A New Machairodont from the Palmetto Fauna (Early Pliocene) of Florida, with Comments on the Origin of the Smilodontini (Mammalia, Carnivora, Felidae)

South-central Florida’s latest Hemphillian Palmetto Fauna includes two machairodontine felids, the lion-sized Machairodus coloradensis and a smaller, jaguar-sized species, initially referred to Megantereon hesperus based on a single, relatively incomplete mandible. This made the latter the oldest record of Megantereon, suggesting a New World origin of the genus. Subsequent workers variously accepted or rejected this identification and biogeographic scenario. Fortunately, new material, which preserves previously unknown characters, is now known for the smaller taxon. The most parsimonious results of a phylogenetic analysis using 37 cranio-mandibular characters from 13 taxa place it in the Smilodontini, like the original study; however, as the sister-taxon to Megantereon and Smilodon. Accordingly, we formally describe Rhizosmilodon fiteae gen. et sp. nov. Rhizosmilodon, Megantereon, and Smilodon ( =  Smilodontini) share synapomorphies relative to their sister-taxon Machairodontini: serrations smaller and restricted to canines; offset of P3 with P4 and p4 with m1; complete verticalization of mandibular symphysis; m1 shortened and robust with widest point anterior to notch; and extreme posterior “lean” to p3/p4. Rhizosmilodon has small anterior and posterior accessory cusps on p4, a relatively large lower canine, and small, non-procumbent lower incisors; all more primitive states than in Megantereon and Smilodon. The former also differs from Megantereon and Smilodon gracilis by having a very small mandibular flange. Rhizosmilodon is the oldest known member of the Smilodontini, suggesting that the tribe originated in North America. Two more derived, similar-sized species evolved in parallel during the Blancan, Megantereon hesperus and Smilodon gracilis. The former is rarer, known only from the north-central and northwestern US, and presumably dispersed into the Old World. The latter is known from the eastern and southern US, and dispersed into South America.


Introduction
Most vertebrate fossils from the Upper Bone Valley Formation of Central Florida are recovered as isolated finds, either in situ from exposures created by mining operations or out of stratigraphic context in spoil piles [1][2][3][4][5][6]. Rarely are they found in sufficient concentration to allow quarrying. However, the largest known such concentration was discovered in the spring of 1989 in the Fort Meade Mine of Gardinier Inc. Thousands of vertebrate fossils were found in a single bed about 0.8 m thick that covered an area of about 2,000 m 2 . This assemblage, called the Whidden Creek Local Fauna, includes 33 mammalian taxa, of which 11 are carnivorans (Table 1). However, carnivorans account for only about 5% of the ca. 900 identifiable mammalian fossils from the Whidden Creek LF, in contrast to proportionally greater representation by perissodactyls (,33%), artiodactyls (,37%), and cetaceans (,16%). The composite vertebrate fossil assemblage from the Upper Bone Valley Formation is referred to as the Palmetto Fauna [3,6].
Among the carnivorans recovered from the Palmetto Fauna are the remains of a jaguar-sized machairodont cat [6]. Berta and Galiano [7] originally identified this felid as Megantereon hesperus based on a single partial ramus containing p3, p4 and the anterior portion of m1, making it the earliest record of the widespread genus Megantereon, which suggests a New World origin for this machairodont. While their identification as Megantereon was followed by some workers [8,9], it was questioned by others due to the fragmentary nature of the material [6,10,11,12], and more recent work even suggested that it exhibited close affinities to a specimen identified as ''Paramachairodus'' sp. from Arizona [13]. Just as important, is the Palmetto machairodont's relationship to other members of the tribe Smilodontini, including Smilodon. Primarily because though Smilodon is an iconic member of the Pliocene and Pleistocene American faunas, its highly derived state has made interpreting its origin, and relationship to other closely related taxa, problematic [7,[9][10][11][12][14][15][16][17][18][19].
More specifically, though Megantereon is well known from the Old World, its origins are uncertain [9,11]. Several authors have suggested at least a sister relationship between Megantereon and Smilodon [7,17,18]; with some further suggesting that the former is ancestral to the latter [11,15,17]. In addition, there are suggestions that Paramachaerodus is a basal member of the Smilodontini [11,13,19], and that it gave rise to Megantereon [7,10,13], which in turn gave rise to Smilodon [11,13,15,17]. Still, others have suggested an African origin for Megantereon [10] with a subsequent migration into the New World, thereby complicating the resolution of the relationships. However, if the Palmetto machairodont is indeed Megantereon, then a New World origin for at least Megantereon [9] would be supported. It is also possible that both Megantereon and Smilodon originated in North America [13]. Following several others [10,12] who disagreed with the identification of the Palmetto machairodont and the suggestion that Megantereon dispersed from the Old World into the New, we sought to resolve some of these questions by reevaluating the former. Consequently, the recovery of additional specimens of the Palmetto machairodont (Figure 1), which provide characters not previously observed, affords the opportunity to revisit the systematic position of this taxon and to address some of the issues within the tribe.

Methods
Basic taxonomy follows that of [11,18,19]. The acronym 'UF' refers to specimens cataloged into the vertebrate paleontology collection of the Florida Museum of Natural History, University of Florida, Gainesville.
Cladistical analyses were run using PAUP 4.0 Beta 10 and MacClade. The most parsimonious (shortest) trees were evaluated using a heuristic search based on combined branch length.
Nomenclatural Acts -The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers)  can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http:// zoobank.org/". The LSID for this publication is: urn:lsid:zoobank.org:pub: urn:lsid:zoobank.org:pub:41E89F59-8327-4282-8BC3-E3A195C436D2. The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central, LOCKSS.

Results and Discussion
Phylogeny -Scores are listed in Table 2. Results ( Figure 2) agree with the revision of the late Miocene felid Paramachaerodus [20], with Promegantereon falling out as a distinct genus (from the former), and both falling basal to all other machairodonts. Of the trees evaluated, 2 most parsimonious trees were retained ( Figure 2, Appendix S2). The first most parsimonious tree used in this study had: TL = 89, CI = 0.63, RI = 0.74, and RC = 0.47. A second most parsimonious tree is provided in Appendix S2. Holotype. UF 124634, partial right ramus including mandibular symphysis, c1, m1, and alveoli for p3 and p4 ( Figure 1D-F).
Type Locality. Diagnosis. Nearly complete verticalization of mandibular symphysis; presence of a weak, but clear mandibular flange; suggestion of crenulations on at least lower canines; lower canine remains large but is moderately compressed laterally; nonprocumbent incisor arcade; incisors small; P3 and P4 not aligned (implied by offset of p4 and m1); lack of p2; p3 elongate, but smaller than p4 (,2/3 length); posterior lean of p3 and p4 towards m1; lack of anterior accessory cusp on p3; presence of a small posterior accessory cusp on p3; anterior accessory cusp on p4 small; p3 and p4 off-set (not in straight line); p4 long and bladelike, yet with distal widening around posterior accessory cusp; p4 and m1 also off-set; m1 talonid variably present as small accessory cuspid to slight raised bump at the base of the tooth, never fully formed metaconid; m1 robust as in Smilodon with widest point at or anterior to the carnassial notch; and m1 paraconid remains shorter than protoconid.
Body mass estimates. Following Christiansen and Harris [21], anterior-posterior diameters (APD) for the referred humerus (UF 133938) and tibia (UF 133939) of 26.3 mm and 18.6 mm, generated body mass estimates of between 76.6-85.0 kg. and 55.7-58.3 kg. respectively (Figure 3). These estimates are similar to observed body masses in extant medium-sized felids such as Puma concolor and Panthera onca, and also overlap the lower half of the estimated range of Smilodon gracilis.
Humeri. Though not directly associated with any of the cranial material, the two isolated humeri are here referred to the R. fiteae for the following reasons: presence of medial projection of the epicondylar region and straight pectoral ridge (as opposed to strong to sharp distal curve prior to joining the deltoid ridge typical of most pantherine cats, [14]), suggests machairodont affinities. Moreover, the strong curve to the shaft and overall robustness; exaggerated medial projection of the epicondylar region well beyond the entepicondylar foramen, typical of members of Smilodon [14,22]; and thin and elongate wall of bone closing the entepicondylar foramen [14] supports its inclusion within the Smilodontini. Consequently, either there is yet another unknown smilodontine machairodont cat in the Palmetto Fauna, or these humeri do indeed represent the taxon described here. As all medium-sized felid dental remains in the Palmetto Fauna appear to represent a single taxon, R. fiteae, all felid post-cranial elements whose size corresponds to these teeth and jaws are also referred to R. fiteae.
Tree. Notable characters grouping Machairodontini: exaggerated serrations on all teeth; straightening and verticalization of the mandibular symphysis; alignment of P3 with P4 and p4 with m1; elongate and trench-like m1; elongate paraconid (approximately equal to protoconid); slight posterior ''lean'' to p3/p4. For Smilodontini, including R. fiteae: serrations restricted to canines and not as large as in Machairodontini; offset of P3 with P4 and p4 with m1; complete verticalization of mandibular symphysis; m1 shortened and robust with widest point anterior to notch; extreme posterior ''lean'' to p3/p4. In sum, R. fiteae lies basal within Smilodontini (in accord with its older geologic age) and is distinct from taxa within both Megantereon and Smilodon. Like other members of the Smilodontini: serrations restricted to canines and not as large as in Machairodontini; offset of P3 with P4 implied by that of p4 with m1; Less derived features not typical of more advanced machairodonts, yet present in R. fiteae: small  anterior and posterior accessory cusps on p4; retention of a large lower canine; and small, non-procumbent incisors. Rhizosmilodon fiteae differs from other members of Smilodontini in polarity depending on the feature. Specifically: Canine serrations appear to have been very minor (if even present) in R. fiteae (specimen worn) and S. gracilis. Clear serrations are exhibited by S. populator and S. fatalis, but apparently are lost in species of Megantereon. It should be noted that S. gracilis exhibits perhaps the beginning of serrations with upper canines characterized by extreme flattening and elongation, with thin enamel ridge running along the anterior and posterior margins of the tooth. Light crenulations (alternating thick and thin areas) are visible along these ridges on several nearly unworn upper canines of S. gracilis (e.g., UF 84189, 86843), -perhaps the precursor to true serrations. Moreover, the recently reported middle Pleistocene record of Smilodon from Venezuela [19] was described as possessing fine serrations on the upper canine, yet also exhibiting morphology otherwise more consistent with S. gracilis, as opposed to S. fatalis. Should this record be correct, then it supports the possibility of serrations or crenulations emerging within Smilodontini in, or close to, R. fiteae. Earlier suggestions of Megantereon as ancestral to Smilodon [11] suffered with the issue of the former lacking serrations, yet some earlier machairodonts exhibiting them (hence representing the acquisition, loss, then re-acquisition of serrations). Evenly spaced valleys and associated ridges, which are nearly equally spaced, are visible on the lower canine of the R. fiteae; however wear obscures definitive confirmation of true crenulations. If R. fiteae did indeed exhibit very minor serrations or crenulations, the loss of such features in species of Megatereon, and contemporaneous development of them in species of Smilodon, is not as problematic.
The large size of the lower canine and only moderate development of the mandibular flange suggest a ''saber-tooth'' configuration more pronounced than species within Promegantereon or Paramachaerodus; similar to species within Machairodus, where the canine is flattened and elongate; but not as extreme as the various species of Smilodon or Megantereon. Moreover, R. fiteae is primitive in its retention of a fairly large lower canine (incisiform in nearly all later machairodonts). However, the lower canine is noticeably laterally compressed. Because enlarged and laterally compressed upper canines, with reduced but also compressed lower canines, are typical machairodont traits [23,24], taken as a whole the canine configuration exhibited by the R. fiteae can be interpreted as a truly intermediate state.
Typical of advanced machairodonts is the protrusion of the incisors into a strong arch [23][24][25]. Limited space between the outer edge of the jaw and the suture at the mandibular symphysis strongly suggests the retention of very small incisors (primitive among machairodonts) in R. fiteae. Species of Megentereon and S. gracilis, exhibit enlarged incisors with the minor development of an arch, which would be intermediate between those of R. fiteae and later more advanced machairodonts (e.g. S. fatalis, S. populator or Homotherium).
Many of the major differences between R. fiteae and other members of the tribe Smilodontini are exhibited by the lower premolars. Specifically, the distal cusp on the p3 is well developed in R. fiteae. On S. gracilis it is typically present, but is reduced and in some specimens occurs as a series of tiny accessory cusps. Species of Megantereon, typically exhibit the cusp, but also in a greatly reduced state. In S. fatalis and S. populator, the entire tooth is typically lost. On the p4 of R. fitaea, the anterior and distal accessory cusps remain low and small. These are enlarged (proportionally more similar in size to the primary central cusp) in later, more derived machairodonts. In addition, the p4 of R. fiteae and some Megantereon spp. is somewhat long and blade-like with slight widening at the posterior end of the tooth. In species of Smilodon (particularly the later forms), the anterior portion widens as well (also exhibits a larger and more prominent anterior accessory cusp) resulting in an overall more robust tooth.
The lower first molar of R. fiteae is also somewhat intermediate in exhibiting a slight hint of a talonid (metaconid). More primitive machairodonts, such as Promegantereon ogygia or Machairodus aphanistus, typically retain a well-developed metaconid, while the more advanced forms exhibit no evidence of it. In addition, R. fiteae retains a simple m1, whereas some advanced members of the genus Smilodon (S. fatalis and S. populator), develop anterior accessory cusps on the m1 (, parastylids?).
Former identifications. Though several authors have suggested that the original identification of the Palmetto Fauna machairodont as Megantereon hesperus [7] was likely incorrect, few offered (supported) alternative identifications. However, Turner [10] expressed two possibilities based on the material known at that time. First was that the taxon could represent a species of the feline-like machairodont Dinofelis, similar to that observed in South Africa from similar-aged deposits. Presence of the mandibular flange, flattening of the canines, and potential presence of serrations, in addition to lack of characters typical of the Dinofelis [26], now eliminates that taxon. Alternatively, the Palmetto Fauna machairodont could represent Paramachaerodus [10], an early machairodont known to exhibit less derived features. Adding to this, Hodnett [13] described new felid material from Arizona (White Cone specimen) as ''Paramachairodus'' sp. and noted its similarity to the Palmetto Fauna machairodont. However, he described the White Cone specimen as gracile and laterally compressed, whereas the mandibles of R. fiteae are stout and robust for their length. In addition, the thin ridge of bone on the dorsal surface of the ramus between p3 and c1 described on the White Cone specimen [13] is more strongly developed in R. fiteae, but is known to be highly variable among machairodont taxa, so likely carries little taxonomic value [10]. Moreover, the diastema itself is significantly longer and ventrally directed in R. fiteae. The White Cone specimen is also described as having a rounded anterior margin [13], whereas R. fiteae has a moderately developed, yet clear mandibular flange. A small anterior cuspid described on the p3 of the White Cone specimen is lacking in R. fiteae, and the p3 of the former is proportionately larger, relative to the p4, than in the latter. Lastly, the p4 of the White Cone specimen is proportionately smaller than m1 relative to R. fiteae. Taken together, the characters describing the White Cone specimen highlight its distinctiveness from R. fiteae.
Biogeographic implications. The first definitive Megantereon in North America occurs in the Blancan [10,27]. Consequently, our results support two possible evolutionary scenarios: First, the genus Megantereon originated in the New World [7,8], at about the same time (early Blancan) as Smilodon, followed by dispersal to, and diversification in, the Old World. Smilodon then spreads and diversifies in the New World through the Pliocene and Pleistocene. Morphologic similarities of both genera (particularly between S. gracilis and M. hesperus) imply similar lifestyles and paleoecology (see Appendix S3 for additional comments). However, geographic separation allowed each genus to thrive and diversify, while avoiding competition with the other. Moreover, the limited NA records of the genus Megantereon likely represent last holdovers before it was out-competed by members of the genus Smilodon.
Alternately, R. fiteae and the species within Smilodon do have some features in common to the exclusion of species of Megantereon (e.g., the reduced development of the mandibular flange, and persistence of a minor talonid on m1). Consequently, additional characters may amplify the distinction of Megantereon from the remainder of the tribe, suggesting a common ancestor of Megantereon and its sister taxon: Smilodon + R. fiteae. If true, such a relationship would then support an Old World origin of Megantereon and the tribe Smilodontini [10,11] and two dispersals of this tribe into NA, once in the Hemphillian, and then again in the Blancan.

Conclusion
The smaller Palmetto Fauna machairodont does not represent Megantereon hesperus; however, it is basal within the tribe Smilodontini, which also includes the genera Megantereon and Smilodon. Moreover, the unique combination of both conservative and derived characters warrants erection of a new genus and species, Rhizosmilodon fiteae. This very late Hemphillian record is the oldest for the tribe, thereby supporting a North American origin at least 5 million years ago (latest Miocene). More specifically, our results also support suggestions of a North American origin of Smilodon [16] from a common ancestor with Megantereon; also affirming their relationship as sister taxa [7,15,17,18,28]. Our results refute the suggestion that the Palmetto Fauna machairodont is congeneric with ''Paramachairodus'' (now Promegantereon [20]) from Spain [13]. Both Promegantereon and Paramachaerodus exhibit less derived features than R. fiteae and consequently fall distinct from, and more basal to, the Smilodontini. Lastly, R. fiteae differs significantly from the White Cone (Arizona) specimen identified as ''Paramachairodus'' sp. [13], consequently the two taxa are not conspecific.