in the parts of the Wagonhound Member of the Uinta Formation designated as Uinta B1 and Uinta B2 by Osborn (1929). It is possible that further investigation may reveal that the two species occupy slightly different geologic or faunal settings within the Uinta and Washakie Formations, but this will probably require the discovery of new specimens with highly detailed field data.

Since both Sphenocoelus intermedius and S. hyognathus share the derived (widened) morphology of the cranial vertex (while S. uintensis does not), S. intermedius and S. hyognathus can be regarded as immediate sister species. Since S. intermedius and S. hyognathus were the species originally incorporated under the name “ Dolichorhinus ,” it is possible to regard Dolichorhinus as a valid subgenus of Sphenocoelus .