Localities

Daryl Domning and Gary Morgan began collecting vertebrate fossils in 1981 from nearshore marine sediments of the Parachucla Formation along the banks of the Suwannee River in northern Florida. They made several collecting trips during most years throughout the 1980s and early 1990s. The original purpose of these trips was to prospect for sirenian fossils that are well known from these strata (Reinhart, 1976; Domning, 1989a, 1989b, 1997). A rich and varied marine vertebrate fauna occurs in association with the fossil sirenians; however, from a biostratigraphic perspective, the most significant aspect of the fauna from these beds is the rare, but regular, occurrence of land mammals.

Morgan (1989) named the White Springs L.F. for the latest Oligocene (early Arikareean) vertebrate fauna recovered from sediments exposed along the banks of the Suwannee River in the vicinity of White Springs. The White Springs L.F. includes vertebrate fossils collected from the Parachucla Formation along about 10 km of the Suwannee River. Most of the fossils referred to the White Springs L.F. were derived from three sites, designated White Springs 1A, 3A, and 3B (Morgan, 1989). Precise locality information for these sites is on file in the UF Vertebrate Paleontology Collection.

The White Springs 1A and 3B sites have produced diverse faunas of nearshore marine vertebrates (sharks, rays, bony fish, marine crocodiles, and sirenians), along with small, but important, samples of terrestrial mammals. White Springs 1A (UF locality CO48) extends for several hundred meters along the east bank of the Suwannee River near White Springs. This site has produced two taxa of dugongid sirenians, including a skull of Dioplotherium manigaulti and a skull and partial skeleton of Metaxytherium sp., as well as an associated fauna of both marine and terrestrial vertebrates (Domning, 1989a, 1989b; Morgan, 1989). Routine screenwashing of sediments from the plaster jacket in which the Metaxytherium skeleton was collected produced an associated fauna of nearshore marine vertebrates, as well as an isolated rodent tooth. The discovery of a terrestrial mammal at White Springs 1A was unexpected and led to intensive screenwashing to augment the terrestrial component of the fauna. In addition to the marine taxa, the White Springs 1A fauna now consists of a horse, shrew, small carnivore, rabbit, and at least five species of rodents.

White Springs 3B (UF locality CO61) is a small site (<100 m²) located in the middle of the Suwannee River near White Springs. This site also has produced a varied marine vertebrate fauna, but is better known for the presence of articulated partial skeletons and isolated elements of the oreodont, Mesoreodon floridensis, n. sp., described herein. A partial skeleton of a medium-sized camelid (cf. Oxydactylus sp.) was the first fossil discovered at the White Springs 3B site. Two species of smaller camels, an equid, and a small rhinoceros also have been recovered from White Springs 3B, primarily represented by fragmentary fossils. UF field crews also collected and screenwashed a large amount of sediment from White Springs 3B, and recovered a fairly diverse sample of small terrestrial vertebrates, including a typhlopid snake, two species of boid snakes, a marsupial, two species of bats, and seven species of rodents.

The vertebrate-bearing strata of the Parachucla Formation have been traced along the banks of the Suwannee River for the entire distance between White Springs 1A and 3B, confirming that these two sites are essentially lateral equivalents, occurring stratigraphically within 2 m of one another. These two sites also share several species of land mammals, including a horse and four rodents. Therefore, the vertebrate faunas from White Springs sites 1A and 3B are combined into the White Springs L.F. (Morgan, 1989).

Stratigraphy

Morgan (1989) referred the stratigraphic interval that produces the White Springs L.F. to the Porters Landing Member of the Parachucla Formation (following Huddlestun, 1988). The lithology of the bone-bearing unit is a light gray to buff to brown, slightly clayey, fine quartz sand with a minor percentage of fine phosphatic grains. At the two primary vertebrate sites, White Springs 1A and 3B, invertebrate fossils are rare except for poorly preserved silicified oysters and pectens. However, in some places along the Suwannee River in the White Springs area, laterally equivalent beds of this same unit consist of a lighter colored, grayish calcareous shelly sand 1–2 m thick that contains a rich fauna of marine invertebrates (Portell, 1989; Zullo and Portell, 1991).

In his description of the Porters Landing Member of the Parachucla Formation at the type locality at Porters Landing on the Savannah River in Effingham County, Georgia, Huddlestun (1988) noted that the southernmost occurrence of this unit was an outcrop containing fossiliferous shell beds on the upper Suwannee River at White Springs in northeastern Florida. Portell (1989) published a stratigraphic section of these shell beds measured about 2 km east of the White Springs 3B site. The strata Portell referred to the Parachucla Formation are about 3 m thick where his section was measured, but reach a maximum thickness of about 5 m in outcrop elsewhere in the White Springs area. Morgan (1989) expanded Huddlestun's concept of the Porters Landing Member of the Parachucla Formation in the White Springs area to include all strata exposed on the banks and in the bed of the Suwannee River that unconformably overlie the Suwannee Limestone and underlie Huddlestun's (1988) unnamed dolostone, clay, and sand of the Hawthorne Group. Geologic units that are laterally equivalent, and at least in part temporally equivalent, to the Parachucla Formation include the Penney Farms Formation in northern Florida, the Chattahoochee Formation in the Florida panhandle, and the Tampa Member (formerly the Tampa Limestone) of the Arcadia Formation in central Florida (Huddlestun, 1988; Scott, 1988).

We follow previous lithostratigraphic studies (Huddlestun, 1988; Scott, 1988, 1989) and place the fossil-bearing strata along the Suwannee River in the vicinity of White Springs in the Parachucla Formation. However, we suspect that detailed field work in this region may show that these strata should instead be placed in the Penney Farms Formation of Scott (1988), a unit of similar lithology to the Parachucla Formation. The Penney Farms Formation is primarily a subsurface unit, and has been identified from two cores in the Suwannee River area, one in Columbia County about 10 km southeast of White Springs, and one in Hamilton County about 15 km northwest of White Springs (Scott, 1988). Surface exposures of the Penney Farms Formation occur at the Martin-Anthony roadcut in Marion County (Scott, 1988), the site that produced the oreodont Phenacocoelus luskensis (MacFadden, 1980). Differences in Sr-isotope age estimates from mollusk shells led Jones et al. (1993) to question the placement of the White Springs strata in the Porters Landing Member of the Parachucla Formation. Strontium isotope age estimates derived from mollusks at White Springs and the Penney Farms Formation at Martin-Anthony are very similar (24.4 Ma and 24.6 Ma, respectively), and both are significantly older than strontium ages on mollusks (20.2 Ma) from the type locality of the Porters Landing Member in Georgia (Jones et al. 1993).

Class Mammalia Linnaeus, 1758

Order Artiodactyla Owen, 1848

Family †Merycoicocontidae Hay, 1902

Subfamily †Merycoidodontinae Hay, 1902

Mesoreodon Scott 1893

Mesoreodon floridensis, new species Figures 15.1–15.13; tables 15.1–15.3

Comparisons with Other Florida Oreodonts

The earliest oreodonts known from Florida come from the Whitneyan I-75 L.F. in Alachua County and are represented by three teeth, including a lower canine (UF 97410), p2 (UF 97411), and upper canine (UF 97409). The p2 of the I-75 oreodont is of similar size to, but transversely narrower than, Mesoreodon floridensis. The overall cusp morphology of the I-75 oreodont and M. floridensis are similar. The upper canine from I-75 is much smaller than the other specimens from I-75 and M. floridensis, and, as previously noted (Patton, 1969), may represent a second, much smaller species. Nevertheless, given the fragmentary material from I-75, not much can be said about phylogenetic affinities of these oreodonts relative to M. floridensis.

Several Arikareean oreodonts occur in Florida. Cowhouse Slough, located in Hillsborough County, contains one oreodont RP3, UF 40138. This tooth is significantly larger and, in particular, more anteroposteriorly elongated than in M. floridensis. The Brooksville sites contain several oreodont teeth. Three canines, UF 163761–163763, are of similar size, but more bladelike and compressed than in M. floridensis. A RM1, UF 18480, is of similar size to M. floridensis; the former also has a strong lingual cingulum whereas in M. floridensis the lingual cingulum is poorly developed. After White Springs, the Buda L.F. contains the second best sample of oreodont teeth and postcranials from Florida. Upper canine UF 16902 is of generally similar size and morphology to that of M. floridensis. A P3, UF 18424, is similar in morphology to M. floridensis. A P4, 16903, is generally similar in morphology but about 10% longer than M. floridensis. Two M3s, UF 16931, are about 10% larger than M. floridensis. These Buda M3s lack a lingual cingulum whereas there is a corresponding well-developed, continuous cingulum in M. floridensis. Both the Buda oreodont and M. floridensis have well-developed anterior cingula. The Buda oreodont appears to be slightly higher crowned than M. floridensis. Another site, the Martin-Anthony roadcut, produced an oreodont palate with C, P1–P3, and M1–M2 referred to Phenacocoelus luskensis (MacFadden, 1980). M. floridensis is about 20% larger than P. luskensis (including both the specimen from Florida and the referred material in the AMNH). The P1 and P2 of the P. luskensis sample are more transversely compressed and not as crowded (imbricated) as in M. floridensis.

Maglio (1966) described several fragmentary oreodont fossils from Thomas Farm. Although extremely rare, there is a small sample of the Thomas Farm oreodont specimens in the UF collection that can be compared to Mesoreodon floridensis, including an upper canine (UF 173986), an upper molar (UF 59027), and five astragali (UF 156231, 157535, 175112, 180405, 181041). The Thomas Farm oreodont fossils are much smaller than comparable elements of M. floridensis. As Maglio (1966) noted, the Thomas Farm oreodont is of similar size to Merychyus minimus, best known from late Arikareean faunas of the Great Plains. This is demonstrated by the comparative size of the astragalus (figs. 15.10, 15.11). The latest-known Florida oreodont is Ticholeptus sp. from the Barstovian Willacoochee Creek Fauna in the Florida panhandle (Bryant, 1991). Represented by a M3, UF 116823, Ticholeptus is about 10% smaller and lacks the labial cingulum seen in M. floridensis. The M3 anterior selene is of similar size, whereas the posterior selene of Florida Ticholeptus is reduced relative to M. floridensis. In the midcontinent, oreodonts decline dramatically in abundance after the Barstovian. The absence of oreodonts in Florida during the Clarendonian may result from a lack of sites of that age.

Comparison of Mesoreodon floridensis with Other Merycoidodontidae

The systematics of the Family Merycoidodontidae were reviewed by Thorpe (1937) and in a series of monographs published between 1940 and 1968 (see Schultz and Falkenbach, 1968). The latter studies were exhaustive in scope, and encompassed a tremendous number of important specimens from many relevant localities. The Schultz and Falkenbach monographs, however, are characterized by a typological approach to nomenclature, in which a plethora of species were named and in our opinion were grossly oversplit (also see Stevens and Stevens, 1996; Lander, 1998). Oreodont dentitions are relatively conservative, and consequently, oreodont systematics are based principally on skull characters. The problem with the use of skull characters, however, is that available sample sizes are usually small, and consequently population variation is poorly understood. Similarly, the effects of postmortem crushing can distort the original cranial morphologies and result in a nonbiological component of variation.

Two recent studies have attempted to unravel the complex systematics of the Merycoidodontidae. Lander (1998) presents a systematic overview of this group, but seems to include in his species concepts previously named species encompassing a broad scope of morphologies. Relative to the Florida oreodont described here, Lander's (1998) concept of Eporeodon comes closest. However, based on our examination of relevant Eporeodon specimens in the AMNH collections, particularly from the type area in the John Day basin of Oregon, and while of similar size (e.g., SKL = 208.4 mm, N = 10, vs. 206.3 mm for M. floridensis [table 15.2]), Eporeodon has numerous characters, including less retracted nasal notch, relatively deep preorbital fossa, transversely broader skull, much more inflated auditory bulla, and more elongated occipital area than the Florida oreodont.

We essentially follow Stevens and Stevens (1996) in determining the alpha-level taxonomy of the Florida oreodont. (As mentioned above, however, we refer to previous named taxa when discussing specific morphologies, e.g., “Desmatochoerus monroecreekensis,” sensu Schulz and Falkenbach [1954].) Their suite of 18 skull, mandibular, and dental characters was particularly useful in making morphological comparisons, and we followed their scheme in our study. As a first approximation, the Florida oreodont fits within the concept of ?Mesoreodon minor. During our studies of the AMNH collections, we found that ?M. minor as envisioned by Stevens and Stevens (1996) encompasses a broad range of variation, which generally falls into two morphotypes. The first of these is characterized by M. cheeki Schlaikjer and M. cheeki scotti Schultz and Falkenbach, which have a morphology generally similar to Merycoidodon (although with well-developed bullae) and are of moderate size. The second group is characterized by species such as Desmatochoerus monroecreekensis Schultz and Falkenbach and D. wyomingensis Schultz and Falkenbach. This group has distinctive characteristics such as a well-developed sagittal ridge, constricted nuchal crest, posteriorly directed occipital region, and expanded zygomatic arches. Of these two morphotypes contained within Stevens and Stevens (1996) concept of ?Mesoreodon minor, the Florida oreodont is most similar to the first group, including M. cheeki (e.g., F:AM 45430, also see Schultz and Falkenbach 1949) from the Arikareean of Goshen Hole, Wyoming and Pseudodesmatochoerus longiceps (AMNH 9732) from the late Arikareean (Harrison equivalent) of Montana. There are, however, a few exceptions to these comparisons that do not allow an unequivocal match for M. floridensis. With regard to M. cheeki, this species lacks a facial vacuity, whereas this feature appears to be developed in M. floridensis. With regard to P. longiceps, the associated postcranials (e.g., astragalus; see fig. 15.11) are about 10% larger than in M. floridensis. Although we use the generic allocation Mesoreodon and believe that the new Florida oreodont is close to the overall morphology demonstrated in the “M. cheeki complex,” M. floridensis differs significantly in the cranial and dental characters described above and hence a new species is justified.

Oreodont systematics have traditionally been based on the morphology and relative proportions of crania, and to a lesser extent, dentitions and postcranial elements. One cranial character that has been generally overlooked as being of phylogentic significance is the development of the facial vacuity. Using extant artiodactyls as an analog, the facial vacuity in oreodonts probably served to house a scent gland (Webb, 1965). When present in oreodonts, this facial vacuity is located in the preorbital cheek region on various combinations of the nasal, premaxillary, maxillary, frontal, and/or lacrimal bones. The presence or absence, relative size, and shape of these facial vacuities vary within the Merycoidodontidae. Schultz and Falkenbach (1954) considered the development of the facial vacuity to have no phylogenetic significance, whereas Stevens and Stevens (1996; personal commun., 2001) assert the contrary point of view. Thus, the discussion of the systematic utility of oreodont facial vacuities is similar to that of the preorbital facial fossae in fossil Equidae, with schools of thought for (e.g., MacFadden, 1982) and against (e.g., Forsten, 1982). As described above, Mesoreodon floridensis appears to have facial vacuities. In this regard, M. flordensis is similar to taxa such as Pseudodesmatochoerus longiceps, and different from Mesoreodon cheeki and M. chelonyx, the latter of which are central to the revised concept of this genus (Stevens and Stevens, 1996). If future studies of oreodonts indicate that the presence of a facial vacuity is outside the concept of Mesoreodon, then the species M. floridensis as currently envisioned may need to be reassigned to include it and similar forms such as Pseudodesmatochoerus longiceps in a new taxon. Resolution and further clarification of the taxonomic utility and distribution of the facial vacuity in oreodonts would require a study outside the intended scope of this paper. For the present time, we prefer to accept a broader, and taxonomically more conservative concept of Mesoreodon that includes some species that lack the facial vacuity (e.g., M. cheeki) and others in which this structure is present (e.g., M. floridensis and Pseudodesmatochoerus longiceps).

Vertebrate Biostratigraphy

There have been significant advances recently in the biostratigraphy of Arikareean vertebrate faunas. Precision in dating these faunas has improved through the use of high-resolution geochronology, specifically argon-argon (⁴⁰Ar/³⁹Ar) radioisotopic dates (e.g., Woodburne and Swisher, 1995), magnetic polarity stratigraphy (MacFadden and Hunt, 1998), detailed biochronologic data from Arikareean sites in both the Great Plains (Tedford et al., 1987, 1996) and the Gulf Coastal Plain (Albright, 1998; Hayes, 2000), and correlation with marine units, principally in the southeastern United States (Tedford and Hunter, 1984; Morgan, 1993, 1994). Tedford et al. (1987) and Woodburne and Swisher (1995) subdivided the Arikareean into four “subages” as follows: early early Arikareean (Ar1); late early Arikareean (Ar2); early late Arikareean (Ar3); and late late Arikareean (Ar4). These “subages” and their abbreviations are followed here.

Morgan (1989, 1993, 1994) reviewed the vertebrate fauna from White Springs and made preliminary biochronologic comparisons with other Arikareean faunas from Florida and the Great Plains. Albright (1998) reviewed all Arikareean vertebrate faunas from the Gulf Coastal Plain of Florida and Texas, including White Springs. Hayes (2000) made comparisons of several taxa of small mammals from White Springs with similar taxa from the Arikareean Brooksville 2 L.F. The vertebrate fauna from White Springs is here compared to the faunas from other Arikareean localities in the northern peninsular Florida. The location, age, and references for Florida Arikareean faunas are: Brooksville 2 L.F., Hernando County (Ar2; Hayes, 2000); Cowhouse Slough L.F., Hillsborough County (Ar2; Morgan, 1993; Albright, 1998; Hayes, 2000); SB-1A/Live Oak L.F., Suwannee County (Ar2; Frailey, 1978; Hayes, 2000; Frailey named the SB-1A L.F.; however, the same site previously had been collected and recorded in the UF locality files as the Live Oak L.F., and both names are used here to eliminate any confusion); Buda L.F., Alachua County (Ar3; Frailey, 1979; Albright, 1998); and Franklin Phosphate Pit Number 2, Alachua County (Ar3; Albright, 1998). The White Springs L.F. consists of 57 vertebrate species, including 14 Chondrichthyes (five rays and nine sharks); 10 Osteichthyes; six Reptilia (a land tortoise, three snakes, and two crocodilians); and 27 mammals (table 15.3). The following discussion on the biostratigraphy of the White Springs L.F. covers all age diagnostic vertebrates, in particular terrestrial mammals.

A single upper cheek tooth (UF 121427) from White Springs 1A is identified as a palaeolagine rabbit based on the presence of a crescentic valley on the upper cheek teeth, a primitive character not found in the more advanced archaeolagines (e.g., Archaeolagus). Compared to Arikareean species of Palaeolagus, the White Springs tooth is lower crowned, lacks cement, has a buccal root, and is thus identified as a small species of Megalagus. Although smaller, the White Springs tooth is morphologically similar, particularly in the well-developed hypostria and crescentic valley and presence of a buccal root, to the upper cheek teeth of the recently described species Megalagus abaconis from two Florida early Arikareean sites, Brooksville 2 and Cowhouse Slough (Hayes, 2000). Both Megalagus and Palaeolagus became extinct at the end of the early Arikareean (Dawson, 1958).

Seven sciurid teeth representing two genera occur in White Springs 3B. Three teeth represent the large sciurid Protosciurus. Frailey (1978) described and illustrated a Protosciurus mandible with p4 from the SB-1A/Live Oak L.F. Lower p4s of the White Springs and SB-1A/Live Oak Protosciurus are similar in size and morphology, but neither has been identified to species. Korth (1992) reported an undescribed species of Protosciurus from the Ar2 McCann Canyon L.F. in Nebraska. Protosciurus is restricted to Chadronian through early Arikareean faunas. Four teeth of a tiny sciurid are referred to Nototamias. The type species of Nototamias, N. hulberti, is from the early Hemingfordian Thomas Farm L.F. in northern Florida (Pratt and Morgan, 1989). Korth (1992) described the species Nototamias quadratus from the McCann Canyon L.F. in Nebraska, and also mentioned the occurrence of N. quadratus in the Ar2 Monroe Creek Fauna.

Morgan (1989) referred three species of geomyoid rodents from White Springs to the Heteromyidae, including small, medium, and large species. More detailed study of this sample (A. Pratt, personal. commun.) has shown that these three species may represent two families, including Heteromyidae and Heliscomyidae. The Heliscomyidae is represented at White Springs by a small species referred to the genus Heliscomys. The White Springs Heliscomys is similar in size and morphology to species of Heliscomys from the Orellan (early Oligocene) of the western United States. The range of Heliscomys is from the Chadronian (late Eocene) through the early Arikareean (late Oligocene), with the possible exception of several isolated teeth from the Hemingfordian of Saskatchewan (Korth et al., 1991). Two other rodent species previously referred to the Heteromyidae (Morgan, 1989) occur at White Springs, neither of which has been identified to genus. The medium-sized species is similar to a species from SB-1A/Live Oak and the largest is similar to a species from Buda.

Five teeth of a large eomyid rodent, three from White Springs 1A and two from White Springs 3B, are referred to the genus Arikareeomys, which is known outside of Florida only by the type species A. skinneri from the Arikareean McCann Canyon L.F. in Nebraska (Korth, 1992). Arikareeomys is a large eomyid with high crowned teeth in which both the upper and lower cheek teeth are characterized by two enamel lakes that are separated by a central, transverse valley. The White Springs sample is clearly referable to Arikareeomys, as no other similar large eomyid is known from North America. However, the Florida teeth differ in minor details from A. skinneri from Nebraska and appear to represent an undescribed species. Arikareeomys occurs in four Florida Arikareean faunas, including White Springs, Cowhouse Slough, SB-1A/Live Oak, and Buda. White Springs, Buda, and SB-1A/Live Oak all have the same Arikareeomys, whereas the Cowhouse Slough taxon is a distinct species.

A large, brachydont “cricetid” rodent of the genus Leidymys is represented at White Springs 3B by more than 10 isolated molars. Leidymys generally is referred to the subfamily Eucricetodontinae, Family Cricetidae (e.g., Martin, 1980; Korth, 1992). The M1 of the White Springs Leidymys is similar to L. cerasus from the McCann Canyon L.F. in Nebraska (Korth, 1992), although it differs from the latter species in slightly smaller size and several dental details, and almost certainly represents an undescribed species. Leidymys is typical of Whitneyan and early Arikareean faunas (Martin, 1980). L. cerasus from McCann Canyon is the youngest known species in the genus (Korth, 1992).

Compared to oreodonts and camels, equids and rhinoceroses are very rare in the White Springs L.F. Morgan (1989) reported several partial teeth of a primitive Parahippus-grade horse from White Springs 1A. Albright (1998) later referred these teeth to Anchippus texanus, a species also known from two other Arikareean sites on the Gulf Coastal Plain, the Franklin Phosphate Pit Number 2 site in Florida, and the Ar3 Toledo Bend L.F. from easternmost Texas. A partial upper tooth (UF 206417) from White Springs 3B is tentatively referred to the Arikareean rhinoceros Diceratherium.

There are three species of camelids in the White Springs L.F., all of which were found only at White Springs 3B. A partial articulated hind limb (UF 205725), seems referable to the large, long-limbed camelid Oxydactylus based on its large size compared to other Arikareean camelids, elongation of the tibia (total length 460 mm), and unfused metatarsals. An intermediate-sized camelid is represented by two associated lower premolars (UF 205724), associated distal hind limb (UF 205728), and isolated astragalus (UF 205727). This species is similar to a camelid tentatively referred to Gentilicamelus from the Buda L.F. (Frailey, 1979). A small calcaneum (UF 205729) is similar in size and characters to those of the tiny camelid Nothokemas waldropi from SB-1A/Live Oak (Frailey, 1978).

One of the most important mammals bearing on the age of White Springs is the newly described oreodont Mesoreodon floridensis. This species is endemic to Florida, but the genus Mesoreodon is otherwise restricted to late Whitneyan and early Arikareean (Ar1 and Ar2) faunas in the western United States (Stevens and Stevens, 1996). Mesoreodon is more typical of the early Arikareean, including faunas from the Gering Formation in Wyoming and Nebraska, the Cabbage Patch Formation in Montana, and the John Day Formation in Oregon. Of the two western species referred to Mesoreodon by Stevens and Stevens (1996), the early Arikareean M. minor is most similar to M. floridensis.

Three genera of sirenians, Crenatosiren, Dioplotherium, and Metaxytherium, co-occur in the White Springs L.F. (Domning 1988, 1989a, 1989b, 1991, 1997). Crenatosiren olseni, originally described from White Springs 3A (Reinhart, 1976; Domning, 1997) has since been reported from the Ashley and Chandler Bridge formations in South Carolina, both of which are late Oligocene (Chattian) in age (Domning, 1997). Domning (1989a) reported a nearly complete skull of the sirenian Dioplotherium manigaulti from White Springs 1A, a species that also occurs in the Oligocene Ashley and Chandler Bridge formations in South Carolina. The third sirenian from White Springs, Metaxytherium, has the longest stratigraphic range of the three genera, occurring in the New World from the late Oligocene through the late Miocene (Domning, 1988).

Morgan (1989) referred two teeth of the large shark Carcharodon from the Parachucla Formation near White Springs to the Oligocene and early Miocene species C. angustidens. The White Springs record of C. angustidens is one of the youngest occurrences of this species, and indicates that the fauna is no younger than early Miocene. The most common shark in the White Springs L.F., a small undescribed species of Carcharinus, is known elsewhere only from the Oligocene (Whitneyan) I-75 L.F. in Gainesville, Alachua County, Florida (Tessman, 1969).

In summary, the biochronology of the White Springs L.F. clearly establishes an Arikareean age for this assemblage. Certain of the genera and species allow for a more precise placement within the Arikareean, which is important because the Arikareean is about 10 million years long (29–19 Ma; Tedford et al., 1987; MacFadden and Hunt, 1998). Almost all of the age-diagnostic mammals from White Springs indicate a “medial” Arikareean age, either late early Arikareean (Ar2; 28–24 Ma) or early late Arikareean (Ar3; 24–22 Ma). This time range is significant because it includes the Oligocene-Miocene boundary at 23.8 Ma (Berggren et al., 1995) and is not well represented in the midcontinental sequence.

Mammalian taxa from White Springs that indicate an early Arikareean age are: Mesoreodon floridensis; the lagomorph Megalagus sp.; the rodents Arikareeomys new species, Heliscomys sp., Leidymys new species, and Protosciurus sp.; and the sirenian Crenatosiren olseni. Among these genera, Mesoreodon, Megalagus, Arikareeomys, Leidymys, and Protosciurus do not occur after the early Arikareean, and Crenatosiren is found elsewhere only in late Oligocene (Chattian) marine faunas. With the exception of a Hemingfordian record from Canada (Korth et al., 1991), Heliscomys is otherwise restricted to early Arikareean and older sites, and the White Springs Heliscomys resembles Oligocene (Orellan) species of the genus. Korth (1992) reported three of these genera, Arikareeomys, Leidymys, and Protosciurus, from the McCann Canyon L.F. of Nebraska, which he regarded as Ar3 in age. However, Korth noted the similarity of certain of the small mammals from McCann Canyon to those from the somewhat older Ar2 Monroe Creek Fauna of Nebraska and South Dakota. We follow Tedford et al. (1996), who considered the McCann Canyon L.F. to be correlative with the Monroe Creek Fauna, and thus early Arikareean, rather than late Arikareean.

First appearances of either immigrant or autochthonous taxa are generally considered to be more important in establishing the age of faunas than are last appearances. Therefore, it would be significant from a biostratigraphic standpoint if the White Springs L.F. contained genera or species that had their first appearance in the late Arikareean. Only two genera of mammals from White Springs appear to be late Arikareean first occurrences (Tedford et al., 1987; Albright, 1998), the horse Anchippus and the camel Oxydactylus. However, the presence of these two genera at White Springs must be considered tentative, as Anchippus is represented by two partial teeth and Oxydactylus is known only from postcranial material. The remainder of the White Springs mammalian fauna includes six genera (Arikareeomys, Crenatosiren, Leidymys, Megalagus, Mesoreodon, and Protosciurus) that are unknown after the early Arikareean, and a seventh genus (Heliscomys) that is most typical of early Arikareean and older faunas but may persist into somewhat younger faunas. In summary, the weight of the biochronologic evidence indicates that the White Springs Local Fauna is late Ar2 in age, probably between 25 and 24 Ma (latest Oligocene, Chattian equivalent), given evidence cited below. Correlative faunas from the Great Plains are the Monroe Creek Fauna and the McCann Canyon L.F.

Invertebrate Biostratigraphy

Among the 65 taxa of marine invertebrates from the Porters Landing Member of the Parachucla Formation in the vicinity of White Springs, Portell (1989) identified several biostratigraphically diagnostic species of mollusks, including Chlamys acanikos and Ostrea normalis. Both of these bivalves have been identified from localities in Florida and Georgia that are considered to be early Miocene in age, including the Penney Farms Formation in the Devils Millhopper and at the Martin-Anthony roadcut in northern Florida, the lower part of the Torreya Formation in the Florida Panhandle, and the type locality of the Porters Landing Member in Georgia (Portell, 1989). Zullo and Portell (1991) reported three species of barnacles from the Porters Landing Member at White Springs. They also identified these same three barnacle taxa from the laterally equivalent Penney Farms Formation from a core in Columbia County, Florida, about 18 km east of White Springs. One of these barnacles, referred to the genus Solidobalanus, belongs to a complex of species that ranges from the middle Eocene through the early Oligocene on the eastern Gulf Coastal Plain. They referred another species to Concavus crassostricola, a barnacle known elsewhere only from the early Miocene (Aquitanian) of North Carolina.

Cores from the Penney Farms Formation in Nassau County, Florida, about 80 km northeast of White Springs (Scott, 1988), and the Porters Landing Member of the Parachucla Formation in southeastern Georgia, about 300 km northeast of White Springs (Huddlestun, 1988), both contain planktonic foraminifera. Comparison with subtropical planktonic foraminifera results in a correlation of these two units to either upper Zone N4 (upper Globorotalia kugleri Zone) or lower Zone N5 (lower Catapsydrax dissimilis Zone). This correlation suggests an early Miocene (late Aquitanian) age of about 22–21 Ma based on the time scale of Berggren et al. (1995). This is considerably younger than the age for White Springs L.F. suggested by the mammalian biostratigraphy. The correlation of the foraminifera-bearing strata from Penney Farms Formation in Nassau County and the type locality of the Parachucla Formation in Georgia is not questioned. However, the correlation of these units with the Parachucla Formation at White Springs, which has not produced diagnostic planktonic foraminifera, is open to question, particularly considering the age discrepancy indicated by both the mammalian fauna and the strontium isotope analysis below.

Strontium Isotope Geochronology

Jones et al. (1993) used strontium isotopes (⁸⁷Sr/⁸⁶Sr) to determine the age of macroinvertebrate fossils from the type locality of the Porters Landing Member of the Parachucla Formation in Georgia and from supposedly correlative strata exposed at White Springs, which are directly correlative with the strata that produced the White Springs L.F. Strontium isotopic analyses of two shells of the oyster Ostrea normalis from White Springs give an age estimate of 24.4 Ma (latest Oligocene, Chattian). The strontium isotopic age calculated for a specimen of Ostrea normalis from Porters Landing, Georgia yielded a significantly younger date of 20.2 Ma (early Miocene, Burdigalian). The errors associated with these strontium isotopic age estimates, which range from ±0.5 m.y. to ±1.0 m.y., are not nearly large enough to account for the 4+ Ma difference in age (Jones et al., 1993) between correlative units in Georgia and White Springs. Strontium isotopic ages on the bivalves Ostrea normalis and Chlamys acanikos from the Penney Farms Formation at the Martin-Anthony roadcut in Marion County yielded an age estimate of 24.6 Ma, very similar to the age of the Parachucla Formation at White Springs. The mollusks dated from the Martin-Anthony roadcut were collected from the same marine limestone (unit 9) that produced a partial skull of the oreodont Phenacocoelus luskensis (see MacFadden, 1980).

Paleoecology

Of the 57 vertebrate species in the White Springs L.F., 27 species are marine (five rays, nine sharks, nine bony fish, one crocodilian, and three sirenians), two species are freshwater (one gar and one alligator), and 28 species are terrestrial (a land tortoise, three snakes, and 24 mammals). The terrestrial members of the vertebrate fauna are concentrated at two sites, White Springs 1A and White Springs 3B. The fauna from the remainder of the Parachucla Formation along the Suwannee River between these two sites consists primarily of marine taxa, including both vertebrates and an extensive invertebrate fauna (Portell, 1989).

The vertebrate fauna from White Springs 1A is predominantly marine, with the most abundant fossils consisting of isolated teeth of sharks, dasyatid rays, and bony fish. The most complete mammals from this site are sirenians, including a skull and partial articulated skeleton of Metaxytherium and a complete skull of Dioplotherium. The terrestrial component of White Springs 1A mostly consists of isolated teeth and postcranial elements of small mammals, often heavily waterworn. Several partial teeth of the horse Anchippus are the only large land mammals from this site. The marine fauna from White Springs 1A consists of nearshore taxa. Deposition probably occurred in a marginal marine habitat such as delta or coastal lagoon, with the terrestrial taxa transported into a shallow marine depositional environment by a coastal river.

The vertebrate fauna from White Springs 3B includes a larger sample of terrestrial mammals than does White Springs 1A. Furthermore, the most completely and best preserved specimens are from the terrestrial component of White Springs 3B, including partial articulated skeletons of the oreodont Mesoreodon floridensis and a large camelid. The presence of partial skeletons of land mammals at White Springs 3B suggests that the carcasses were not transported far and were buried rather quickly. This site also contains several partial jaws and maxillae and numerous well-preserved isolated teeth of small mammals, further indicating quiet-water depositional conditions and limited transport. Several of the small mammals from White Springs 3B, including two species of squirrels, suggest forested habitat. White Springs 3B also has a diverse marine vertebrate fauna of sharks, rays, and bony fish, similar to the fauna from White Springs 1A. Most of the marine vertebrate fossils also are well preserved and indicate limited transport and waterwear. The predominantly marine nature of the Parachucla Formation and the difficulty in explaining how a diverse marine fauna could be transported and deposited inland under freshwater conditions indicates a marine depositional environment for White Springs 3B as well. A quiet coastal lagoon surrounded by deciduous forests with small freshwater streams emptying into it is a possible depositional environment that could explain the diverse and well-preserved faunas of both terrestrial mammals and marine vertebrates.

Mounted Skeleton and Reconstruction of Mesoreodon floridensis

During preparations for the opening of the “Hall of Florida Fossils” at the Florida Museum of Natural History, a mounted skeleton of Mesoreodon floridensis was articulated for exhibition (fig. 15.12). The skull, jaws, some vertebrae, and limbs consist of a composite of White Spring specimens (see “Referred Material”) and most of the axial skeleton consists of vertebrae from Merycoidodon culbertsoni from western Nebraska, or reconstructed ribs. Merycoidodon culbertsoni is a close relative of M. floridensis, similar in morphology and overall size, and was therefore appropriate to fill in the missing skeletal elements.

Mesoreodon floridensis is a medium-sized mammal, with a head-body length of 1070 mm and a mean shoulder height (base of foot [distal-most tip of ungual phalanx], with flexed limb [not extended], to top of scapula) of 475 mm (R = 470 mm, L = 480 mm). Like other oreodonts, the cranium is robust and large relative to overall body size. In the front limb the humerus, radius, and ulna of M. floridensis are all about the same length, but more robust than, Merycoidodon. The mean length of the humerus is 170 mm (R = 170 mm, L = 170 mm) and the mean length of the radius is 125 mm (R = 120 mm, L = 130 mm). In the hind limb, the femur of M. floridensis is of similar length as, but more robust than, Merycoidodon. The tibia of M. floridensis is of similar proportion to Merycoidodon. The mean length of the femur of M. floridensis is 184 mm (R = 187 mm, L = 180 mm) and the mean length of the tibia is 166 mm (R = 167 mm, L = 165 mm). The mean ratio of the forelimb (humerus and radius measured as a unit) compared to that of the hindlimb (femur and tibia) is 0.82. In the fore- and hind foot of M. floridensis, the carpals and tarsals are more elongated, whereas the metacarpals and metatarsals are generally more slender and slightly shorter than in Merycoidodon. The carpus of Mesoreodon floridensis contains a trapezoid that is triangular and positioned proximal to both MC I and II, whereas in Merycoidodon this bone is rectangular and positioned directly proximal to MC II. The metatarsals (MT II, III, and IV) of M. floridensis are more slender and slightly shorter than those of Merycoidodon. The lifelike reconstruction of M. floridensis (fig. 15.13) is generally similar to that of Merycoidodon.

Given the associated fauna and interpreted paleoecology of the White Springs sites, the local habitat of Mesoreodon floridensis probably was a scrub or woodland located near an estuary or lagoon, very close to a full marine environment. The dental morphology of M. floridensis, being relatively short-crowned, indicates a principally browsing diet. Given these parameters, little or no apparent sexual dimorphism, and an estimated body size of 25 kg, M. floridensis probably lived in mixed-sex feeding herds (Janis, 1982).