Leptomerycidae. Chapter 27 (pp. 581-608) in Prothero, D. R., and Emry, R. J. (eds.) The Terrestrial Eocene-Oligocene Transition in North America. Cambridge Univ. Press. (1996).

Timothy H. Heaton
Department of Earth Sciences
University of South Dakota
Vermillion, SD 57069

Robert J. Emry
Department of Paleobiology
Smithsonian Institution
Washington, DC 20560

ABSTRACT

Leptomerycids appear in North America in the Late Duchesnean and extend into the Arikareean. The earliest grade, called Hendryomeryx by some workers, differs from later forms (Leptomeryx) in having lower crowned cheek teeth. Leptomeryx yoderi appeared in the early Chadronian and became a widespread species. This small but high crowned species probably gave rise to all later species of Leptomeryx. At Flagstaff Rim, Wyoming, L. yoderi underwent a gradual anagenetic size increase to become L. mammifer of the middle to late Chadronian. Leptomeryx speciosus probably also descended from L. yoderi. Leptomeryx yoderi and L. speciosus are of similar size and differ only in the shape of the entoconulid on the last molar. Both L. speciosus and L. mammifer became widespread species and are found at Pipestone Springs, Montana and in the Cypress Hills, Saskatchewan. Leptomeryx mammifer went extinct before the end of the Chadronian, leaving L. speciosus as the only known surviving species. At Douglas, Wyoming, L. speciosus can be seen undergoing gradual anagenesis across the Chadronian-Orellan boundary to become L. evansi, the type species of the genus. Leptomeryx evansi is probably the only valid Orellan species, and it is found in great numbers in the Big Badlands of South Dakota and at other Orellan exposures of the Great Plains.

In the past, the taxonomy of the leptomerycids has been confused by the enormous variation in the ridge configuration of the lower premolars, especially P3. Our statistical study of large samples demonstrates that these are merely individual variations within all species of Leptomeryx and therefore have little taxonomic value. Individual specimens cannot be identified based on premolar morphology, but, using statistical samples, the proportion of different character states can help in establishing ancestor/descendent relationships.

INTRODUCTION
Leptomeryx, one of the smaller and more common artiodactyls of the Chadronian and Orellan land mammal ages, comprises a cohesive group of about seven species, the more primitive of which have been placed in the genus Hendryomeryx by some workers (e.g. Black, 1978; Storer, 1981). The genotypic species, L. evansi, is common in the Big Badlands of South Dakota and at other Orellan localities in the Great Plains. Chadronian forms are more diverse and are found in deposits from Saskatchewan to Texas, but the most abundant material with the best stratigraphic control comes from eastern Wyoming.

Because Leptomeryx is a hornless artiodactyl, only the teeth have been used to distinguish species. Dentaries are the most common elements recovered, and many of these show signs of scavenging. Complete dentaries include four incisiform teeth (I1-3 and C1), a peg-like first premolar (P1) within the diastema, three anteroposteriorly elongate premolars (P2-4), and three selenodont molars (M1-3). Most commonly the anterior part with incisors is missing, and the preserved part contains P2-M3 (Figure 1b). Less commonly the P2 is also preserved (Figure 1d). The lower cheek teeth form the basis of most studies. Upper dentitions are less common and include three elongate premolars and three molars.

The sporadic distribution of fossil deposits makes environmental and evolutionary reconstruction difficult because geographic and chronologic variations cannot always be distinguished and because important transitions are not always recorded (see Heaton, this volume). This problem is exacerbated for the Chadronian-Orellan transition because the most fossiliferous Chadronian deposits are located in isolated basins of the Rocky Mountains while the most productive Orellan localities occur in the more climatically uniform Great Plains. Fortunately, Leptomeryx material from central and eastern Wyoming spans the late Duchesnean through early Orellan, and the species found there occur in deposits of similar age throughout the region.

During its existence Leptomeryx underwent several important morphological transformations that were interspersed with periods of stasis. The timing and nature of these transformations is the subject of this paper. The evolutionary history of Leptomeryx, as we believe it can best be reconstructed, is discussed in three parts. In chronological order they are: 1) the development of Leptomeryx from Hendryomeryx in the late Duchesnean or earliest Chadronian; 2) the apparent cladogenesis of L. yoderi into L. speciosus and L. mammifer in the early to middle Chadronian; and 3) the development of L. evansi from L. speciosus during the late Chadronian to early Orellan. The later history of leptomerycids is described by Taylor and Webb (1976).

Collections Used
AMNHAmerican Museum of Natural History, New York, New York
CMCarnegie Museum of Natural History, Pittsburgh, Pennsylvania
MCZMuseum of Comparative Zoology, Harvard University, Cambridge, Massachusetts
MPUMUniversity of Montana Museum of Paleontology, Missoula, Montana
NMCNational Museums of Canada, Ottawa, Ontario
PTMPioneer Trails Museum, Bowman, North Dakota
ROMRoyal Ontario Museum, Toronto, Ontario
SDSMSouth Dakota School of Mines and Technology, Rapid City, South Dakota
SMNHRoyal Saskatchewan Museum, Regina, Saskatchewan
USNMU.S. National Museum, Smithsonian Institution, Washington, D.C.

THE DEVELOPMENT OF LEPTOMERYX FROM HENDRYOMERYX
The ancestry of the leptomerycids is uncertain. Emry et al. (1987) noted their sudden appearance in North America and suggested the Eocene genus Archaeomeryx of Asia as a possible ancestor.

We have not studied the earliest specimens of Leptomeryx (e.g. Hendryomeryx) in detail, and they are less relevant to this volume than the later forms. All are small in size. Most of the material is incomplete and difficult to evaluate, especially when one considers the enormous amount of individual variation found in later species. We will therefore limit our coverage of the earliest specimens to a review of the literature, a report of new material, and some comments on taxonomy.

Wilson (1974) described a new species, Leptomeryx defordi, based on what was then the earliest known leptomerycid material, from the Porvenir local fauna of Trans-Pecos Texas. He described it as a primitive species, "upper molars with incomplete selenes," and "in lower molars the metaconid is a distinct cusp but joined to the posterior internal crest by a ridge at right angles to the tooth."

Black (1978) reported even earlier leptomerycid dentitions from Badwater Creek in central Wyoming and chose to give them, as well as the Porvenir material and two specimens from McCarty's Mountain, Montana, the name Hendryomeryx. According to Black "Hendryomeryx is less advanced than the Oligocene Leptomeryx of the plains and intermontane regions in having lower crowned, less selenodont molars, and a weak metaconid-entoconid lophid." Black gave his Badwater Creek material the specific name H. wilsoni.

Storer (1981), in his review of leptomerycids from the Cypress Hills, Saskatchewan, assigned Cope's (1889) small species Leptomeryx esulcatus to Hendryomeryx and gave the genus a fuller description. Storer described the molars of Hendryomeryx, in addition to being lower crowned, as "much shallower-patterned than in Leptomeryx, with the bottoms of the fossettes and fossettids easily visible; in Leptomeryx, valley walls are much steeper, and valley bottoms are not easily visible except in very worn specimens." Tooth length measurements for the three species assigned by Storer (1981) to Hendryomeryx are reported in Table 1. Hendryomeryx esulcatus occurs with a middle Chadronian fauna, and Storer indicated that its teeth were higher crowned than in H. wilsoni. The inclusion of L. esulcatus in Hendryomeryx would extend this genus into the middle Chadronian when more derived species were also present.

We consider Black's (1978) decision to separate Hendryomeryx from Leptomeryx to be unfortunate because the differences between the species assigned to this genus and the Chadronian species retained in Leptomeryx are hardly worthy of generic distinction. Leptomeryx evansi of the Orellan is more distinct from the Chadronian leptomerycids than any of the Chadronian and Duchesnean species are from one another, and even these differences are only of specific value. On the whole, the species of Leptomeryx exhibit remarkable uniformity, so there is no utility in dividing the genus.

Undescribed material from several Duchesnean and early Chadronian localities will need to be considered when unraveling early Leptomeryx evolution. Black (1978) reported that "in the McCarty's Mountain fauna in Montana an undescribed species of Hendryomeryx appears to be present (CM 1057 and 31397) together with a species of Leptomeryx," though Storer (1984) considered both of these to represent Leptomeryx. The West Canyon Creek fauna of late Duchesnean or early Chadronian age in central Wyoming (under study by Emry) includes several Leptomeryx specimens, possibly of L. (H.) wilsoni. Tooth lengths for the West Canyon Creek sample are listed in Table 2. A new sample has also been found in the Medicine Pole Hills of North Dakota (discussed below).

Because late Duchesnean and early Chadronian faunas are few in number and do not contain abundant leptomerycid material, tracing the origin and early development of Leptomeryx is a difficult task. All we know at present is that the transition to a later grade involved an increase in crown height and a deepening of valleys in the molars (Storer, 1981).

THE CLADOGENESIS OF LEPTOMERYX YODERI INTO L. SPECIOSUS AND L. MAMMIFER
The main thrust of our study is a detailed statistical analysis of Chadronian Leptomeryx dentaries from Flagstaff Rim and other localities in central Wyoming (see map on page ??). Emry (1970, 1973, 1992) supervised the collection of nearly all this material, keeping careful stratigraphic records, and Heaton (1989) measured and analyzed these specimens as a Smithsonian Institution postdoctoral research project. The sample size (1400 dentaries from Flagstaff Rim alone), the number of characters measured (89 on complete specimens), and the stratigraphic control allow us to examine individual vs. specific variation and evolutionary change at a level of detail that is rarely possible in paleontological studies.

As noted by Emry (1970, 1973), there are several species of Leptomeryx at Flagstaff Rim, and at least one lineage clearly represents an evolutionary transition from one species to another. Figures 2 through 4 are bivariate plots of Leptomeryx showing the size of specimens from different stratigraphic levels. The size is based on length of the cheek tooth series (P2-M3) in Figure 2, length of the molars (M1-3) in Figure 3, and length of M2 (the most abundantly represented tooth) in Figure 4. Figure 5 is a similar plot using M2 length for Leptomeryx from the Ledge Creek locality, 11 miles SSE of Flagstaff Rim. In each plot a unimodal size distribution is found at the lowest level while a bimodal distribution, which includes a larger species, is found throughout the upper part of the section. Levels in between exhibit an intermediate amount of size variation.

At Flagstaff Rim the single tooth measurement (Figure 4) provides the largest sample size, while the full cheek tooth row measurement (Figure 2) provides the cleanest separation between two coexisting species in the upper part of the section. The length of the molars (Figure 3) makes a good compromise and is used in a series of size histograms in Figure 6. Because the two species from the upper portion of the Flagstaff Rim section are in stasis (or become only slightly larger up section), they are grouped into a single histogram (Figure 6d). Specimens from the lower part of the section are concentrated in three quarries that exhibit different size distribution patterns, so they are shown (together with specimens from nearby stratigraphic levels) in three separate histograms (Figure 6a-c). We will refer to these samples as Quarry A ("Low Pocket"; oldest), Quarry B ("Dry Hole Pocket"; middle), and Quarry C ("B-44 Pocket"; youngest).

Of the three quarries, Quarry A has the largest sample size, shows the greatest morphologic uniformity, and comes closest to exhibiting a normal size distribution (Figure 6a). Emry (1973) referred this population to Leptomeryx yoderi because of its similarity in morphology and age to the type specimen from the Yoder local fauna, located 125 miles ESE of Flagstaff Rim. Leptomeryx yoderi was described by Schlaikjer (1935), and more material from the type locality was referred to this species by Kihm (1987). Statistical data on the Quarry A sample and on L. yoderi from the Yoder local fauna are presented in Table 3.

The upper portion of the Flagstaff Rim section exhibits a distinctly bimodal size distribution that is best interpreted as representing two species. Similar bimodal size distributions are found in samples of equivalent age from Ledge Creek, Wyoming (Figure 5); Pipestone Springs, Montana (Matthew, 1903; Tabrum and Fields, 1980); and Calf Creek in the Cypress Hills, Saskatchewan (Storer, 1981). Emry (1973) considered the smaller of the two species from Flagstaff Rim to be the same as the smaller species from Pipestone Springs, which Matthew (1903) had provisionally referred to Leptomeryx esulcatus. However, Storer (1981) instead assigned this material to L. speciosus, a species originally described from the Cypress Hills by Lambe (1908), and this name is used here (see Emry et al., 1987). Statistical data on this sample and on similar samples from Ledge Creek and Pipestone Springs are shown in Table 4.

Emry (1973) referred the larger of the two species from the upper part of the Flagstaff Rim section to Leptomeryx mammifer because of its similarity to the sample of similar size from Pipestone Springs, which Matthew (1903) referred to that species. The type specimen of L. mammifer was described by Cope (1885) from the Cypress Hills, but it is very fragmentary and may not be diagnostic of this species (Storer, 1981). We have not examined the type but will use the name L. mammifer until this taxonomic problem is resolved. Table 5 presents statistics for this larger species from Flagstaff Rim, Ledge Creek, and Pipestone Springs.

The samples from Quarries B and C (Figures 6b-c) are more difficult to interpret because of their skewed size distributions, but they seem to exhibit intermediacy between the Quarry A sample and the sample from the upper part of the Flagstaff Rim section. Emry (1973) suggested that Leptomeryx from Flagstaff Rim represents two coexisting lineages, both of which underwent anagenetic increases in size. In particular, he considered L. mammifer to be a descendant of L. yoderi. Even more significantly, this sequence of four chronologically distinct samples may represent a case of sympatric speciation, and this is the reason for our unusually detailed analysis.

Methods
In analyzing Leptomeryx from Flagstaff Rim, we first sought to measure the range of variation in each supposed species to consider the possibility that two or more species of similar size could be mixed. Specimens from all levels at Flagstaff Rim and samples from several other Chadronian and Orellan localities were examined to assess the total range of variation, and measurable characters were chosen to evaluate this variation quantitatively. Objective measurements chosen were the anteroposterior length and the anterior and posterior transverse width of each cheek tooth, the pre-protoconid length of the premolars, the length of the anterior selene of the molars, and the depth of the jaw below the tooth row in three places (Table 6, Figure 7a-b). Variations within the occlusal surfaces of the teeth were harder to quantify, so a series of subjective scores were developed to facilitate this (Figure 7c).

Six subjective measurements were made on P2, with corresponding measurements on P3. The first five refer to the configuration of one or more ridges that extend posteriorly from the protoconid (discussed below). The sixth measurement is the size of the cingulum on the posterolingual corner of the tooth. Eight additional measurements were made on P3. The first refers to a forking at the tip of the ridge extending anteriorly from the paraconid. The others refer to five accessory ridges and two accessory cusps found on some specimens. Twelve subjective measurements were made on P4: the anterior fork (as in P3), five accessory ridges, five accessory cusps, and an anterolingual cingulum (Table 6, Table 7c).

Three subjective measurements were made on M1 and M2: 1) development of the "Paleomeryx fold," which extends posteriorly off the protoconid to meet the anterior extension of the hypoconid; 2) presence of an accessory cusp (referred to as a "buccal column" by Storer, 1981) in the valley between the protoconid and hypoconid on the labial margin of the tooth; and 3) presence of an accessory cusp or transverse ridge at the posterior end of the tooth. M3 has a posterolophid behind the two selenes. Six subjective measurements were made on this tooth, two of which are development of the "Paleomeryx fold" and presence of the labial accessory cusp as in M1 and M2. The third is the presence of an additional labial accessory cusp that is sometimes found in the valley between the hypoconid and the hypoconulid (Figure 7c). The final three subjective measurements relate to the shape of the entoconulid and the valley posterior to it on the lingual side of the posterolophid (discussed below).

Dentaries from Flagstaff Rim (Chadronian) and Douglas (Orellan) were studied extensively prior to the selection of measurements in order to include all possible variations. Very few variants were found that could not be measured with this set of objective and subjective measurements. All measurements of accessory cusps, ridges, folds, cingula, and forks were scored with an integer from zero (feature absent) to five (feature exceptionally large or well-developed), though some did not exhibit this entire range of variation (Table 6). The only subjective measurements with a different scoring system are the position of the primary ridge behind the protoconid on P2 and P3, the last three measurements on M3, and the wear stage.

All our examinations of Leptomeryx teeth were conducted using a binocular microscope. Objective measurements were made using a single pair of fine-tip metal calipers. The bulk of material was measured at the Smithsonian's National Museum of Natural History, though some specimens were measured by Heaton at the University of South Dakota or on visits to the American Museum of National History in New York City.

Premolar Variations
The greatest morphological variation is exhibited in P3. One, two, or three ridges extend posteriorly from the protoconid and approach or connect with the entoconid and/or hypoconid (Figure 8). In most cases a single ridge dominates, but the position of this ridge varies continuously from a labial to a lingual position on the tooth. We chose to score the position of the primary ridge from 1 (fully labial) to 7 (fully lingual), and we also assigned a score from 0 to 5 to the strength of the connection between the primary ridge and the entoconid (Table 6). Because as many as three ridges can be present, we also gave a size score from 0 to 5 to the labial, medial, and lingual ridges: 0, absent, to 5, large and connected to the entoconid and/or hypoconid. In specimens with fewer than three ridges the assignment was sometimes arbitrary. These two independent scoring systems helped us to assess the enormous variety of configurations found in the posterior ridges of P3. P2 was also scored using this system. It has the same ridge configuration as P3 but exhibits much less variation.

This effort served only to document that the ridges on P3 are highly variable in all the species studied and therefore have little taxonomic value. We base this conclusion on several observations: 1) roughly the same range of variation is found in all species studied, 2) this range of variation exhibits nearly every conceivable intermediate condition, and 3) the configuration of the ridges, within any sample, does not correlate well with any other characters on P3 or any other tooth. Several variations in the P3 ridges are shown in Figure 8.

Table 7 shows counts of specimens with the main ridge behind the P3 protoconid in various lateral positions for samples from Flagstaff Rim, Pipestone Springs, and Douglas. The vast majority of Chadronian Leptomeryx have the primary ridge in a lingual position, while most Orellan specimens (L. evansi) have it in a medial to labial position. But the range of variation covers nearly the entire spectrum in most large samples with no indication of bimodality, so this is not a reliable character for distinguishing any two species. It is only useful in comparing species with large samples for evidence of relatedness. Of the ten samples compared in Table 7, the earliest (L. yoderi) and the latest (L. evansi, discussed later) exhibit the widest and most curious ranges of variation.

The primary P3 ridge in L. yoderi from Quarry A very commonly occurs in a fully lingual to a fully medial position (Table 7), and this same variation is seen in other early Chadronian populations of Leptomeryx from the Yoder local fauna of Wyoming (Kihm, 1987), the Medicine Pole Hills local fauna of North Dakota (Pearson and Hoganson, 1995), and the Southfork local fauna of Saskatchewan (Storer, 1984). The medial position of the ridge becomes less common up section from Quarry A and is unusual in L. speciosus and L. mammifer from Flagstaff Rim, Ledge Creek, and Pipestone Springs (Table 7). In this character L. speciosus and L. mammifer are more similar to one another than either is to L. yoderi.

Table 7 considers only the most prominent ridge behind the protoconid of P3, which in most cases is much larger than any others. In some cases two ridges connect the protoconid with the entoconid and/or hypoconid and are of nearly equal size, but only the larger one was considered for this measurement. Because as many as three ridges commonly extend posteriorly from the protoconid, separate scores were given for their sizes (Table 6). A presentation of these statistics comparable to that in Table 7 would not be useful because it would fail to account for the relationship among the ridges in individual specimens. Table 8, instead, shows statistics for two ridges based on samples for which the third ridge is fully connected to the entoconid (scored as 5). Only four samples were deemed large enough to be considered: L. yoderi from Quarry A, L. speciosus and L. mammifer from the upper section at Flagstaff Rim, and L. evansi (discussed below) from above the persistent white layer (PWL) at Douglas. These are the best tabular data we can provide to show the enormous individual variation in the ridges of P3.

Though not included in Table 8, we have found that P3s of L. speciosus from Pipestone Springs exhibit the same range of variation seen at Flagstaff Rim, including multiple ridges connecting the protoconid with the entoconid and/or hypoconid. This variation was noted by Tabrum and Fields (1980). The same is true for a population with a smaller mean size from Canyon Ferry, Montana (White, 1954; Table 2), where three of six P3s have two connecting ridges. Clark (1937) noted variation in the size of the accessory ridges behind the P3 protoconid in small Leptomeryx from the Chadronian of South Dakota. The large sample of L. mammifer from Ledge Creek exhibits the same variations seen at Flagstaff Rim, including multiple ridges in various configurations. Extreme variation in P3 ridges is the rule rather than the exception in populations of Leptomeryx.

The Chadronian Leptomeryx species more commonly have a fully- connected lingual ridge on P3 while Orellan L. evansi more commonly has a fully-connected medial ridge, and this is reflected in the totals at the bottom of Table 8. While L. yoderi exhibits more variation than the other Chadronian populations in the position of the primary ridge (Table 7), it shows less variation in the additional ridges (Table 8). Not a single specimen from Quarry A at Flagstaff Rim has two ridges that connect the protoconid with the posterior end of the tooth, though three such specimens have been found in Quarry B, one of which (USNM 366176, Figure 8d) is positively identified as L. yoderi. All Quarry A specimens with a fully-connected lingual ridge have a large but non- connected labial ridge and, in some cases, a smaller medial ridge. All specimens with the connecting ridge in a more medial position have moderate-sized labial and lingual ridges. No bimodality is exhibited in any of the L. yoderi columns of the table. Specimens with two connecting ridges are much more common in the other three species, as are bimodal distributions. This again suggests that L. speciosus and L. mammifer are more similar to each other than either one is to L. yoderi.

Leptomeryx evansi exhibits the greatest bimodality, with additional ridges being either absent or fairly large, so this species displays the greatest variability in both Table 7 and Table 8. As discussed below, L. evansi is easily recognized based on several characters and is most certainly a single species, so premolar variations should not be used to suggest that any of these samples represent coexisting species of the same size. What Table 8 shows, instead, is that for every specimen with a second ridge connecting the protoconid with the entoconid and/or hypoconid, there are many specimens in the population with a second ridge that is large and nearly makes such a connection. The only bimodality exhibited is in the presence/absence of secondary ridges, and since secondary ridges take on an almost infinite variety of shapes and orientations (Figure 8), their absence in some specimens is hardly surprising.

This study demonstrates that previous attempts to use the ridges on P3 for taxonomic distinction are completely invalid. Matthew (1903) stated:
In the third lower premolar the protoconid has two posterior ridges, of which [in Leptomeryx speciosus and L. mammifer] the internal one connects with the heel, and the external one does not; while in L. evansi and the other species from the Oreodon and Leptauchenia Beds, the external ridge connects with the heel, and the internal one does not. In the lower jaw I have observed no entirely constant distinctions, except [this one] in P3.
Matthew, using limited samples, failed to note that P3 has from one to three ridges extending posteriorly from the protoconid and that the position of the connecting ridge (or ridges) can be found in a labial, medial, or lingual position in all of the species he studied. While his observations are true for the majority of specimens, this is clearly not a distinguishing feature even between the rather distinct Leptomeryx species of the Chadronian and the Orellan.

Attempts have also been made to distinguish the various Chadronian species of Leptomeryx using premolars. Storer (1981, 1984) listed three possible characters to distinguish L. yoderi from L. speciosus: 1) small paraconid on P2, 2) less prominent labial "heel" on the P4 hypoconid, and 3) lack of an enclosed basin behind the P3 protoconid (as is occasionally found in L. speciosus when two ridges connect the protoconid and entoconid). Kihm (1987), in his study of L. yoderi from the Yoder local fauna, dismissed the first two of Storer's distinctions as individual variation, which is also confirmed by our sample, but he found the distinction in P3 to be valid. We found the same to be true for the large sample from Quarry A at Flagstaff Rim as indicated by the lack of any L. yoderi specimens with two fully-connected ridges as discussed above. However, one positively identified specimen (USNM 366176, Figure 8d) and two probable specimens (USNM 366201 and 366203) of L. yoderi from Quarry B have the enclosed basin. This distinction appears to be of little value, anyway, because the vast majority of L. speciosus P3s also lack the enclosed basin between two ridges (Table 8).

Kihm (1987) proposed two additional distinguishing characters for L. yoderi, bifurcation of the talonid crest on P2 and presence of a labial cingulum on P3, but as he suspected, larger samples demonstrate that these characters are variable in all species. We have found many premolar variations that have not been previously reported, and it is our firm opinion that no two species of Leptomeryx can be positively distinguished using characters of these teeth. Some character traits are found more commonly in one species than another, however, and these may be useful in tracing the evolution of certain lineages.

Molar Variations
Storer (1981, 1984) stated that an accessory labial cusp (buccal column) in the valley between the protoconid and hypoconid of the molars is more common and larger in Leptomeryx speciosus than in L. yoderi or L. mammifer. We found the opposite to be true (Table 9), though the range of variation is roughly the same for each species measured. This cusp is always present on the M1 of L. yoderi from Quarry A. A corresponding accessory cusp between the hypoconid and hypoconulid of M3 is more common and generally larger in L. speciosus (and L. evansi), however (Table 9). As with premolar variations, large samples demonstrate that these characters of the molars are not species diagnostic.

Storer (1981) and Kihm (1987) provisionally accepted Leptomeryx yoderi as a species distinct from L. speciosus, though all of the differences they listed are actually individual variations. As can be seen in Figures 2-6 and Tables 3, 4, and 6, these two species have very similar size ranges. Our study was unable to identify any distinguishing character between the two except one that Emry discovered prior to our joint work (Emry, 1970; cited by Storer, 1984). This involves a difference in the configuration of a small cusp (entoconulid) and valley on the lingual side of the posterolophid of M3 (Figure 9, Table 10). In this character L. yoderi and the much larger L. mammifer are identical while L. speciosus is distinct from both. This similarity, together with populations of intermediate size from Quarries B and C at Flagstaff Rim (Figures 2-4 and 6), demonstrate beyond reasonable doubt that L. mammifer is a direct descendent of L. yoderi.

In the Leptomeryx yoderi/mammifer lineage the entoconulid is long and narrow, and the anterior and posterior ridges extending from its peak slope downward at about the same angle when viewed lingually. In L. speciosus the entoconulid is much more rounded and pronounced, and it slopes steeply downward posteriorly (Figure 9). We assigned subjective scores for the shape (V-shaped vs. U-shaped) and depth of the valley behind the entoconulid (Subjective Measurements 42 and 43 in Table 6), but these did not help in distinguishing the two lineages. Ultimately we simply assigned each M3 to the lineage to which it belongs based on the shape of the entoconulid (Subjective Measurement 44 in Table 6).

This distinguishing character on M3 is not perfect because a few specimens exhibit an intermediate condition or an odd condition that cannot be classified. There are also many specimens that cannot be classified due to excessive wear or damage to the posterolophid. Nevertheless, the distinctive shape of the M3 entoconulid exhibits remarkable consistency in middle Chadronian populations where size can be used to identify species independently. The large sample of Leptomeryx speciosus from Pipestone Springs and the smaller sample from Canyon Ferry are identical to L. speciosus from Flagstaff Rim in this respect. For early Chadronian samples, this is the only character we have found useful for distinguishing the similar-sized L. yoderi and L. speciosus. It may also hold the clue to their relationship.

As can be seen in Table 10, the large sample from Quarry A at Flagstaff Rim consists entirely of Leptomeryx yoderi type M3s with the exception of three intermediate and two odd cases. Not a single specimen matches L. speciosus. This uniformity, combined with the normal size distribution at Quarry A (Figure 6a), strongly suggests that only a single species is represented (L. yoderi). In the smaller sample from Quarry B, three M3s match L. speciosus, and they are among the smallest specimens in this left-skewed sample (Figure 6b). The same is true for Quarry C (Figure 6c). Only in the upper part of the section at Flagstaff Rim are M3s of L. speciosus and L. mammifer non- overlapping in size (Tables 4-5 and 10).

Kihm (1987) assigned all Leptomeryx from the Yoder fauna to L. yoderi except for two unusually small, low crowned specimens (Tables 2-3). Based on this he correlated the Yoder local fauna with the lower quarries at Flagstaff Rim. Kihm (1987) did not mention the M3 posterolophid configuration of his sample, so this will be described here. The majority of M3s from the Yoder local fauna, including the type specimen of L. yoderi (MCZ 2095), match the L. yoderi/mammifer entoconulid pattern of Figure 9 (SDSM 5343, 5346, 5348, 53295, 53304). The two smallest specimens (SDSM 8659 and 53300) do not match either configuration and may be a different species as Kihm (1987) noted. The two largest of Kihm's "L. yoderi" (SDSM 5345 and 8435) actually match L. speciosus. This suggests that Kihm's correlation between the Yoder local fauna and the Flagstaff Rim section may be suspect, especially when one considers that the localities are only 125 miles apart and that Leptomeryx was a highly mobile animal.

The absence of the small Yoder leptomerycid from all levels at Flagstaff Rim might suggest that the Yoder local fauna is older, predating the extinction of this species. But the joint occurrence of M3s matching both Leptomeryx yoderi and L. speciosus correlates the Yoder local fauna only to Quarry B at Flagstaff Rim. (The large sample from Quarry A lacks L. speciosus while the L. yoderi/mammifer specimens from Quarry C are too big to match L. yoderi from the Yoder local fauna.) We considered the possibility that the Yoder local fauna contains a mixture of several ages, but this seems unlikely because 1) the fossils come from a relatively small area, 2) the fossils have excellent preservation, and 3) the three "morphs" are sometimes found together in small quarry samples.

The M3s of Leptomeryx speciosus from the Yoder local fauna differ from those of Quarry B in being larger than L. yoderi specimens from the same localities (Table 10). Though it is statistically improbable, this could be due to sampling error as all of the Yoder M3s fall within the size range of L. speciosus from the upper part of the Flagstaff Rim section (Table 4).

The situation becomes more complex when one considers the Raben Ranch local fauna of northwest Nebraska, 160 miles east of Flagstaff Rim and 75 miles NNE of Yoder. Ostrander (1980) recovered four isolated M3s of "Leptomeryx sp." from Raben Ranch, two of which match L. yoderi (both SDSM 10191) and two of which match L. speciosus (SDSM 10193 and 10210) in the configuration of the M3 entoconulid (Table 10). The odd thing is that all these teeth are relatively small (M3 length of 9.2 to 9.7 mm). The shortest M3 from Flagstaff Rim is 9.6 mm long (from Quarry B), and the shortest M3s from Pipestone Springs and Canyon Ferry are 9.5 mm long. In size the Raben Ranch specimens match the small Yoder species (Table 3), but in cusp height and other morphological characters they clearly match L. yoderi and L. speciosus. It is peculiar that three geographically close Chadronian localities are known to contain a mixture of the two M3 types (Figure 9), but that none of the three exhibit the same size distribution.

Localities in Saskatchewan and North Dakota also contain Leptomeryx with probable affinities to L. yoderi. Storer (1984) reported a dentary with P3-M3 (ROM 23207) and an isolated M2 (SMNH P1276.3) of L. cf. yoderi from the Southfork local fauna of Saskatchewan, together with several smaller specimens that he referred to L. cf. blacki after a species from California named by Stock (1949). The single complete M3 of each species (ROM 23207 and SMNH P1185.8, respectively) matches L. yoderi in entoconulid pattern (Table 10; Storer, 1984, fig. 2) as well as in crown height and other features. The smallest specimen that Storer referred to L. cf. blacki is no smaller than the Raben Ranch Leptomeryx and may represent the same species (i.e. a small variant of L. yoderi or a closely-related species). We find no justification based on morphology or size range to regard the Southfork Leptomeryx as two different species. If a single species is represented, however, it would have a smaller mean size than L. yoderi of Wyoming.

Storer (1984) based his identification of Leptomeryx cf. blacki in part on the lack of a P1-P2 diastema in SMNH P1185.7. The position of P1 is highly variable in all Flagstaff Rim species. For example, in the 31 specimens of L. yoderi from Quarry A in which the P1-P2 diastema can be measured, its length ranges from 2.7 to 6.5 mm with a mean value of 4.7 mm and a coefficient of variation of 18.3% (Emry 1970). Because this character is so variable, we do not consider the lack of a P1-P2 diastema to be significant.

Leptomeryx has recently been discovered from the Medicine Pole Hills of southwestern North Dakota (Pearson and Hoganson, 1995). In size, the teeth match L. yoderi from the Yoder local fauna and from Quarry A at Flagstaff Rim (Table 10). Of four M3s discovered so far, two (PTM 663 and 1525) match L. yoderi in entoconulid pattern, and the others are too heavily worn to be certain. The posterolophid of M3 in both unworn specimens is smaller relative to the rest of the tooth than in L. yoderi and has a less developed entoconulid with a shallower valley behind it. The molars are very high crowned like L. yoderi, however. The Medicine Pole Hills Leptomeryx may represent an early grade of L. yoderi or possibly a new species.

Evolutionary Relationships
The fact that Leptomeryx yoderi and L. mammifer share the same M3 posterolophid pattern, are distinguishable only by size, and show a gradual size increase up section at Flagstaff Rim is sufficient to document that they are chronospecies resulting from anagenetic evolution. Although Emry (1973) referred to the intermediate sample from Quarry C as an unnamed species, we have decided not to establish a species name for this sample. Such a species would differ from L. yoderi and L. mammifer only in its intermediate mean size, and there is so much overlap in the size ranges that this species could be identified only in samples large enough to define statistically. Therefore, we simply recognize the Quarry C form as a temporally restricted sample of the anagenetic continuum linking L. yoderi with L. mammifer.

The relationship of Leptomeryx yoderi to L. speciosus is more problematic. Only a single character (shape of the M3 entoconulid, Figure 9) reliably distinguishes between them, and that not in every case (Table 10). Leptomeryx speciosus, like L. mammifer, is abundant in the middle Chadronian, and the most likely early Chadronian ancestor for both species is L. yoderi. This raises the possibility that the Flagstaff Rim section illustrates a case of sympatric speciation in Leptomeryx. The principal questions are: 1) whether the Quarry B specimens with L. speciosus type M3 entoconulids descended from the L. yoderi population represented by the Quarry A sample, and 2) whether the Quarry B sample represents two coexisting species (L. yoderi and L. speciosus) or a single species (L. yoderi) with newly-acquired variation (incipient L. speciosus). An alternative explanation is that L. speciosus evolved elsewhere (still probably from L. yoderi) and migrated into the Flagstaff Rim region after the deposition of the Quarry A deposit.

We will not review the literature on speciation except to state that many biologists discount the possibility of sympatric speciation (Cracraft, 1989; Lynch, 1989). The division of an interbreeding population into two species would seem particularly unlikely in a prolific and mobile artiodactyl like Leptomeryx, though it might be possible for distinct herds to develop genetic differences while still occupying the same region. This would be a borderline case of allopatry/sympatry, sometimes called microallopatry. Genetically isolated sympatric populations, or sibling species, have been found in less mobile vertebrates such as salamanders (Larson, 1989). Such cases are usually attributed to habitat fragmentation (another form of microallopatry) which would be less likely to effect an artiodactyl.

The fact that the Leptomeryx sample from Quarry B, like the other samples shown in Table 10, contains few intermediate or odd M3s compared to the number of easily-classified specimens argues for two distinct, coexisting species rather than a single, morphologically diverse species. The asymmetry of the size distribution (Figure 6) and the fact that all of the L. speciosus type M3s are among the smallest specimens in the long left tail of that distribution (Table 10) also suggest the presence of two species. If the Quarry B and Yoder samples are combined (on the assumption of similar age), then M3s of the two types have similar size ranges, but as stated above, correlation of the Yoder fauna with any level at Flagstaff Rim is problematic.

We do not know the duration of the hiatus between Quarry A and B at Flagstaff Rim because no datable ashes exist below Quarry C, so it is not possible to know the length of time available for evolutionary change to take place (Emry 1973, 1992). Leptomeryx yoderi underwent only a slight increase in mean size between Quarry A and B, much less than between Quarry B and C or between Quarry C and the upper portion of the Flagstaff Rim section (Figure 6). Sympatry of two similar-sized species may have driven or accelerated the increase in size of the L. yoderi/mammifer lineage, however. The only hope of unraveling this relationship is to compare the three Flagstaff Rim leptomerycids to see if L. speciosus shares any characters with L. yoderi or L. mammifer that the other lacks.

The most obvious similarity between Leptomeryx yoderi and L. speciosus is their size, although the latter is slightly smaller on average (Tables 3-4, Figure 6). If Quarry B contains two species that can be distinguished based on the shape of the the M3 entoconulid, as suspected, then a size comparison can be made between these two species and L. yoderi from Quarry A. As can be seen in Table 10, L. yoderi from Quarry B ranges to larger size than in Quarry A, while L. speciosus averages smaller than Quarry A L. yoderi (0.2 mm difference in length of M3 in each case). These size range extensions are significant because the Quarry B sample contains so few specimens compared to the Quarry A sample. Above Quarry B, L. speciosus remains roughly the same size while the L. yoderi/mammifer lineage undergoes a marked increase in size. But between Quarries A and B the divergence is nearly symmetric. The only other unique similarity between L. yoderi (Quarry A) and L. speciosus (and L. evansi) is that some specimens have the main ridge behind the P3 protoconid in a more labial than lingual position; no specimen of L. mammifer with this condition has yet been found (Table 7).

Leptomeryx speciosus and L. mammifer share several unique similarities. Unlike L. yoderi from Quarry A, both have relatively fewer P3s with the main ridge in a medial position (Table 7), and both include P3s with two ridges connecting the protoconid with the entoconid and/or hypoconid (Table 8). Both these characters first appear in Quarry B, so they could be shared, derived characters (though they are relatively uncommon in both species). Rarely L. speciosus and L. mammifer show some development of the Paleomeryx fold in the molars (Table 11, discussed below), but no cases of this have been found below Quarry C.

In contrast to these somewhat dubious similarities, Leptomeryx yoderi and L. mammifer share the same unique configuration of the M3 entoconulid (Figure 9, Table 10) in nearly every specimen. They also have a similar distribution pattern of accessory labial cusps in the molars (Table 9) and are less likely than L. speciosus to have a large medial ridge behind the protoconid of P3 when a fully-connected lingual ridge is present (Table 8). The only significant difference between L. yoderi and L. mammifer is size, and at Flagstaff Rim the size transition is well documented. No comparable graded link has been found between L. yoderi and L. speciosus. While L. yoderi remains the most likely ancestor for L. speciosus, it is uncertain whether L. yoderi speciated sympatrically or, alternately, whether L. speciosus developed from L. yoderi or a related species in another location.

In conclusion, the dominant Chadronian species of Leptomeryx are closely related and represent a grade of evolution beyond their Duchesnean counterparts. Six stages in the evolution of Chadronian Leptomeryx seem to be indicated: 1) earliest Chadronian: Leptomeryx from the Medicine Pole Hills of North Dakota may represent an early grade of L. yoderi or its immediate ancestor; 2) middle early Chadronian: L. yoderi became a widespread species and is the only species found in Quarry A at Flagstaff Rim; 3) late early Chadronian: Leptomeryx speciosus appeared, probably a descendant of L. yoderi, and these two similar-sized species are found together in Quarry B; the Yoder local fauna, considered early Chadronian by Kihm (1987), and the Raben Ranch local fauna, considered middle Chadronian by Ostrander (1980), seem to best fit this stage, though the match is imperfect in both cases; 4) early middle Chadronian: L. yoderi increased in size, approaching L. mammifer, and is found with L. speciosus in Quarry C; 5) middle middle to early late Chadronian: L. mammifer developed as a species distinctly larger than L. speciosus, and the two species became widespread and stable; both are found throughout the upper part of the Flagstaff Rim section, at Pipestone Springs, and in the Calf Creek local fauna of the Cypress Hills; and 6) latest Chadronian: L. mammifer became extinct, and L. speciosus underwent some modifications approaching the Orellan L. evansi grade (discussed below).

THE DEVELOPMENT OF LEPTOMERYX EVANSI FROM LEPTOMERYX SPECIOSUS
The highly fossiliferous Orellan deposits of the Great Plains contain large numbers of a small leptomerycid that includes the type specimen of the genotypic species, Leptomeryx evansi Leidy (1853). Clark and Kietzke (1967) found L. evansi to be the most abundant animal in the open plains fauna of the South Dakota badlands. Emry et al. (1987) believed this species to be restricted to the Orellan, attributing reports in older and younger rocks to misidentifications.

Leptomeryx evansi is slightly smaller on average than L. speciosus and has a number of unique characters. First, the primary ridge behind the P3 protoconid is usually more labial than lingual (Table 7), though as in the Chadronian species, there is considerable variation in this feature. It commonly has two ridges connecting the P3 protoconid with the entoconid and/or hypoconid, one medial and one lingual (Table 8). Secondly, most specimens have a well-developed Paleomeryx fold on each of the lower molars (Table 11). Leptomeryx evansi also tends to have more strongly developed vertical crenulations around the tooth crowns, especially on the lingual side of the lower teeth, though this is a variable feature. Crenulations on the lingual side of the metaconid and entoconid of the lower molars tend to be better developed in L. evansi than in other species (Figure 11). In the critical feature of the M3 posterolophid, L. evansi matches L. speciosus rather than L. yoderi and L. mammifer (Figure 9). In fact the M3 entoconulid tends to be even broader and more rounded posteriorly in L. evansi than in L. speciosus, or at least more consistently so.

We have not found any variation that would suggest that more than a single species of Leptomeryx existed in the Orellan of the Great Plains. The distinguishing characters are variable in their degree of development, just as in the Chadronian species, but the range of variation is generally continuous rather than bimodal, and the degree of development of the various characters on individuals is not well correlated. For example, there are specimens with the major ridge on P3 in a labial position that have little or no development of the Paleomeryx fold on the molars (USNM 443715), and there are specimens with the P3 ridge in the lingual position with strongly-developed Paleomeryx folds (AMNH 127017). The Paleomeryx fold can even be strongly developed on some molars of a jaw and entirely absent on adjacent ones (AMNH 127019, USNM 443664), though this is unusual. The type specimen of L. evansi (USNM 157) lacks the P3 but has molars with prominent Paleomeryx folds. Several other Orellan and Whitneyan species were named by Cook (1934) and Frick (1937). We have not examined the type specimens of these species but suspect that they are synonyms of L. evansi.

South and east of Douglas, Wyoming, about 90 miles east of Flagstaff Rim, White River deposition is apparently continuous from middle Chadronian through Orellan time (Evanoff et al., 1992). The Chadronian/Orellan transition is approximately at a volcanic ash bed referred to as the persistent white layer (PWL), "Glory Hole Ash," purplish-white ash, 100-foot white ash, or No. 5 tuff (Evanoff et al., 1992:117). Except for the uppermost part of the Chadron Formation, the entire section is fossiliferous (Kron, 1978; Evanoff et al., 1992).

The lower part of the Chadron Formation at Douglas is temporally equivalent to at least part of the upper portion of the section at Flagstaff Rim (Prothero and Swisher, 1992; Evanoff et al., 1992), though the exact correlation is uncertain due in part to the paucity of fossils in the uppermost levels at Flagstaff Rim. Leptomeryx speciosus appears to persist later at Flagstaff Rim than does L. mammifer (Figures 2-4; see also Emry, 1992), and a population with a slightly smaller mean size than L. speciosus of Flagstaff Rim is found below the PWL at Douglas (Figure 11). The L. evansi population above the PWL at Douglas has a mean size that is slightly smaller yet, and the size ranges are similar. A comparison of these three successive populations suggests the strong possibility that L. evansi descended from L. speciosus in the latest Chadronian. This relationship has also been proposed by Prothero (1985a, 1985b, and unpublished data).

Our study of Chadronian Leptomeryx has shown that the most diagnostic character for distinguishing species of similar size is the configuration of the M3 entoconulid. Because the variant found in L. speciosus is a derived character that first appears in that species and is shared by L. evansi, a strong case can be made for an ancestor/descendant relationship. The few M3s found below the PWL at Douglas all share this condition as well. Another similarity between the two species is found in the labial accessory cusps on the molars (Table 9). Both have less frequent or smaller cusps between the protoconid and hypoconid than is seen in L. yoderi and L. mammifer (being even more uncommon in L. evansi, especially on M1), and both more commonly have an accessory cusp between the hypoconid and hypoconulid of M3 (Table 9).

If the Leptomeryx population from below the PWL at Douglas is intermediate between L. speciosus from Flagstaff Rim and L. evansi, then an even stronger case can be made that these are chronospecies. Unfortunately this population is very small, but it does exhibit remarkable intermediacy. In mean size it is intermediate between Flagstaff Rim L. speciosus and L. evansi (Tables 4 and 12), though somewhat closer to L. evansi. Only four specimens include P3, all of which have a fully lingual main ridge behind the protoconid as in L. speciosus (Table 7). In terms of the Paleomeryx fold this population is perfectly intermediate, having the fold much more commonly than in L. speciosus but much less commonly than in L. evansi (Table 11). The folds even become more common higher in the Douglas section (less common 12 to 16 m below the PWL; more common 9 to 12 m below the PWL). Because three of the four P3s come from the lowest level at Douglas (16 m below the PWL), the distinguishing characters of this tooth may also undergo gradual change. Evidence for such a conclusion comes from specimens of L. evansi found above the PWL at Douglas. Those P3s that have the dominant ridge in the lingual position (as in L. speciosus) occur more frequently near the Chadronian/Orellan boundary than they do higher in the Douglas section.

Assigning the population from below the PWL at Douglas to a species would be a judgement call between Leptomeryx speciosus and L. evansi. Taxonomically the same sort of dilemma is faced as with the L. yoderi/mammifer population from Quarry C at Flagstaff Rim. We therefore conclude that L. speciosus and L. evansi are chronospecies exhibiting gradual anagenesis across the Chadronian/Orellan boundary.

Chadronian Leptomeryx resembling Leptomeryx evansi
Before leaving the topic of L. evansi, we need to address some unusual leptomerycid specimens from the Pipestone Springs area that resemble this species in several characters. If L. evansi or a likely precursor distinct from L. speciosus could be found in the early or middle Chadronian, then the genealogy presented above would be undermined.

MPUM 3296 is a dentary from Little Pipestone Creek (near Pipestone Springs) that is as large as the largest L. mammifer specimens from Flagstaff Rim and exhibits several L. evansi features: 1) the main ridge behind the P3 protoconid is labial rather than lingual, 2) the Paleomeryx fold is moderately developed on the molars, 3) the enamel is strongly crenulated (though no more than in some L. mammifer specimens), and 4) the M3 posterolophid matches L. speciosus and L. evansi more closely than L. mammifer. This specimen differs from L. evansi and the other species of Leptomeryx, however, in lacking a deep and relatively wide depression in the posterolophid of M3. MPUM 3296 also has an unusual accessory ridge extending posteriorly from the metaconid on M1.

USNM 19001, a large M3 from Pipestone Springs, has stronger development of the Paleomeryx fold than any M3 from Flagstaff Rim and has strongly crenulated enamel. The entoconulid of USNM 19001 has a rounded posterior edge as in L. speciosus and L. evansi, but unlike all other Leptomeryx specimens we have examined, the entoconulid has a small ridge projecting into the medial basin toward the hypoconulid. So while the posterolophids of these molars match L. evansi in some respects, in other ways they are entirely unique.

The P3 ridge configuration on MPUM 3296 and the crenulated enamel on both specimens can be dismissed as individual variations in Leptomeryx mammifer, as can the Paleomeryx fold because these are occasionally found on L. mammifer molars from Flagstaff Rim (Tables 7 and 11). The M3 entoconulid is more problematic, as is its presence in specimens that possess several other L. evansi characters. There are three possible explanations: 1) these large specimens are merely unusual variants of L. mammifer, 2) there is a second large species of Leptomeryx from the Pipestone Springs area which has developed features paralleling L. evansi, or 3) this species shares a unique common ancestry with L. evansi independent of L. speciosus and L. mammifer. We view the third possibility as very unlikely. Leptomeryx evansi has never been found in the Chadronian, and we have provided substantial evidence that it evolved from L. speciosus across the Chadronian/Orellan boundary. The L. evansi features seen in MUPM 3296 and USNM 19001 can all be found as variants in L. speciosus and/or L. mammifer (Tables 8-11). The vast size difference and lack of intermediate populations argue against a close relationship with L. evansi.

The sample of large Leptomeryx specimens from Pipestone Springs and other Montana localities is very small compared the sample from Flagstaff Rim (see Table 5, though not all Pipestone Springs material is included). The presence of greater morphological variation in this much smaller sample is perhaps the strongest evidence that more than one species is represented in the Montana material. However, larger samples are needed to determine for certain whether these specimens with unusual combinations of characters comprise a new species or merely represent individual or geographic variations in L. mammifer.

SUMMARY AND CONCLUSIONS
The leptomerycids, a cohesive group of about seven species, first appear in North America in the Duchesnean, become abundant in the Chadronian and Orellan, are less common in the Whitneyan (Cook, 1934; Frick, 1937), and are present but rare in the Arikareean (Taylor and Webb, 1976). Individual variation is often more pronounced than differences between species, and this has long caused taxonomic confusion. Our study of large samples from Flagstaff Rim, Wyoming, has succeeded in determining which characters are useful for identification of species, and this information can now be applied to smaller samples. Most importantly we have documented that premolar ridge configurations, especially on P3, are highly variable in each species studied and therefore offer little help in identification, their prominence notwithstanding.

The early evolution of the group is obscured by small sample sizes and a predominance of isolated teeth rather than full dentitions. Duchesnean forms, called Hendryomeryx by some authors, exhibit a slightly lower-crowned condition than their Chadronian and Orellan counterparts. These lower-crowned forms appear to extend into the Chadronian in the Cypress Hills, Saskatchewan (Storer, 1981) and at Yoder, Wyoming (Kihm, 1987).

Leptomeryx yoderi became the dominant species of the early Chadronian and is the only species found in the enormous sample from Quarry A at Flagstaff Rim. It is slightly larger than its predecessors and has higher-crowned, steeper-walled teeth. This species underwent a gradual size increase and evolved into L. mammifer of the middle Chadronian. This anagenetic shift is well-documented at Flagstaff Rim. Prior to this increase in size, L. yoderi apparently gave rise to L. speciosus, a species of similar size that can only be distinguished from it by a change in the shape of the entoconulid on the last molar. Leptomeryx speciosus and L. mammifer are found together in middle Chadronian deposits from Wyoming to Saskatchewan. Large samples exhibit clear bimodality in size, so that individual specimens can be distinguished by size as well as by M3 entoconulid configuration.

Leptomeryx mammifer became extinct before the end of the Chadronian, leaving L. speciosus as the only known surviving species of the genus. Leptomeryx speciosus underwent a size reduction at the end of the Chadronian and gradually evolved into L. evansi, apparently the only Leptomeryx species of the Orellan. This anagenetic transition is documented by an intermediate population from the latest Chadronian of Douglas, Wyoming, where L. evansi characters become more common stratigraphically higher in the section. The most distinctive characters of L. evansi, the labial ridge behind the protoconid on P3 and Paleomeryx fold on the molars, are only rarely found in L. speciosus and almost never in the same specimens.

Changes in Leptomeryx across the Duchesnean/Chadronian boundary can be accounted for by normal evolutionary adaptation. Leptomeryx yoderi, being larger and possessing higher-crowned teeth than its contemporaries, may well have out-competed them. The Chadronian/Orellan transition, however, is marked by the extinction of a large species and the size reduction and morphological modification of a smaller species. No new predators or competitors appear at this boundary, so these changes must be attributable to climatic factors. Evanoff et al. (1992) provide evidence for a fairly rapid change from moist subtropical to semiarid warm temperate climate at the Chadronian/Orellan boundary. The standardization in L. evansi of characters that are rare in its predecessors, as well as the decrease in size, suggests strong selection pressure, a population bottleneck, or both. Nevertheless, L. evansi survived this transition to become one of the most common and widespread species of the Orellan.

ACKNOWLEDGEMENTS
We thank Robert W. Purdy of the Smithsonian Institution, Richard H. Tedford of the American Museum of Natural History, James E. Martin and Janet L. Whitmore of the South Dakota School of Mines and Technology, Charles R. Schaff of the Harvard University Museum of Comparative Zoology, and Dean A. Pearson of the Pioneer Trails Museum (Bowman, North Dakota) for providing us with Leptomeryx specimens and associated data in their care. John E. Storer and Alan R. Tabrum provided helpful critiques of the manuscript even though they didn't always agree with our conclusions.

Partial funding for this project was provided by a Smithsonian Institution postdoctoral fellowship to the senior author.

LITERATURE CITED

Black, C. C. 1978. Paleontology and geology of the Badwater Creek Area, central Wyoming. Part 14. The Artiodactyls. Annals of Carnegie Museum, 47:223-259.

Clark, J. 1937. The stratigraphy and paleontology of the Chadron Formation in the Big Badlands of South Dakota. Annals of Carnegie Museum, 25:261-350.

Clark, J., and K. K. Kietzke. 1967. Paleoecology of the Lower Nodular Zone, Brule Formation, in the Big Badlands of South Dakota. Pp. 111-137 in Clark, J., J. R. Beerbower, and K. K. Kietzke (eds.), Oligocene Sedimentation, Stratigraphy, Paleoecology and Paleoclimatology of the Big Badlands of South Dakota. Fieldiana: Geology Memoirs, 5:1-158.

Cook, H. J. 1934. New artiodactyls from the Oligocene and lower Miocene of Nebraska. American Midland Naturalist, 15:148-165.

Cope, E. D. 1885. The Vertebrata of the Swift Current Creek region of the Cypress Hills. Annual Report of the Geological and Natural History Survey of Canada, 1:79-85.

Cope, E. D. 1889. The Vertebrata of the Swift Current River. American Naturalist, 23:151-155.

Cracraft, J. 1989. Speciation and its ontogeny: the empirical consequences of alternative species concepts for understanding patterns and processes of differentiation. P. 28-59 in D. Otte and J. A. Ender (eds.) Speciation and its Consequences. Sinuar Associates, Sunderland, Massachusetts, 679 p.

Emry, R. J. 1970. Stratigraphy and Paleontology of the Flagstaff Rim Area, Natrona County, Wyoming. Unpublished Ph.D. dissertation, Columbia University, New York, 195 p.

Emry, R. J. 1973. Stratigraphy and preliminary biostratigraphy of the Flagstaff Rim area, Natrona County, Wyoming. Smithsonian Contributions to Paleobiology, 18:1-43.

Emry, R. J. 1992. Mammalian range zones in the Chadronian White River Formation at Flagstaff Rim, Wyoming. Pp. 106-115 in D. R. Prothero and W. A. Berggren (eds.), Eocene-Oligocene Climatic and Biotic Evolution. Princeton University Press, 568 p.

Emry, R. J., P. R. Bjork, and L. S. Russell. 1987. The Chadronian, Orellan, and Whitneyan Land Mammal Ages; pp. 119-152 in M. O. Woodburne (ed.), Cenozoic Mammals of North America: Geochronology and Biostratigraphy. University of California Press, Berkeley, 336 p.

Evanoff, E., D. R. Prothero and R. H. Lander. 1992. Eocene-Oligocene climatic change in North America: the White River Formation near Douglas, east-central Wyoming. Pp. 116-130 in D. R. Prothero and W. A. Berggren (eds.), Eocene-Oligocene Climatic and Biotic Evolution. Princeton University Press, 568 p.

Frick, C. 1937. Horned ruminants of North America. American Museum of Natural History Bulletin, 69:1-669.

Heaton, T. H. 1989. Cladogenesis in a lineage of Leptomeryx (Artiodactyla, Mammalia) from the Chadronian of Flagstaff Rim, Natrona County, Wyoming (Abstract). Journal of Vertebrate Paleontology, 9:24A-25A.

Kihm, A. J. 1987. Mammalian paleontology and geology of the Yoder Member, Chadron Formation, east-central Wyoming. Dakoterra, 3:28-45.

Kron, D. G. 1978. Oligocene vertebrate paleontology of the Dilts Ranch area, Converse County, Wyoming. Unpublished M.S. thesis, University of Wyoming, Laramie, 185 p.

Lambe, L. M. 1908. The vertebrata of the Oligocene of the Cypress Hills, Saskatchewan. Contributions to Canadian Paleontology, 3:1-65.

Larson, A. 1989. The relationship between speciation and morphological evolution. Pp. 579-598 in D. Otte and J. A. Ender (eds.) Speciation and its Consequences. Sinuar Associates, Sunderland, Massachusetts, 679 p.

Leidy, J. 1853. Remarks on a collection of fossil Mammalia from Nebraska. Proceedings of the Academy of Natural Sciences of Philadelphia, 6:392-394.

Lynch, J. D. 1989. The gauge of speciation: on the frequencies of modes of speciation. Pp. 527-553 in D. Otte and J. A. Ender (eds.) Speciation and its Consequences. Sinuar Associates, Sunderland, Massachusetts, 679 p.

Matthew, W. D. 1903. The fauna of the Titanotherium beds of Pipestone Springs, Montana. Bulletin of the American Museum of Natural History, 19:197-226.

Ostrander, G. 1980. Mammalia of the early Oligocene (Chadronian) Raben Ranch local fauna. Unpublished M.S. thesis, South Dakota School of Mines and Technology, Rapid City, 288 p.

Pearson, D. A., and J. W. Hoganson. 1995. The Medicine Pole Hills local fauna: Chadron Formation (Eocene: Chadronian), Bowman County, North Dakota. Proceedings of the North Dakota Academy of Science, 49:65.

Prothero, D. R. 1985a. Mid-Oligocene extinction event in North American land mammals. Science, 229:550-551.

Prothero, D. R. 1985b. North American mammalian diversity and Eocene-Oligocene extinctions. Paleobiology, 11(4):389-405.

Prothero, D. R., and C. C. Swisher III. 1992. Magnetostratigraphy and geochronology of the terrestrial Eocene-Oligocene transition in North America. Pp. 46-73 in D. R. Prothero and W. A. Berggren (eds.), Eocene-Oligocene Climatic and Biotic Evolution. Princeton University Press, 568 p.

Schlaikjer, E. M. 1935. Contributions to the stratigraphy and paleontology of the Goshen Hole Area, Wyoming. III. A new basal Oligocene formation. Bulletin of the Museum of Comparative Zoology, 76:71-93.

Stock, C. 1949. Mammalian fauna from the Titus Canyon Formation, California. Contributions to Paleontology, 8:231-244.

Storer, J. E. 1981. Leptomerycid Artiodactyla of the Calf Creek Local Fauna (Cypress Hills Formation, Oligocene, Chadronian), Saskatchewan. Saskatchewan Museum of Natural History Contributions, 3:1-32.

Storer, J. E. 1984. Fossil mammals of the Southfork local fauna (early Chadronian) of Saskatchewan. Canadian Journal of Earth Sciences, 21:1400-1405.

Tabrum, A. R., and R. W. Fields. 1980. Revised mammalian faunal list for the Pipestone Springs local fauna (Chadronian, early Oligocene), Jefferson County, Montana. Northwest Geology, 9:45-51.

Taylor, B. E., and S. D. Webb. 1976. Miocene Leptomerycidae (Artiodactyla, Ruminantia) and their relationships. American Museum Novitates, 2596:1-22.

White, T. E. 1954. Preliminary analysis of the fossil vertebrates of the Canyon Ferry Reservoir area. Proceedings of the U.S. National Museum, 103:395-438.

Wilson, J. A. 1974. Early Tertiary vertebrate faunas, Vieja Group and Buck Hill Group, Trans-Pecos Texas: Protoceratidae, Camelidae, Hypertragulidae. Texas Memorial Museum Bulletin, 23:1-34.


Figure 1. Comparison of two dentaries of Leptomeryx from Flagstaff Rim, Wyoming, in occlusal and lingual view showing cheek tooth series. A and B are of L. speciosus (AMNH, Frick Collection Field No. B.H. 43-859-AN). C and D are of L. mammifer (AMNH, Frick Collection Field No. B.H. 57-1427-1). A and C are x2; B and D are natural size.

Figure 2. Bivariate plot of 129 Leptomeryx specimens from Flagstaff Rim, Wyoming, showing the length of P2-M3 vs. level in the stratigraphic section. Specimens from the lower part of the section are concentrated in three quarry deposits, the lowest of which contains only L. yoderi. This measurement of six consecutive teeth allows for excellent resolution between L. speciosus (small) and L. mammifer (large) in the upper part of the section, but the sample size is small. See Tables 3-6 for tooth size statistics on each species.

Figure 3. Bivariate plot of 466 Leptomeryx specimens from Flagstaff Rim, Wyoming, showing the length of M1-3 vs. level in the stratigraphic section. The measurement of three teeth increases the sample size over that seen in Figure 2 and still allows good resolution between L. speciosus and L. mammifer. See histograms of four stratigraphic samples based on this measurement in Figure 6.

Figure 4. Bivariate plot of 928 Leptomeryx specimens from Flagstaff Rim, Wyoming, showing the length of M2, the most abundantly preserved tooth, vs. level in the stratigraphic section. The measurement of a single tooth provides an enormous sample size but reduces the resolution between L. speciosus and L. mammifer seen in Figures 2-3.

Figure 5. Bivariate plot of 111 Leptomeryx specimens from Ledge Creek, Wyoming, showing the length of M2 vs. level in the stratigraphic section. Leptomeryx mammifer (large) greatly outnumbers L. speciosus (small) in the upper part of the section. Comparisons for the lower part of the section are obscured by the small sample size.

Figure 6. Size histograms, based on M1-3 length, of four stratigraphic samples of Leptomeryx from Flagstaff Rim, Wyoming. Quarry A ("Low Pocket") is a localized deposit (probably a single carnivore den) containing Leptomeryx almost exclusively. Quarry B ("Dry Hole Pocket") and Quarry C ("B-44 Pocket") are also localized deposits; a few specimens from above and below Quarry C are included in that sample. Because L. speciosus and L. mammifer remain in relative stasis throughout the upper part of the section (Figures 2-4), all specimens from above 50 m have been grouped into histogram d. Only L. yoderi is found in Quarry A, whereas two species coexist in the other three samples. Leptomeryx yoderi can be traced through size increase to L. mammifer in this diagram. Leptomeryx speciosus remains about the same size in samples B, C, and D. See Tables 3-6 for tooth size statistics on each species.

Figure 7. Three diagrams of the lower dentition of Leptomeryx in occlusal view: a. tooth cusps referred to in the text, b. objective measurements made with calipers, and c. subjective measurements (scores) given to various accessory cusps, ridges, and other variable features. See Table 6 for a list of the measurements illustrated and summary statistics.

Figure 8. Diagram of twelve Leptomeryx P3s showing variability of the ridge(s) extending posteriorly from the protoconid. See Tables 7-8 for statistics concerning these ridges. MCZ 2095 has the most typical pattern seen in L. yoderi, L. speciosus, and L. mammifer, while ND 238.4 shows the most typical pattern for L. evansi. But all species studied contain variants such as those seen here, so these differences are not species diagnostic.

Figure 9. Diagrams of Leptomeryx M3s in lingual and occlusal view showing the different configuration of the entoconulid in L. yoderi/mammifer and L. speciosus/evansi. See Table 10 for counts of specimens with these configurations from several localities.

Figure 10. Diagrams of Leptomeryx M2s in occlusal view showing the development of the Paleomeryx fold and the greater incidence of crenulations in L. evansi compared to L. speciosus. Statistics concerning the Paleomeryx fold can be found in Table 11. The accessory cusp in the labial valley is found more commonly in L. speciosus than in L. evansi (Table 9).

Figure 11. Bivariate plot of 116 Leptomeryx specimens from Douglas, Wyoming, showing the length of M2 vs. level in the stratigraphic section. The Chadronian/Orellan boundary is marked by the persistent white layer (PWL). The Orellan specimens are L. evansi while the Chadronian specimens are a population showing characters intermediate between L. speciosus and L. evansi. See Table 12 for size statistics on these two stratigraphic populations.

Table 1. Tooth length measurements (mm) for the three species assigned to "Hendryomeryx": H. defordi from Trans-Pecos Texas (late Duchesnean), H. wilsoni from Badwater Creek, Wyoming (early Duchesnean), and H. esulcatus from the Cypress Hills, Saskatchewan (middle Chadronian). All Trans-Pecos Texas and Badwater leptomerycid material is included here, whereas the Cypress Hills measurements represent those specimens assigned by Storer (1981) to Hendryomeryx rather than Leptomeryx.

Table 2. Tooth length measurements (mm) for small leptomerycids from two early Chadronian localities in Wyoming and one locality of less certain age in Montana. All West Canyon Creek and Canyon Ferry leptomerycid material is included here, whereas the Yoder measurements represent only those specimens considered by Kihm (1987) to be too small for Leptomeryx yoderi. The West Canyon Creek and Canyon Ferry material was measured by Heaton.

Table 3. Tooth length measurements (mm) for two populations and the type specimen of Leptomeryx yoderi, all from eastern Wyoming. The type is a distorted dentary from the Yoder local fauna collected by Schlaikjer (1935). All leptomerycid material from Quarry A at Flagstaff Rim (Figure 6a) and from the Yoder local fauna is included except for two Yoder dentaries considered by Kihm (1987) to be too small to be L. yoderi (see Table 2). The Flagstaff Rim and type material was measured by Heaton.

Table 4. Tooth length measurements (mm) for three populations of Leptomeryx speciosus from Flagstaff Rim (Figure 6d) and Ledge Creek (Figure 5), Wyoming, and Pipestone Springs, Montana. Included are all specimens judged on size and morphology to be L. speciosus rather than L. mammifer. All measurements were made by Heaton.

Table 5. Tooth length measurements (mm) for three populations of Leptomeryx mammifer from Flagstaff Rim (Figure 6d) and Ledge Creek (Figure 5), Wyoming, and Pipestone Springs, Montana. Included are all specimens judged on size and morphology to be L. mammifer rather than L. speciosus. All measurements were made by Heaton. The type specimen of L. mammifer from the Cypress Hills of Saskatchewan (NMC 6278) has an M2 length of 10.2; an M3 from the Calf Creek local fauna (SMNH P1585.1230) has a length of 15.2 mm (Storer, 1981).

Table 6. List of objective (44) and subjective (45) measurements taken on Leptomeryx dentaries used in this study, including summary statistics of four species. See Figure 7 for illustrations. The L. yoderi sample is from Quarry A, and the L. mammifer and L. speciosus samples are from the upper part of the section at Flagstaff Rim, Wyoming. The L. evansi sample is from above the PWL at Douglas, Wyoming.

Table 7. Lateral position of largest ridge extending posteriorly from the protoconid of P3 for ten samples of Leptomeryx. Some variations are illustrated in Figure 8. The sample from Quarry B at Flagstaff Rim is a mixture of L. yoderi and L. speciosus, most of which cannot be distinguished. The sample from below the PWL at Douglas is a species intermediate between L. speciosus and L. evansi (Table 12).

Table 8. Size of secondary ridges extending posteriorly from the protoconid of P3 of Leptomeryx. Some variations are illustrated in Figure 8. Each P3 was given a score for a labial, medial, and lingual ridge from 0 (absent) to 5 (large and connecting to entoconid and/or hypoconid). In this table, scores for two ridges are given for the sample in which the third ridge (lingual or medial) was given a score of 5. Only four samples were considered large enough for inclusion: L. yoderi (Y), L. mammifer (M), and L. speciosus (S) from Flagstaff Rim, and L. evansi (E) from Douglas, Wyoming.

Table 9. Size of accessory cusps (buccal columns) on the labial margin of the tooth in the valley between the protoconid and hypoconid of the molars (hypoconid and hypoconulid of M3 posterior; Figure 7) for four species of Leptomeryx: L. yoderi (Y), L. mammifer (M), and L. speciosus (S) from Flagstaff Rim, and L. evansi (E) from Douglas, Wyoming. Each cusp was given a score from 0 (absent) to 4 (large).

Table 10. Classification of Leptomeryx M3s from Flagstaff Rim and several other localities that contain L. yoderi based on the configuration of the entoconulid as illustrated in Figure 9. The categories are: 1) L. speciosus configuration (S), 2) L. yoderi and L. mammifer configuration (Y/M), 3) intermediate or mixed characters of L. speciosus and L. yoderi/mammifer (I), and 4) odd configuration not matching either type (O). The two samples from the upper section at Flagstaff Rim include only USNM specimens. Size ranges (length of M3 in mm) are given for the "S" and "Y-M" samples. The two "O" specimens from the Yoder local fauna are the two small specimens listed in Table 2.

Table 11. Degree of development of the "Paleomeryx fold" (Figure 10) on the molars of eight populations of Leptomeryx ranging from 0 (absent) to 4 (large). All specimens from Quarries A and B from Flagstaff Rim are combined, as are the two species from Quarry C. Specimens from the upper part of the section at Flagstaff Rim and from Pipestone Springs are separated by size into L. speciosus (S) and L. mammifer (M). Specimens from Douglas are separated into those below the persistent white layer (PWL) and those above it, which are L. evansi (E). Counts are shown separately for each molar and combined.

Table 12. Tooth length measurements (mm) for two stratigraphically restricted populations of Leptomeryx from the Douglas area, Wyoming, and the type specimen of L. evansi from the Big Badlands of South Dakota. The population from above the persistent white layer (PWL) is L. evansi, whereas the population from below the PWL appears to be ancestral to that species. All measurements were made by Heaton.


Timothy H. Heaton: E-mail, Home page, Phone (605) 677-6122, FAX (605) 677-6121