Early Miocene land mammals from eastern North America are exceedingly rare. Over the past several decades a small, but significant, vertebrate fauna has been recovered by paleontologists and citizen scientists from the Belgrade Formation at the Martin Marietta Belgrade Quarry in eastern North Carolina. This assemblage has 12 land mammal taxa, including beaver (Castoridae), stem lagomorph, carnivorans (Mustelidae, Ailuridae), horses (Equidae), rhinoceros (Rhinocerotidae), tapir (Tapiridae), peccary (Tayassuidae), anthracothere (Anthracotheriidae), entelodont (Entelodontidae), and protoceratid (Protoceratidae). Taken together, the biochronology of this Maysville Local Fauna indicates a late Arikareean (Ar3/Ar4) to early Hemingfordian (He1) North American Land Mammal Age (NALMA). This interval, which includes the Runningwater Chronofauna, documents numerous important Holarctic immigrants, including Amphictis , Craterogale , and cf. Menoceras found at this locality. Strontium isotope stratigraphy (SIS) of shark teeth collected in situ from the Belgrade Formation yield an age of 21.4 ± 0.13 Ma, which validates the age of interbedded land mammals within this unit. It also is consistent with the late Arikareean (Ar3/Ar4) biochronology and Aquitanian Neogene marine stage. New SIS analyses of oysters ( Striostrea gigantissima ) and clams ( Chione ) from this mine, previously assigned to late Oligocene or Late Miocene, are significantly older (28.0 ± 0.22 Ma and 27.6 ± 0.26 Ma, respectively) than the land mammals. Depending upon stratigraphic interpretations, these may confirm an older marine facies within the Belgrade Formation. This locality is important because of its marine and terrestrial tie-ins that facilitate intercalibration of both NALMAs and Cenozoic marine stages.

We describe a new genus and species of mellivorine mustelid from the late early Hemphillian (Hh2, c. 7.5 Ma) Box T Local Fauna of Texas, USA. Our phylogenetic analysis indicates that it is a basal eomellivorin. Moralesictis intrepidus is intermediate in size between Eomellivora and Mellivora capensis, the smaller extant honey badger. It is characterised by its strong muzzle, relatively convex skull profile, robust and unicuspid P1-P3, and a distinctive P4. The P4 has a mesially located robust and enlarged protocone, and is shorter in relative to the M1 width. The M1 occlusal morphology is primitive for a mellivorine. These traits distinguish it from other large-sized mustelids such as Eomellivora wimani from the Hemphillian (Hh3-4) of California, Plesiogulo spp. from several Hh3-4 North American localities, and Hoplictis cf. grangerensis from the late Clarendonian (Cl2), Nebraska. It also differs from the Barstovian Legionarictis fortidens and undescribed North American Clarendonian specimens of Sthenictis and Mionictis. Moralesictis intrepidus, previously identified as Eomellivora, unquestionably represents the oldest immigrant mellivorine to North America (Hh 2) and implies the entrance of Eomellivora to North America as a separate immigration event in Hh3-4.http://zoobank.org/urn:lsid:zoobank.org:pub:96AEE149-5B98-4779-AF65-152E8680527D

A capybara jaw (Rodentia, Caviidae, Hydrochoerinae) and an isolated lower tooth fragment were recovered from dredge spoils on Pelican Island in Galveston Bay, Gulf Coast of Texas. The attached matrix on the jaw and most of the dredged material is clay, indicative of the latest Pleistocene Beaumont Formation which underlies the Holocene deposits. In the latest Pleistocene (Rancholabrean) of the United States, capybaras are known from along the Gulf and Atlantic coasts in Texas, Florida, Georgia, South Carolina, and Virginia.These Rancholabrean capybaras have been assigned to three different taxa: Neochoerus pinckneyi Hay, 1923 for the largest size specimens, Hydrochoerus holmesi Simpson, 1928 for the somewhat smaller specimens, or Neochoerus aesopi Leidy, 1853, a name that had been considered invalid because it was not adequately diagnosed, if there is only one late Pleistocene species of North American capybara. The p4 in the new Texas mandible is much smaller than the genotype of N. pinckneyi from the Aransas River in South Texas and similar in size to H. holmesi from the Pleistocene of Florida and Recent H. hydrochaeris. However cheek tooth size and occlusal pattern in Recent capybaras varies significantly with age. Hydrochoerus holmesi cannot be distinguished from N. pinckneyi. None of the late Pleistocene capybaras of South Carolina, including the type and referred specimens of N. aesopi, have any characteristics to differentiate them from H. holmesi or N. pinckneyi. We therefore assign all the Texas and South Carolina specimens to N. aesopi.

Miocene terrestrial mammals are poorly known from the Atlantic Coastal Plain. Fossils of the Order Carnivora from this time and region are especially rare. We describe a carnivoran mandible with a p4 from the late Oligocene or early early Miocene Belgrade Formation in Jones County, North Carolina. Comparisons are made with carnivoran jaws with similar premolar and molar lengths from the late Oligocene and Miocene of North America and Eurasia. These indicate that the North Carolina jaw is assignable to the Ailuridae, a family whose only living member is the red panda. The jaw is tentatively referred to Amphictis, a genus known elsewhere from the late Oligocene and early Miocene of Europe and the early Miocene (Hemingfordian) of North America. The North Carolina mandible compares best with the late Oligocene (MP 28) Amphictis ambiguus from Pech du Fraysse, France, the oldest known member of the Family Ailuridae, and with the early Miocene (MN 1-MN 2a) A. schlosseri from southwestern Germany. This identification is compatible with a late late Arikareean (Ar4, early Miocene, MN 2-3 equivalent) age assignment for the other terrestrial mammals of the Belgrade Formation.

A capybara jaw (Rodentia, Caviidae, Hydrochoerinae) and an isolated lower tooth fragment were recovered from dredge spoils on Pelican Island in Galveston Bay, Gulf Coast of Texas. The attached matrix on the jaw and most of the dredged material is clay, indicative of the latest Pleistocene Beaumont Formation which underlies the Holocene deposits. In the latest Pleistocene (Rancholabrean) of the United States, capybaras are known from along the Gulf and Atlantic coasts in Texas, Florida, Georgia, South Carolina, and Virginia. These Rancholabrean capybaras have been assigned to three different taxa: Neochoerus pinckneyi Hay, 1923 for the largest size specimens, Hydrochoerus holmesi Simpson, 1928 for the somewhat smaller specimens, or Neochoerus aesopi Leidy, 1853, a name that had been considered invalid because it was not adequately diagnosed, if there is only one late Pleistocene species of North American capybara.The p4 in the new Texas mandible is much smaller than the genotype of N. pinckneyi from the Aransas River in South Texas and similar in size to H. holmesi from the Pleistocene of Florida and Recent H. hydrochaeris. However cheek tooth size and occlusal pattern in Recent capybaras varies significantly with age. Hydrochoerus holmesi cannot be distinguished from N. pinckneyi. None of the late Pleistocene capybaras of South Carolina, including the type and referred specimens of N. aesopi, have any characteristics to differentiate them from H. holmesi or N. pinckneyi. We therefore assign all the Texas and South Carolina specimens to N. aesopi.

During 2014–2015, we trapped Texas pocket gophers (Geomys personatus) in coastal South Texas for a translocation study (Hansler et al. 2017). In the 2017 paper, the trapping method was briefly described in three sentences. Herein we provide more detail on the technique and discuss a preliminary experiment demonstrating that cantaloupe is an effective bait for pocket gophers (Rodentia: Geomyidae). Because of their fossorial lifestyle (Williams 1982), pocket gophers are inherently difficult to live trap at the surface because they rarely come above ground. They are usually live-captured by digging and inserting a live trap in their burrow (Sherman 1941; Baker & Williams 1972; Hickman 1979) sometimes using a specially constructed device (Connior & Risch 2009; Moore et al. 2019). Burrow depths for Geomys personatus average 240 cm and range from 80–440 cm (Kennerly 1954). We used standard Sherman live traps near the surface. Baits, such as granola or peanut butter, used for trapping non-fossorial rodents yielded little success in the original study and were not even discussed (Hansler et al. 2017). We observed that captive gophers housed in aquaria seemed to prefer cantaloupe over conventional feed such as rat biscuits, granola, pelleted rodent foods, and a variety of other fruits and vegetables. We therefore began using cantaloupe as bait and it appeared that we captured more gophers with cantaloupe than we did using other baits or no bait. To evaluate whether this observation was correct, we conducted a trial in fall of 2015 in the Flour Bluff area of Corpus Christi, Texas. This study was designed to compare using cantaloupe with using granola and gopher scat bait.Active gopher mounds were located by identifying those that were darker in color due to the moisture still present in the sand. An open entrance to the burrow or freshly plugged entrance signaled that a gopher was active and present in the burrow system below. Wearing gloves so we would not spread human scent in the gopher burrow, we dug into the surface entrance locating where the tunnel bifurcates. We then widened that tunnel to accommodate the trap. Non-collapsible Sherman traps were turned upside down so the foot treadle that springs the door closed was on the roof of the trap. We then coated the floor of the trap with moist sand obtained from inside the gopher's tunnel. Gophers are more likely to enter the trap if the floor is covered in sand rather than left as exposed metal (Sherman 1941). Soil cannot be used on the treadle side of the trap as this would impede the spring mechanism and the door will not close. Placing the trap so the treadle is on the roof allows for a gopher to enter the trap, thus hitting the treadle with its back as it turns around, and consequently causing the door to shut and trap the gopher inside. After coating the floor with sand, we placed the bait in the back of the trap and inserted the trap into the burrow up to the point where the tunnel bifurcates. This ensured that a gopher traveling from either direction would encounter the trap thus reducing the number of traps needed for each burrow system. We filled in the hole behind the trap leaving only a small slit for light and outside air to enter the trap which seems to encourage gophers to investigate the trap. In our previous study, traps were usually investigated within 30 min of setting. If traps were not filled or triggered, or a gopher was not trapped within that time period, the trap usually remained untouched even if left for hours or days. Because of the threat of fire ants and exposure to extreme temperatures for any trapped gophers, we did not leave traps in the ground for more than one hour. After one hour, we pulled up the traps to check for gophers. Gophers that were trapped were weighed, sexed, and released immediately into the tunnel where they were trapped. Traps were soaked in soapy water, rinsed thoroughly, and dried between sessions.We set 23 sets of three traps each for a total of 69 traps. Every set had one trap with a small slice of cantaloupe, one trap with a tablespoon of granola, and one trap with a tablespoon of gopher scat. We put three traps for every trial in close proximity to ensure that terrain and location was not a mitigating factor for trapping.Results of the trapping are presented in Table 1. Four gophers were trapped with cantaloupe; no gophers were trapped in any of the other traps. This supports our initial observations that cantaloupe is a more effective bait for trapping than granola or scat. This agrees with the previous cantaloupe-baited trapping results of approximately one trapped gopher out of every six traps set (Hansler et al. 2017). Scat as bait operated as a deterrent to the gophers with 14 out of 23 scat baited traps being filled with dirt. The traps set with scat were almost twice as likely to be filled with dirt as both the cantaloupe and the granola baited traps. Granola baited traps were almost twice as likely to be uninvestigated as those baited with cantaloupe or scat. Cantaloupe may be an effective bait because of the moisture content of the melon. Pocket gophers do not drink free water (Hanley 1944; DeVries & Sikes 2009) and may prefer foods with high water content to meet their metabolic water demand. In addition, because the scat baited traps seemed to offend the gophers, scat may be employed as a deterrent to keep gophers out of areas such as gardens or yards.Connior & Risch (2009) compared trapping success for their box traps (made of wood and measuring approximately 40–50 cm long by 12 cm wide by 10 cm deep) with the pipe traps used by Baker &Williams (1972). Both had a 25–50% capture success for their trials. Our cantaloupe baited traps had a 17% success rate (4/23). However, Connior & Risch (2009) checked each trap up to six times, with traps reset if plugged with dirt. For our project, each trap was only checked once, one hour after it was set, as noted above. An additional advantage with our method, other than not having to dig down to locate the main tunnel and having to refill the hole immediately, is that it is easy to set a set a large number of traps, especially if light weight Sherman traps are used.We thank Scott Henke, Caesar Kleberg Wildlife Research Institute, Department of Animal, Rangeland and Wildlife Sciences Texas A&M University-Kingsville for supervising and supporting Tara Hansler's graduate project which was the basis of the original study. Collection and use of animals in this study were approved by the Texas A&M University-Kingsville Animal Care and Use Committee (ACUC 2009-04-29A). We also thank the anonymous reviewers whose comments improved the final version of this manuscript.

One new species of ailurid and two new genera and species of oligobunine mustelids are recognized from the early Hemingfordian (He1) Miller Local Fauna, Dixie/Levy County, Florida (U.S.A.). The ailurid is congeneric with the European early Miocene Amphictis but is smaller than the Old World species. One new oligobunine is a small hypercarnivore with a reduced metaconid and talonid on m1 and is related to Promartes and Oligobunis. The second is a small hypocarnivore with an enlarged metaconid on m1 and a relatively large m2 and is is related to Brachypsalis. Four other mustelids new to the Miller Local Fauna are the oligobunines cf. Brachypsalis sp. and cf. Oligobunis sp. and the leptarctines Craterogale cf. simus and Leptarctus ancipidens. These small carnivorans most closely resemble those of the early Hemingfordian Runningwater Formation of the better known northern Great Plains faunas and add further support that the Miller Local Fauna is older than the late early or early late Hemingfordian Thomas Farm.

Geomys personatus maritimus (Maritime Pocket Gopher) is a genetically distinct subspecies of pocket gopher that occurs only in deep, sandy soils located in Nueces and Kleberg counties of southern Texas. The US Fish and Wildlife Service considers it a species of concern. Pocket gophers are considered a nuisance because they dig burrows and create mounds in landscaped areas. Lethal control options are not advised because of the Maritime Pocket Gopher's rarity status and recent public attitude disfavoring lethal methods. However, short-distance translocation might be a management option, but research to determine its viability is lacking. We captured 15 Maritime Pocket Gophers from athletic fields and commercial properties in Corpus Christi, TX. For all captured gophers, we subcutaneously or intraperitoneally implanted a radio transmitter, translocated the animals to private property within 2 km from their capture site, and radio-tracked their movements for <= 4 months to determine if their behavior and activity differed from 4 control gophers. Subcutaneous transmitters implanted in the scapular region were lost by 86% of the gophers (n = 12 of 14), while 100% (n = 5) of the gophers retained intraperitoneal-implanted transmitters. Relocated Maritime Pocket Gophers did not return to their site of origin. Gopher movements generally were away from their homing lines (i.e., an imaginary line drawn between each translocated gopher's capture site and release site), and they did not become successively closer to their respective sites of capture. Therefore, short-distance translocation has the potential to be a management option for nuisance gophers.

We describe cranial and mandibular remains of three undescribed individuals of the giant mustelid Megalictis ferox Matthew, 1907 from the latest Arikareean (Ar4), Early Miocene mammal fauna of Nebraska, and Wyoming (USA) housed at the American Museum of Natural History (New York, USA). Our phylogenetic hypothesis indicates that Ar4 specimens assigned to M. ferox constitute a monophyletic group. We assign three additional species previously referred to Paroligobunis to Megalictis: M. simplicidens, M. frazieri, and “M.” petersoni. The node containing these four species of Megalictis and Oligobunis forms the Oligobuninae. We test the hypothesis that Oligobuninae (Megalictis and Oligobunis) is a stem mustelid taxon. Our results indicate that the Oligobuninae form the sister clade to the crown extant mustelids. Based on the cranium, M. ferox is a jaguar-size mustelid and the largest terrestrial mustelid known to have existed. This new material also sheds light on a new ecomorphological interpretation of M. ferox as a bone-crushing durophage (similar to hyenas), rather than a cat-like hypercarnivore, as had been previously described. The relative large size of M. ferox, together with a stout rostrum and mandible made it one of the more powerful predators of the Early Miocene of the Great Plains of North America.

This paper describes the Dinero Local Fauna, a small collection of fossils from a roadcut in the Goliad Formation in Live Oak County, Texas. It is from higher in the section than the classic early Clarendonian Lapara Creek Fauna. Four mammalian taxa are present: the rodent Ceratogaulus cf. anecdotus, and the equids Pseudhipparion skinneri, Cormohipparion cf. ingenuum, and Calippus cf. cerasinus. The co-occurrence of these taxa is consistent with the late Clarendonian age assignment.

Cernictis repenningi is a new species of mustelid carnivore from the middle Hemphillian (late Miocene) Bidahochi Formation of northeastern Arizona. Cernictis is a late Miocene immigrant from the Old World, where it apparently is represented by undescribed material from China. The upper dentition of this material indicates that Cernictis is a primitive member of the tribe Galictini, which includes one or two extant Neotropical genera, five extinct New World genera, and three or four extinct Old World genera.

Cernictis repenningi is a new species of mustelid carnivore from the middle Hemphillian (late Miocene) Bidahochi Formation of northeastern Arizona. Cernictis is a late Miocene immigrant from the Old World, where it apparently is represented by undescribed material from China. The upper dentition of this material indicates that Cernictis is a primitive member of the tribe Galictini, which includes one or two extant Neotropical genera, five extinct New World genera, and three or four extinct Old World genera.

Leptarctus mummorum, new species, from the late Clarendonian of northwestern Kansas is described for the unusual mustelid genus Leptarctus Leidy, 1856. It is distinguished from other species of the genus by its larger size, temporal crests meeting along the central axis of the skull. relatively longer rostrum and bullar processes, depression anterior to the orbit, oblique ridge along the orbital wall, tipper third premolar (P3) with an accessory internal accessory cusp, and upper fourth premolar (P4) with a greatly enlarged parastyle. It appears that previous suggestions that Leptarctus was strictly herbivorous and arboreal arc not supportable based on the morphologies of the skull and dentition.

The fluvial Goliad Formation crops out along the coastal plain of South Texas. In the 1950s-60s, the Lapara Creek Fauna, from low in the section and including virtually all the then known Goliad fossils, was assigned to the early Clarendonian (12 Ma) North American Land Mammal Age (NALMA). A Labahia Mission Fauna, based only on a tooth and leg bone that were never described, was placed in the succeeding early to mid- dle Hemphillian NALMA. At that time the Miocene-Pliocene boundary was placed at about 13 Ma, which meant the Goliad Formation was Pliocene in age, a correlation still followed by some. In the 1970s, the Mio-Pliocene boundary was set at about 5.5 Ma, and the Goliad was reassigned to the middle (Lapara Creek) to late (Labahia Mission) Mio- cene. This paper documents a late Clarendonian (10 Ma) Local Fauna (LF) and a latest Hemphillian (5 Ma) LF in the Goliad. The Dinero LF occurs high in the section, below the caliche that caps the Goliad. The concurrent range zone of its taxa (Ceratogaulus anecdotus, Pseudhipparion skinneri, Calippus cf. cerasinus, and Cormohipparion cf. in- gennum) is late Clarendonian. The mainly reworked Lake Corpus Christi LF includes the horses Dinohippus cf. mexicanus, Astrohippus stockii, Pseudhipparion simpsoni, Nan- nippus cf. beckensis, N. aztecus, and Neohipparion eurystyle. These support a latest Hemphillian age (early Pliocene) for the caliche cap of the Goliad. The Lapara Creek Fauna correlates with the Textularia "W" benthic foraminiferal biozone; the Dinero LF to the Bolivina-Cibicides biozone; and the Lake Corpus Christi LF to the Bigenerina "6" biozone.

South Texas has a rich late Cenozoic fossil record. At least 18 extinct genera and 27 extinct species of fossil mammals are known from the late Pleistocene alone. Many of these were participants in a major evolutionary event known as the Great American Biotic Interchange, which resulted from the formation of the Panamanian land bridge between North and South America. South American genera that traveled north and are present in the late Pleistocene of South Texas are the giant ground sloths Eremotherium, Paramylodon, and Megalonyx, the large relatives of armadillos Holmesina and Glyptotherium, and the capybara Neochoerus. North American genera that traveled south include the wolf Canis the bears Tremarctos and Arctodus, the saber cat Smilodon, the gomphothere elephant Cuvieronius, the tapir Tapirus, the horse Equus, the peccary Platygonus, and the llamas Palaeolama and Hemiauchenia. The 6.5 ft (2 m) tall, flightless, predatory, ‘terror bird’ Titanis is another South American immigrant. It had previously been thought that it entered North America about 2 Ma and survived in South Texas until the end of the Pleistocene. However, rare earth element dating of the South Texas Titanis shows that it arrived in North America in the early Pliocene (about 5 Ma), shortly before the completion of the Panamanian land bridge.

The giant flightless terror bird Titanis walleri is known from Florida and Texas during the late Neogene. The age of T. walleri is problematic because this taxon co-occurs with temporally mixed (i.e., time-averaged) faunas at two key sites. Thus, prior to this study, T. walleri from the Santa Fe River, Florida (type locality), was either as old as late Pliocene (ca. 2.2 Ma) or as young as latest Pleistocene (ca. 15 ka). Likewise, T. walleri from the Nueces River, Texas, was either early Pliocene (ca. 5 Ma) or latest Pleistocene (ca. 15 ka). In order to better resolve this age range, the rare earth element (REE) patterns of T. walleri from the Santa Fe River, Florida, were compared to two biochronologically distinctive groups (late Pliocene versus late Pleistocene) of fossil mammals from the same locality. Similarly, the REE patterns of T. walleri from Texas were compared to two groups (early Pliocene versus latest Pleistocene) of fossil mammals from the same locality. The REE patterns of T. walleri from Florida are indistinguishable from those of the co-occurring late Pliocene mammals. Likewise, the REE pattern of T. walleri from Texas is indistinguishable from those of the co-occurring early Pliocene mammals. Given these REE constraints, the revised age of T. walleri is early Pliocene in Texas (ca. 5 Ma) and late Pliocene (ca. 2.2-1.8 Ma) in Florida. As such, T. walleri is interpreted as an early immigrant during the Great American Interchange prior to the formation of the Isthmian land bridge. No evidence currently exists for Pleistocene T. walleri in North America.

South Texas has a rich late Cenozoic fossil record. At least 18 extinct genera and 27 extinct species of fossil mammals are known from the late Pleistocene alone. Many of these were participants in a major evolutionary event known as the Great American Biotic Interchange, which resulted from the formation of the Panamanian land bridge between North and South America. South American genera that traveled north and are present in the late Pleistocene of South Texas are the giant ground sloths Eremotherium, Paramylodon, and Megalonyx, the large relatives of armadillos Holmesina and Glyp- totherium, and the capybara Neochoerus. North American genera that traveled south include the wolf Canis the bears Tremarctos and Arctodus, the saber cat Smilodon, the gomphothere elephant Cuvieronius, the tapir Tapirus, the horse Equus, the peccary Platygonus, and the llamas Palaeolama and Hemiauchenia. The 6.5 ft (2 m) tall, flightless, predatory, 'terror bird' Titanis is another South American immigrant. It had previously been thought that it entered North America about 2 Ma and survived in South Texas until the end of the Pleistocene. However, rare earth element dating of the South Texas Titanis shows that it arrived in North America in the early Pliocene (about 5 Ma), shortly before the completion of the Panamanian land bridge.

Twenty-one species of extinct, large, late Pleistocene mammals have been identified from sand and gravel pits in terrace and valley-fill deposits of the Lower Nueces River of South Texas. A log from the alluvial valley-fill unit that produced most of the fossils has been 14 C dated at 13,230 ± 110 YBP. The distal end of a mammoth tusk from this unit exhibits modifications that may not be the result of natural processes and may have been altered by contemporaneous humans. The 185 mm long tusk tip is rounded on the distal end, suggesting possible use as burnisher-billet. The tusk segment also has a curved, rounded, and polished groove 3 cm wide and 1 cm deep transverse to the long axis. The groove may have resulted from being used as a thong stretcher, hide softener, or shaft straightener. Alternatively, the groove may be a natural feature, formed by the stripping of leaves from twigs, an activity noted in living elephants. Sections of a limb bone have marks that may have been caused by butchering activities.

Eleven genera and twelve species of Carnivora are known from the late Miocene Love Bone Bed Local Fauna, Alachua County, Florida. Taxa from there described in detail for the first time include the canid cf. Urocyon sp., the hemicyonine ursid cf. Plithocyon sp., and the mustelids Leptarctus webbi n. sp., Hoplictis sp., and ?Sthenictis near ?S. lacota. Postcrania of the nimravid Barbourofelis indicate that it had a subdigitigrade posture and most likely stalked and ambushed its prey in dense cover. The postcranial morphology of Nimravides (Felidae) is most similar to the jaguar, Panthera onca. The carnivorans strongly support a latest Clarendonian age assignment for the Love Bone Bed. Although the Love Bone Bed local fauna does show some evidence of endemism at the species level, it demonstrates that by the late Clarendonian, Florida had become part of the Clarendonian chronofauna of the midcontinent, in contrast to the higher endemism present in the early Miocene and in the later Miocene and Pliocene of Florida.