Palaeosyops, Leidy 1870

Mader, Bryn J., 2010, A species-level revision of the North American brontotheres Eotitanops and Palaeosyops (Mammalia, Perissodactyla), Zootaxa 2339, pp. 1-43 : 16-36

Genus PALAEOSYOPS Leidy 1870

= Limnohyus Marsh 1872

= Limnohyops Marsh 1890 = Eometarhinus Osborn 1919

Age: Bridgerian.

Subage: Gardnerbuttean, Blacksforkian, and Twinbuttean.

Type species: P. paludosus Leidy 1870 .

Included species: P. fontinalis ( Cope 1873a) ; P. robustus ( Marsh 1872) .

Diagnosis: Medium-sized (length P2 to M3 approximately 123 to 165 mm) brontothere with six upper and lower incisors; large canines; very small or no upper diastema and a moderate lower diastema (mostly between p1 and p2); unmolarized premolars; large paraconules on the molars; hypocone or pseudohypocone variably present on M3. Palaeosyops is distinguished from all other brontotheres by the following synapomorphies: strongly brachycephalic skull; robust zygomatic arches that are sharply curved; sharply curved nasals that taper distally; and a low convexity or dome in the region of the frontoparietal border (see Figure 6 ).

Discussion: Leidy based the type species of Palaeosyops , P. paludosus , on four isolated teeth collected from Church Buttes in the Green River Basin of Wyoming ( Leidy 1870). Of these cotypes, Osborn (1929) selected USNM 759, a lower second molar, as the lectotype for the type species.

Based on the length and width of the lectotype molar, Mader (1989) concluded that the type belongs to the same taxon as the medium-sized, brachycephalic brontotheres later named Limnohyus and Limnohyops by Marsh (see Figure 6 ). Because the lectotype appeared to be identifiable, Mader accepted the name Palaeosyops as valid, but noted that if the lectotype should prove to be inadequate for diagnostic purposes, the next available name would be Limnohyus Marsh (1872; type species L. robustus ), which is based on a relatively complete skull (YPM 11122). The lectotype of Palaeosyops paludosus is from the Blacks Fork Member of the Bridger Formation, and because only one brontothere genus has been reliably recorded from this level, it is almost certain that Palaeosyops is a valid taxon.

Within two years after Leidy's naming of Palaeosyops paludosus, Marsh recognized that some of the specimens that had been referred to Palaeosyops had a hypocone on the M3 and some did not. Marsh (1872) proposed that the name Palaeosyops be restricted to those animals with a hypocone and that the name Limnohyus be given to those without it. Leidy (1872b), however, pointed out that the absence of a hypocone was a character originally attributed to Palaeosyops (the type series of Palaeosyops paludosus included an M3 that lacked a hypocone) and could not be used to define a new genus. Marsh later (1890) reversed his previous position and applied the name Palaeosyops to specimens without the hypocone on M3 and gave the new generic name Limnohyops to those with one.

Earle (1891; 1892) recognized both Limnohyops and Palaeosyops as valid genera as did Osborn (1908; 1929). Osborn (1929) diagnosed the genus Limnohyops as follows:

Brachycephalic; grinding teeth persistently brachydont; conules on the molars persistent, usually lophoid; third superior molar subquadrate and usually with distinct hypocone. Proportions of skull and skeleton moderately robust. Manus slender. Five sacral vertebrae (type).

Osborn's (1929) formal diagnosis for the genus Palaeosyops , which he incorrectly characterized as specific rather than generic characters (p. 312), was very similar to the above:

Brachycephalic. Grinding teeth persistently brachydont; metaconules on the molars persistent or absent; third superior molar without hypocone. Skull and skeleton robust; feet broad; manus with well-developed fifth digit; lunar resting subequally on magnum and unciform. Four sacral vertebrae.

Osborn (1929, p. 302) stated that the ancestral members of these two genera from Bridger B (Blacks Fork Member of the Bridger Formation) are difficult to distinguish from one another and elsewhere (p. 303) commented that the early species of both genera are so similar that they might be included within a single genus. According to Osborn, however, distinctions gradually began to develop between the two genera until, in the upper strata of the Bridger Formation, the two forms are quite distinct. Osborn listed ten characters (pp. 302–303) that supposedly distinguish species of Limnohyops and Palaeosyops from higher geologic levels (such as P. robustus and L. laticeps from Bridger D). These characters (not quoted verbatim) are:

1. Upper and lower molars of Palaeosyops relatively larger, more rounded, and more robust.

2. Vertical striations on the cones of the upper and especially lower molars more distinct in Palaeosyops .

3. Conules on M1–2 of Palaeosyops more variable, more rounded, and separate; in Limnohyops more constant, lophoid, ridged, or conjoined with the protocone and hypocone. Osborn noted that this distinction is not always reliable.

4. In Limnohyops hypocone of M3 typically present; in Palaeosyops , typically absent. In Limnohyops metaconule of M3 extremely reduced (owing to the large size of the adjacent hypocone); in Palaeosyops metaconule usually present and sometimes in such close proximity to the raised posterior cingulum that it appears similar to a hypocone (Osborn called such a structure a "pseudohypocone"). Thus the M3 of Limnohyops is more quadrate and that of Palaeosyops is generally more triangular.

5. In Palaeosyops parastyle rounded and obliquely placed across the outer angle of the crown; in Limnohyops parastyle sharp and extending outward as a ridge.

6. In Palaeosyops nasals taper toward the extremities and are more pointed, while in Limnohyops the sides of the nasals are more parallel and they are more truncate at the extremities.

7. In Palaeosyops the suborbital bar is rounded and the maxillary process extends across its lower side as a broad splint. In Limnohyops the bar is more depressed and slightly rectangular in section and the maxillary process extends across its lower side as a thin splint.

8. In Palaeosyops there is a strong median convexity near the frontoparietal junction, while in Limnohyops the top of the cranium is slightly concave.

9. The sagittal crest of Palaeosyops is lower, broader, and passes more rapidly into the temporal ridges whereas in Limnohyops the crest is higher and thinner and extends well forward before spreading into the temporal ridges.

10. Male jaws of Palaeosyops have a more prominent, massive chin and shorter insertion for the digastric than male skulls of Limnohyops . The rami of Palaeosyops (seen from below) are more massive than those of Limnohyops and the anterior border of the coronoid process in Palaeosyops is broader. In more progressive stages of Palaeosyops the anterior border of the coronoid process is hollowed out in front while in progressive stages of Limnohyops the process is narrower, less deeply excavated, and "lies more to the outer side of the line of the molar teeth".

Despite Osborn's recognition of Palaeosyops and Limnohyops as distinct genera, he noted (1929, p. 303) that the divergence between Palaeosyops and Limnohyops is far less than that observed within the modern genus Cervus . Mader (1989) concluded that Palaeosyops and Limnohyops are synonymous and stated that most of the generic distinctions cited by Osborn may be attributed to the vagaries of preservation or to individual variation. Gunnell and Yarborough (2000) also regarded Palaeosyops and Limnohyops as synonyms.

The following can be attributed to individual variation and probably sexual dimorphism in the above characters: the size and massiveness of the molars, the distinctiveness of the striations on the molars, the shape of the sagittal and temporal crests (see Mader 1989), and the shape and massiveness of the lower jaw (characters 1, 2, 9, and 10). The shape of the parastyle (character 5) is also probably attributable to individual variation and possibly sexual dimorphism, but may also reflect differences in wear. The presence or absence of a hypocone on M3 (character 4) is attributable to individual variation, but is not a sexually dimorphic character.

Osborn was incorrect in his assertion that skulls of " Limnohyops " lack the frontoparietal convexity or dome (character 8). Although he illustrated a skull ( Osborn 1929; Figs. 256, 259) that clearly shows a concave rather than convex forehead, the specimen in question (AMNH 11687) is almost crushed flat and this cranial morphology cannot be inferred. As previously noted ( Mader 1989), the frontoparietal area is not well preserved in any of the specimens that Osborn referred to Limnohyops , with most being badly damaged or missing entirely. The dome appears to be present in all specimens of Palaeosyops , but is more prominent in the supposed males.

The most convincing character used by Osborn to distinguish Limnohyops from Palaeosyops is the shape of the suborbital bar and the configuration of the suture pattern between the jugal and maxilla (character 7). Figure 7 illustrates two specimens showing these suture patterns: AMNH 5104, which Osborn referred to Limnohyops (type of L. laevidens ); and AMNH 1516, which Osborn referred to Palaeosyops . In AMNH 1516 a wide flange of the maxilla extends below the jugal and forms the ventral surface of the zygomatic arch beneath the orbit. In AMNH 5104 only a slender projection of the maxilla extends onto the zygomatic arch, and the suture between the jugal and maxilla has a distinct Z-shaped pattern. Furthermore, the suborbital part of the zygomatic arch is broadly rounded in AMNH 1516 and is somewhat angular in AMNH 5104. Although seemingly excellent distinguishing characteristics, these most likely reflect preservational rather than taxonomic differences. In both specimens the morphology of the jugal beneath the orbit is the same and the uppermost contact between the jugal and maxilla in AMNH 5104 is in the same position as the long oblique contact between the jugal and maxilla in AMNH 1516.

I posit that the morphology of both specimens was originally the same but, in AMNH 5104 the ventral part of the wide flange of the maxilla beneath the jugal has been broken away revealing the bottom part of the Z-shaped pattern normally concealed beneath. The ventral part of the maxillary flange is responsible for the roundness of the suborbital bar cited by Osborn as a character for Palaeosyops . When this flange is broken away, however, it results in the angular morphology that Osborn attributed to Limnohyops . Figure 8 illustrates AMNH 5104 showing my suggested reconstruction of the jugal-maxilla contact in this specimen. At least one other specimen that Osborn referred to Limnohyops (AMNH 11687, type of L. priscus ) appears to have the jugal-maxillary region broken in a manner similar to that of AMNH 5104.

Osborn recognized five valid species of Limnohyops and eight of Palaeosyops . These species were generally delineated by size, morphology of the premolars (shape of P2, presence of a mesostyle on some of the premolars, and presence of two lingual cusps on P2), and by the shape of the skull. The differences in premolar morphology cited by Osborn are trivial, however, and many of the differences in cranial morphology are the result of taphonomic deformation.

In 1919 Osborn described a fragmentary skull (AMNH 17412, Figure 9 ) from the Huerfano Basin and identified it as a form ancestral to Metarhinus , which he named Eometarhinus ("Dawn Metarhinus "). In 1929 Osborn upheld this identification, but Robinson (1966), Wallace (1980), Mader (1989; 1998), and Gunnell and Yarborough (2000) synonymized Eometarhinus with Palaeosyops .

Because all of the Huerfano material referred to Palaeosyops is rather poorly preserved it is difficult to demonstrate with certainty that it represents the same genus as Palaeosyops from the Bridger Formation. Although the Huerfano material is here considered to be correctly referred to Palaeosyops , I have previously questioned whether they are truly the same ( Mader 1989). The upper part of the Huerfano Formation, therefore, preserves the earliest known record of Palaeosyops in North America. Unfortunately, only a few specimens have been collected from this stratigraphic level and I have data for only two individuals: AMNH 17411 and AMNH 17425.

The Blacks Fork Member of the Bridger Formation is stratigraphically higher than the Huerfano Formation and is divisible into two stratigraphic levels: Bridger A and Bridger B. The sample of Palaeosyops from Bridger A that was examined for the present study includes some well-preserved cranial material, but is rather small in size and I have collected data for only three individuals from this level: AMNH 5107 (the type of P. fontinalis ), UW 3039, and UW 3091. Most of the Palaeosyops material known from the Blacks Fork Member of the Bridger Formation is from Bridger B. I have data for twenty five individuals from this higher level.

The Twin Buttes Member of the Bridger Formation overlies the Blacks Fork Member and is similarly divisible into two stratigraphic levels: Bridger C and Bridger D. Altogether data for 16 individuals from the Twin Buttes Member were collected for this study: 9 from Bridger C, 4 from Bridger D, and 3 from uncertain stratigraphic levels within the unit.

The sample sizes of Palaeosyops from the Huerfano Formation (assuming the generic assignment is correct) and Bridger A assembled for the present study are too small to make a statistical analysis meaningful. Thus it is not practical to compare specimens from the Huerfano Formation against those from Bridger A using t -tests, nor is it profitable to examine summary statistics for specimens of Palaeosyops from each level. The most complete specimen of Palaeosyops from Bridger A examined for this study (UW 3091, Figure 10 ) appears to be rather different compared to specimens from Bridger B and may be taxonomically distinct. Compared to specimens of Palaeosyops from Bridger B, the lingual cusp on the second upper premolar of UW 3091 is very poorly developed and the metacone is almost lacking (although not all specimens of Palaeosyops from Bridger A have a P2 that is as structurally plesiomorphic). Furthermore, the distinct cranial convexity that I have regarded ( Mader 1989; 1998; present paper) as a synapomorphy of Palaeosyops appears to be very small and may be absent entirely (it is difficult to be certain because of crushing). There is some evidence of a slight swelling in the fronto-parietal region, however (see Figure 10 ). Gunnell and Yarborough (2000) have also described a specimen of Palaeosyops from the lower Bridger (UM 94880) with a very small dome. The small dome (compared to specimens of Palaeosyops from higher levels) could be a species characteristic, but it is also possible that it is a characteristic of females ( Mader 1989).

Other cranial and dental characters of UW 3091 closely match those of specimens from Bridger B including a strongly brachycephalic skull and robust, sharply downturned zygomatic arches. The nasals are missing from the specimen, however, and it is not known whether they are strongly arched and distally tapered as in specimens of Palaeosyops from Bridger B. Gunnell and Yarborough (2000) noted that the nasals of UM 94880 are curved ventrally, but do not appear to taper distally. If correct, then distally tapered nasals could not be a synapomorphic character of Palaeosyops (as formulated here), but would be a shared derived character of the more derived Palaeosyops species ( P. paludosus and P. robustus ). Gunnell and Yarborough’s specimen is crushed, however, (which could flatten and, thus, widen the tips of the nasals) and the type of “ Eometarhinus ” huerfanensis (here referred to Palaeosyops ) has nasals that taper distally (see Figure 9 ).

Because of the small size of specimens of Palaeosyops from Bridger A and the retention of plesiomorphic conditions in at least one specimen of Palaeosyops from this level (UW 3091), I provisionally accept the specimens from Bridger A as representing a species that is distinct from specimens from Bridger B. The type of P. fontinalis is from Bridger A, and I tentatively refer all of the Palaeosyops material from this level to that species. I also tentatively refer specimens of Palaeosyops from the Huerfano Formation to P. fontinalis because of their relatively small size and because there is no basis at present for distinguishing them from specimens of Palaeosyops from Bridger A. Many more fairly complete specimens of Palaeosyops from Bridger A and from the Huerfano Formation will have to be collected and analyzed before either of these conclusions can be accepted with reasonable certainty. If it should later prove that the material from the Huerfano Formation represents a distinct species, then the name P. huerfanensis (Osborn) could probably be applied to it.

Table 7 presents the summary statistics for the sample of Palaeosyops from Bridger B. Almost 90 % of the individual coefficients of variation fall within the ideal range of 4 to 10. Three variables (excluding diastema length) have values of V greater than 10 (rounded to the nearest whole number) and one has a value of V less than 4. Although the average coefficient of variation for the sample is slightly high (6.7) it is within the range that I accept for a single species. Two out of the three variables that have values of V greater than 10 are measurements of canine size. Canine size is often sexually dimorphic in perissodactyls and it is possible that this factor accounts for the high individual values of V and perhaps for the relatively high average value of V. Although I have previously stated that canine size is not sexually dimorphic in brontotheres ( Mader 1989) I have since concluded that canine size is sexually dimorphic in at least some, and possibly all, brontotheres ( Mader 1998). Gunnell and Yarborough offer evidence suggesting that canine size may be dimorphic in Palaeosyops (see Gunnell & Yarborough 2000, Fig. 12 ). If canine size is excluded from the calculation of the average value of V for the sample, then the average value becomes 6.4, which is within the ideal range suggested by Simpson et al. (1960).

a Based, whenever possible, on an average of left and right measurements. b Excluding Diastema Length.

Analysis of the coefficient of variation strongly suggests that the sample is homogeneous and represents only a single species. It is surprising, therefore, that cluster analysis has revealed the presence of two or more size groups among the specimens in the sample. A cluster analysis of all of the variables listed in Table 7 results in a dendrogram ( Figure 11 ) in which all of the specimens are distributed in an essentially homogeneous manner. If, however, cluster analyses are performed on certain groups of variables, then size groups of specimens begin to emerge.

If, for example, a cluster analysis is performed using only the length of the cheek tooth series, length of the cheek tooth series exclusive of P1, length of the molar series, and the basilar length of the skull, then a dendrogram results in which three groups are delineated ( Figure 12 ). Two of these groups (labeled Group 1 and Group 2 on the diagram) join at a distance of 2.5 millimeters and are then joined by a third group (Group 3) at 4.0 millimeters. One specimen (AMNH 5102) groups out separately joining the others at a distance of 7.25 millimeters.

If a third cluster analysis is performed using only the length and width of the first upper molar, then only two distinct size groups emerge ( Figure 13 ). The size difference between these two groups is extremely small, however, and both join on the dendrogram at a distance of only one millimeter. It is difficult to determine from these cluster analyses how many size groups are actually present in the sample. Because the cluster analysis of the first upper molar resulted in the sharpest delineation of size groups, the groups suggested by this analysis were analyzed further.

T -tests (Tables 8 and 9) confirm that for many variables (length of cheek tooth series, length of cheek tooth series exclusive of P1, length of molar series, width of M3, length and width of M2 and M1, and width right P4) the means of the two size groups suggested by the dendrogram in Figure 13 are significantly different. Because the length of the premolar series and all but one of the individual premolar measurements show no significant difference between the two size groups, it is probable that molar dimensions account for the significant difference observed for the length of the cheek tooth series (with or without P1).

a Based, whenever possible, on an average of left and right measurements. b There is no variance in at least one of the two groups being compared.

Taking all t -tests together, there is an 80% chance (see Methods section) that at least one of these significant results is invalid (i.e., a false rejection of the null hypothesis). If, however, the error analysis is restricted to only those measurements that involve the molars (length of cheek tooth series, length of cheek tooth series exclusive of P1, length of molar series, and length and width of individual molars) then there is only a 54% chance that at least one of the significant results is due to a Type I error.

a Based, whenever possible, on an average of left and right measurements. b Insufficient data for t -test. C Separate t -test

The summary statistics for the two size groups suggested by the dendrogram in Figure 13 are presented in Tables 10 and 11. In Table 10, 75 % of the individual values of V (rounded to the nearest whole number) for the larger-size group are within the range expected for a single species. The average value of V for the sample (exclusive of diastema length and variables where n=1) is also within the recognized parameters of a single species (= 4.9) although, as indicated above, canine size might be sexually dimorphic and thus influence this result. If canine size is excluded from the calculation of the average, however, the average value of V remains within the suggested parameters of a single species (= 4.6).

TABLE 10. Summary statistics for the large-size group of Palaeosyops from Bridger B (all measurements in

millimeters).

a Based, whenever possible, on an average of left and right measurements. b Excluding Diastema Length and variables

where n=1.

The summary statistics for the smaller-size group ( Table 11 ), however, tend to indicate that there is less variation in this group than is normally encountered in a single extant mammalian species. Although over half of the individual values of V are within the range of a single species, a rather large fraction (over one quarter) are below 4. Most of the values of V that are below 4 were calculated from samples of reasonably large size (n> 5), suggesting that the values of V reflect most of the variation that is actually present. The average value of V for the smaller-size group (= 6.4) is well within the established range of a single species. Once again canine size may be influencing this result and it will be noted that the values of V for canine size within the smallersize group are relatively high. If canine size is excluded from the calculation of the average, however, the average value of V for the sample remains within the range of a single species (= 5.6).

a Based, whenever possible, on an average of left and right measurements. b Excluding Diastema Length and variables where n=1.

The preceding statistical analyses suggest that there are at least two size groups present within the Bridger B sample. It cannot be determined from these analyses, however, whether these groups represent different species or size groups within a single species (perhaps males and females). Inspection of the specimens in the sample failed to show which of these two possibilities is correct. Although dental material is well preserved in the sample, skulls are in generally poor condition. In most cases it was not possible to determine the sex of the specimens based on the morphology of the temporal crests (see Mader 1989). I also found no morphological characters (cranial or dental) that could be used to define the size groups suggested by the cluster dendrogram in Figure 13 as separate species.

The molar dimensions between the two size groups seem to show a significant difference, which could be an adaptive distinction implying that they are different taxa. However, the size difference between the two groups is actually quite minor (both groups join on the dendrogram at a distance of only one millimeter) and although all t -tests that involve molar measurements (see above) show a significant size difference, there is a 54% chance that at least one of these fifteen positive t -tests is invalid due to a Type I error. Furthermore, if all t -tests performed for this analysis are considered (thirty-one tests), then there is an 80% chance that at least one positive result is in error.

Although it is possible that more than one species may be present in the Bridger B sample, the preceding morphological and statistical analysis did not clearly establish this to be the case. While Osborn (1929) identified several species of Palaeosyops (assigned both to Palaeosyops and Limnohyops ) from Bridger B, none of the characters that he used are adequate for diagnostic purposes. In the absence of convincing statistical and morphological evidence for more than a single species I have chosen to take a conservative approach and recognize only one. The type species of Palaeosyops , P. paludosus , is from Bridger B and I tentatively assign all specimens from this level to this taxon. Gunnell and Yarborough (2000) recognized a second valid species from this level ( P. laevidens ), which I regard as a junior synonym of P. paludosus (see Discussion section for the species P. paludosus ). If later research should show that more than a single species is actually present in Bridger B then there may be some difficulty in establishing which specimens should be referred to P. paludosus . The lectotype of P. paludosus is an isolated m2 and may not be adequate for diagnostic purposes. The name P. paludosus might have to be regarded as a nomen dubium, therefore.

n Group 1 n Group 2 F Probability T -Test a Based, whenever possible, on an average of left and right measurements. b Insufficient data for t -test. c Separate t -test.

The Twin Buttes Member of the Bridger Formation is divisible into two stratigraphic levels: Bridger C and Bridger D. Although it would have been desirable to perform t -test comparisons to determine whether the samples from Bridger C and D are significantly different in size, the sample from Bridger D is rather small (four individuals), making this impractical. T -tests ( Tables 12 and 13) suggest, however, that there is a significant difference in molar size between specimens of Palaeosyops from Bridger C and from Bridger B (Blacks Fork Member of the Bridger Formation). Eighty percent of measurements that involve the molars (length of cheek tooth series, length of cheek tooth series exclusive of P1, length of molar series, and length and width of individual molars) test as significantly different (only the length of the right M2 and width of both M2’s do not). Among those t -tests that concern measurements involving the molars there is a 54% chance that at least one of these significant results is invalid due to a Type I error. Among all thirty t -tests performed for this analysis there is a 79% chance that at least one of the significant results is an invalid rejection of the null hypothesis.

a Based, whenever possible, on an average of left and right measurements. b Excluding Diastema Length and variables where n=1.

The summary statistics for the Bridger C sample are presented in Table 14 . In this table almost two-thirds of the individual values of V are within the accepted range for a single species. A number of premolar measurements, however, have values of V greater than 10, and the average value of V for the sample is also rather high (= 7.0) although within the accepted range of a single species.

Cluster analyses of all of the variables listed in Table 14 results in the dendrogram illustrated in Figure 14 . Most specimens on this dendrogram (ACM 1794, AMNH 12185, USNM 13454, USNM 13465, USNM 16660, and USNM 26138) are distributed in an essentially homogeneous fashion, clustering together at a distance of 1.2 millimeters. Three specimens, however, join this cluster at slightly further distances. USNM 16690 and AMNH 12189 join at 2.0 and 2.1 millimeters respectively, and AMNH 12190 joins at a distance of 3.5 millimeters.

Although the groupings in this cluster dendrogram are not completely homogeneous, there is no clear delineation into size groups. Cluster analysis of various combinations of the characters listed in Table 14 also failed to differentiate size groups. Although some of the individual values of V are rather high in the Bridger C sample, there is not enough evidence to justify the recognition of more than a single species from this stratigraphic level. Specimens of Palaeosyops from Bridger C are apparently distinguished from specimens of Palaeosyops from Bridger B by their larger size (especially the molars) although there is considerable size overlap. Based upon this difference, I provisionally recognize the species of Palaeosyops from Bridger C as being taxonomically distinct from P. paludosus in Bridger B. The earliest new name to be assigned to a specimen of Palaeosyops clearly recorded from Bridger C is Palaeosyops robustus ( Marsh 1872) , and I tentatively refer all specimens of Palaeosyops from this level to that species (see Discussion for P. robustus below).

The sample of Palaeosyops from Bridger D is too small to analyze separately. Table 15, however, presents the summary statistics for specimens of Palaeosyops from the entire Twin Buttes Member of the Bridger Formation including both Bridger C and D. As shown by this table, most of the individual values of V are within the range established for a single species, as is the average value of V for the sample. Only a small fraction of the individual values of V are less than 4 or greater than 10. This result suggests that the sample of Palaeosyops from the Twin Buttes Member of the Bridger Formation is essentially homogeneous and there is no indication that more than a single species is present. I provisionally refer all of the Palaeosyops material from Bridger D, therefore, to P. robustus .

a Based, whenever possible, on an average of left and right measurements. b Excluding Diastema Length and variables where n=1.

Figures

FIGURE 6. Skull of Palaeosyops in A, dorsal, and B, lateral views. After Osborn (1929).

FIGURE 7. Suture pattern between the jugal (dark gray) and maxilla (light gray) in two specimens of Palaeosyops. A, AMNH 5104 (reversed), referred by Osborn to Limnohyops; B, AMNH 1516 (reversed), referred by Osborn to Palaeosyops. Both figures modified from Osborn (1929).

FIGURE 8. Suture pattern between the jugal (dark gray) and maxilla (light gray) in AMNH 5104, a specimen of Palaeosyops paludosus. A, suture pattern as it actually appears on the specimen, probably due to breakage of the maxilla; B, reconstruction of the jugal-maxilla contact as it may have originally appeared. Both figures reversed and modified from Osborn (1929).

FIGURE 9. Holotype (AMNH 17412) of “ Eometarhinus ” huerfanensis (= Palaeosyops fontinalis). A, nasals in dorsal view; B, nasals and facial region of skull in lateral view; and C, palate in ventral view. After Osborn (1929).

FIGURE 10. Skull of Palaeosyops fontinalis (UW 3091) in lateral view.

FIGURE 11. Cluster dendrogram for specimens of Palaeosyops from Bridger B resulting from a cluster analysis of all variables listed in Table 7. a, type of Limnohyops monoconus; b, type of Limnohyops priscus; c, type of Limnohyus laevidens.

FIGURE 12. Cluster dendrogram for specimens of Palaeosyops from Bridger B resulting from a cluster analysis of basilar skull length, length of cheek tooth series, length of cheek tooth series exclusive of P 1, and length of molar series. a, type of Limnohyops priscus; b, type of Limnohyus laevidens; c, type of Limnohyops monoconus.

FIGURE 13. Cluster dendrogram for specimens of Palaeosyops from Bridger B resulting from a cluster analysis of the length and width of the first upper molars. a, type of Limnohyops priscus; b, type of Limnohyops monoconus.

FIGURE 14. Cluster dendrogram for specimens of Palaeosyops from Bridger C resulting from a cluster analysis of all variables listed in Table 14. a, type of P. grangeri.

Tables

TABLE 11. Summary statistics for the small-size group of Palaeosyops from Bridger B (all measurements in millimeters).

Basilar Length Skull a Length Cheek Tooth Series a Length P2 to M3a Length Premolar Series a Length Molar Series a Length Diastema a Length Left M3 Length Right M3 Width Left M3	n Range 1 406.0 – 406.0 3 148.0 – 155.0 4 132.3 – 144.8 5 51.0 – 67.5 8 86.0 – 98.5 3 4.6 – 8.2 6 30.5 – 35.5 8 31.0 – 34.0 6 32.8 – 36.2	M 406.0 151.0 139.4 62.0 89.6 6.5 32.8 32.5 34.4	s V ±0.0 0.0 ±3.6 2.4 ±5.5 3.9 ±6.6 10.6 ±4.0 4.5 ±1.8 28.3 ±1.8 5.5 ±1.2 3.8 ±1.2 3.6
Width Right M3 Length Left M2	6 33.0 – 36.1 5 31.0 – 33.5	34.3 32.1	±1.2 3.5 ±1.3 4.0
Length Right M2 Width Left M2 Width Right M2	7 30.5 – 33.0 6 30.7 – 33.6 5 30.0 – 33.6	31.4 32.5 32.2	±0.8 2.6 ±1.1 3.2 ±1.4 4.2
Length Left M1 Length Right M1 Width Left M1	4 25.0 – 26.0 8 23.5 – 25.0 4 25.3 – 25.9	25.5 24.4 25.6	±0.6 2.3 ±0.6 2.6 ±0.4 1.4
Width Right M1 Length Left P4 Length Right P4	7 25.3 – 27.6 5 16.9 – 19.0 6 16.5 – 19.0	26.2 17.9 17.4	±0.8 2.9 ±0.9 5.2 ±0.9 5.3
Width Left P4 Width Right P4 Length Left P3 Length Right P3	5 21.7 – 24.5 6 22.5 – 24.4 3 15.0 – 17.5 6 13.0 – 17.0	23.3 23.2 16.2 15.2	±1.3 5.5 ±0.7 2.8 ±1.3 7.8 ±1.3 8.7
Width Left P3 Width Right P3 Length Left P2	3 17.4 – 20.0 4 17.9 – 20.1 3 12.0 – 16.5	18.7 18.9 14.2	±1.3 7.1 ±1.0 5.5 ±2.3 15.9
Length Right P2 Width Left P2 Width Right P2	3 11.0 – 17.0 2 14.1 – 15.0 2 13.2 – 14.4	14.0 14.5 13.8	±3.0 21.4 ±0.6 4.4 ±0.9 6.2
Buccal-Lingual Width Left Canine Buccal-Lingual Width Right Canine Mesial-Distal Width Left Canine	2 17.5 – 22.9 0 ––––– 2 19.4 – 24.0	20.2 ––––– 21.7	±3.8 19.0 ±––––– ––––– ±3.4 15.0
Mesial-Distal Width Right Canine AVERAGE V	0 –––––	–––––	±––––– ––––– 6.4b

TABLE 12. F-Test comparisons for Palaeosyops from Bridger B (Group 1) and Bridger C (Group 2).

Basilar Length Skull a Length Cheek Tooth Series a Length P2 to M3a Length Premolar Series a Length Molar Series a Length Diastema a Length Left M3 Length Right M3	4 1 8 3 12 4 11 4 17 4 9 2 17 7 19 7	––––– 2.595 0.384 0.817 0.692 0.161 0.588 0.105	––––– 0.111 0.536 0.367 0.406 0.691 0.443 0.746	–––––b Pooled Pooled Pooled Pooled Pooled Pooled Pooled
Width Left M3 Width Right M3	15 4 17 7	0.128 0.206	0.721 0.650	Pooled Pooled
Length Left M2	15 7	1.490	0.222	Pooled
Length Right M2 Width Left M2	16 8 15 3	0.375 0.131	0.540 0.718	Pooled Pooled
Width Right M2 Length Left M1	13 5 12 4	0.252 0.002	0.616 0.961	Pooled Pooled
Length Right M1 Width Left M1	16 6 13 3	0.022 0.457	0.882 0.501	Pooled Pooled
Width Right M1 Length Left P4	12 5 14 4	0.535 0.017	0.465 0.896	Pooled Pooled
Length Right P4	14 7	4.347	0.037	Separate
Width Left P4 Width Right P4	15 4 14 6	2.368 7.659	0.125 0.006	Pooled Separate
Length Left P3 Length Right P3	12 3 15 6	––––– 2.979	––––– 0.084	–––––b Pooled
Width Left P3	10 4	1.714	0.192	Pooled
Width Right P3 Length Left P2	11 6 11 3	2.738 2.385	0.099 0.126	Pooled Pooled
Length Right P2 Width Left P2	13 4 9 3	0.465 0.867	0.496 0.354	Pooled Pooled
Width Right P2 Buccal-Lingual Width Left Canine	8 4 5 2	2.155 0.197	0.144 0.661	Pooled Pooled
Buccal-Lingual Width Right Canine Mesial-Distal Width Left Canine Mesial-Distal Width Right Canine	2 1 5 2 2 1	––––– 0.256 –––––	––––– 0.618 –––––	–––––b Pooled –––––b
a Based, whenever possible, on an average of being compared.	left and right measurements. b	There is	no variance in one	of the two groups

TABLE 14. Summary statistics for Palaeosyops from Bridger C (all measurements in millimeters).

	n Range	M	s V
Basilar Length Skull a Length Cheek Tooth Series a Length P2 to M3a Length Premolar Series a Length Molar Series a Length Diastema a Length Left M3	1 405.5 – 405.5 3 165.0 – 169.5 4 151.5 – 165.0 4 65.3 – 72.0 4 98.5 – 107.5 2 6.0 – 8.0 7 35.5 – 40.0	405.5 167.7 156.5 67.7 102.5 7.0 37.5	±0.0 0.0 ±2.4 1.4 ±6.3 4.0 ±3.1 4.6 ±3.8 3.7 ±1.4 20.5 ±1.5 4.1
Length Right M3 Width Left M3	7 34.0 – 42.5 4 37.3 – 43.1	38.1 40.3	±2.8 7.3 ±2.7 6.6
Width Right M3	7 37.1 – 42.5	39.4	±2.3 5.8
Length Left M2 Length Right M2	7 33.5 – 41.5 8 30.0 – 38.5	36.0 34.4	±2.8 7.7 ±2.6 7.6
Width Left M2 Width Right M2	3 35.4 – 40.3 5 34.9 – 40.1	37.2 36.1	±2.7 7.2 ±2.2 6.1
Length Left M1	4 27.5 – 31.0	29.1	±1.4 4.9
Length Right M1 Width Left M1	6 27.0 – 31.5 3 29.6 – 33.8	29.2 31.2	±1.8 6.2 ±2.3 7.4
Width Right M1 Length Left P4	5 29.1 – 33.3 4 18.0 – 20.0	31.0 18.9	±2.0 6.5 ±0.9 4.5
Length Right P4 Width Left P4	7 14.5 – 22.0 4 24.4 – 25.9	18.8 25.0	±2.4 12.8 ±0.6 2.5
Width Right P4 Length Left P3	6 19.8 – 30.7 3 17.0 – 17.0	25.4 17.0	±3.6 14.0 ±0.0 0.0
Length Right P3	6 13.5 – 18.5	16.8	±1.9 11.2
Width Left P3 Width Right P3	4 16.5 – 20.7 6 16.0 – 22.6	19.4 20.5	±2.0 10.3 ±2.4 11.7
Length Left P2 Length Right P2	3 12.5 – 17.5 4 14.5 – 17.0	14.7 15.4	±2.6 17.5 ±1.1 7.2
Width Left P2	3 13.0 – 16.2	14.8	±1.7 11.2
Width Right P2 Buccal-Lingual Width Left Canine	4 15.4 – 16.2 2 20.7 – 22.8	15.8 21.8	±0.4 2.6 ±1.5 6.8
Buccal-Lingual Width Right Canine Mesial-Distal Width Left Canine	1 20.4 – 20.4 2 20.0 – 21.7	20.4 20.8	±0.0 0.0 ±1.2 5.9
Mesial-Distal Width Right Canine AVERAGE V	1 21.2 – 21.2	21.2	±0.0 0.0 7.0b