The oldest North American primate and mammalian biogeography during the Paleocene–Eocene Thermal Maximum

Abstract

Undoubted primates first appear almost synchronously in the fossil records of Asia, Europe, and North America. This temporal pattern has complicated efforts to reconstruct the early dispersal history of primates in relation to global climate change and eustatic fluctuations in sea level. Here, I describe fossils from the Tuscahoma Formation on the Gulf Coastal Plain of Mississippi documenting an anatomically primitive species of Teilhardina that is older than other North American and European primates. Consistent with its antiquity, a phylogenetic analysis of dental characters recognizes Teilhardina magnoliana, sp. nov., as the most basal member of this genus currently known from either North America or Europe. Its stratigraphic provenance demonstrates that primates originally colonized North America near the base of the Paleocene–Eocene Thermal Maximum (PETM), but before an important fall in eustatic sea level. Correlation based on carbon isotope stratigraphy and sequence stratigraphy indicates that the earliest North American primates inhabited coastal regions of the continent for thousands of years before they were able to colonize the Rocky Mountain Interior. The transient provincialism displayed by early North American primates corresponds to similar biogeographic patterns noted among fossil plants. Decreased precipitation in the Rocky Mountain Interior during the early part of the PETM may have been an important factor in maintaining biotic provincialism within North America at this time. These results underscore the need to obtain multiple, geographically dispersed records bearing on significant macroevolutionary events such as the PETM.

Primates and several other modern mammalian groups (including artiodactyls and perissodactyls) first appear in the North American fossil record shortly after the Paleocene–Eocene (P/E) boundary (1–3), an interval marked by rapid and dramatic global warming known as the Paleocene–Eocene Thermal Maximum (PETM) (1, 4–6). Although it is generally accepted that the warmer climatic conditions during the PETM facilitated dispersal of primates and other mammals across two high latitude land bridges (Beringia and the North Atlantic land bridge then connecting Greenland and Scotland), identifying where these mammals originated and the pathway they took to achieve their widespread distribution remains contentious (1–3, 7, 8).

The most thoroughly studied North American terrestrial sequence spanning the PETM is in the Bighorn Basin of northwestern Wyoming (3, 4, 9, 10). There, the local first appearance of primates (including the omomyid Teilhardina brandti) occurs in biostratigraphic zone Wa-0, at a stratigraphic level estimated to be 19–25 Kyr above the P/E boundary (1). Correlation using the global carbon isotope excursion (CIE) at the P/E boundary suggests that T. brandti from the Bighorn Basin is slightly younger than the oldest primates currently known from Europe (Teilhardina belgica) and Asia (Teilhardina asiatica) (1, 7, 11). This agrees with phylogenetic analyses of Teilhardina, which recognize T. asiatica as the most basal species, followed by T. belgica and North American species of Teilhardina, respectively (1, 11). These geological and phylogenetic data have been interpreted as evidence that primates dispersed from Asia to Europe to North America during the PETM, and that dispersal of Teilhardina from Europe to North America was made possible by an important drop in eustatic sea level during that interval (1).

Here, I describe an anatomically primitive species of Teilhardina from the Gulf Coastal Plain of Mississippi. This primate is one component of the earliest Eocene Red Hot local fauna, which also includes sharks and rays, bony fishes, snakes, lizards, crocodilians, birds, and a variety of other mammals (12–16). The Red Hot local fauna was recovered by screen-washing fine-grained glauconitic sands forming the uppermost part of the Tuscahoma Formation at an outcrop south of Meridian, MS (15, 17). The distinctive geographic and stratigraphic setting of the Red Hot local fauna provides the basis for more detailed understanding of the initial dispersal of primates and other mammals into North America and the biogeographic response of North American mammals to global warming during the PETM.

Systematic Paleontology

Primates Linnaeus, 1758.

Omomyidae Trouessart, 1879. Teilhardina Simpson, 1940. Teilhardina magnoliana, sp. nov.

Holotype.

Carnegie Museum of Natural History (CM) 70435, an isolated left M₂ (Fig. 1).

Fig. 1.

Composite partial dentition of Teilhardina magnoliana from the Tuscahoma Formation (earliest Eocene), Gulf Coastal Plain of Mississippi. (a) Composite right upper dentition (P³–M²) in occlusal view, based on CM 67854, CM 77210, CM 70431, and CM 70436 (reversed from left side). (b and c) Composite left lower dentition (P₄–M₃) in occlusal (b) and buccal (c) views, based on CM 73229, CM 67856, CM 70435 (holotype), and CM 70427 (reversed from right side).

Hypodigm.

The holotype; CM 67854, isolated right P³; CM 67856, isolated left M₁; CM 67858, isolated right M¹; CM 67860, isolated left M₁; CM 67861, isolated left P³; CM 70422, isolated left M² lacking the metacone; CM 70427, isolated right M₃; CM 70430, isolated right M₃; CM 70431, isolated right M¹; CM 70433, isolated left P⁴ lacking the buccal side of the paracone; CM 70434, isolated right P₄; CM 70436, isolated left M²; CM 73229, isolated left P₄ trigonid; CM 77209, isolated right M₃ lacking part of the hypoconulid lobe; CM 77210, isolated right P⁴; CM 77211, isolated right M¹; CM 77212, isolated right M² lacking the metacone [Fig. 1; see supporting information (SI) Text for detailed descriptions].

Etymology.

For Mississippi, the Magnolia State.

Type Locality.

CM locality 517, uppermost part of the Tuscahoma Formation, Lauderdale County, MS.

Diagnosis.

Trigonid of P₄ differs from that of all other species of Teilhardina except T. asiatica in being abbreviated mesiodistally, with relatively vertical preprotocristid and abrupt transition from buccal surface to postvallid. P₄ differs from that of T. asiatica in having metaconid more closely appressed to the base of the protoconid and weak mesial extension of cristid obliqua partially ascending postvallid. Lower molars differ from those of all other species of Teilhardina except T. asiatica in being lower-crowned. M_1–2 further differ from those of other North American species of Teilhardina in being relatively narrower, although not so narrow as in T. asiatica.

Discussion.

T. magnoliana was a tiny primate. A regression equation predicting body mass from M₁ area among living and fossil tarsioids (18) yields an estimated mean body mass of 28 g for T. magnoliana, placing this species at the lower limit of the range of adult body mass encompassed by living primates (19). Primates that are this small must adopt diets that are rich in calories, and small omomyids like T. magnoliana are thought to have eaten mainly fruits, gums, and insects (20). Our current knowledge of the anatomy of T. magnoliana limits the range of paleobiological inferences we can draw for this species. However, the closely related species T. asiatica possesses relatively small orbits that have been interpreted as evidence for a diurnal activity pattern (11). Tarsal bones of T. belgica suggest that it was an active arboreal quadruped that also engaged in powerful leaping (21).

Results

Phylogenetic analyses indicate that T. magnoliana is a very basal member of the genus Teilhardina (Fig. 2a; for details see SI Text and SI Fig. 3), which is usually regarded as the most basal genus within Omomyidae and one of the most basal primate taxa currently known (1, 11, 22). Notably primitive characters that are retained in T. magnoliana (and T. asiatica) include its diminutive size, relatively low molar crown height, narrow M₂, and the distinctive structure of its P₄ trigonid (including the relatively vertical orientation of the preprotocristid and the very abrupt transition from the buccal surface of the trigonid to the postvallid). T. magnoliana more closely resembles European T. belgica than it does North American T. brandti. T. brandti is slightly larger than T. magnoliana and T. belgica and differs morphologically from both species in having more nearly square occlusal outlines of M_1–2 and stronger buccal cingulids on its lower molars.

Fig. 2.

Phylogenetic, biogeographic, and stratigraphic relationships among basal species of the primate Teilhardina and closely related taxa. (a) Single most parsimonious tree recovered from phylogenetic analysis of 26 dental characters (for details see SI Text and SI Fig.3). In-group branches are color coded to reflect geographic occurrences of terminal taxa (red for Asia, yellow for North America, orange for Europe). (b) Most parsimonious pattern of dispersal for early species of Teilhardina given the tree topology depicted in a, superimposed on a paleogeographic map of the globe near the P/E boundary. Under this scenario, Teilhardina originated in Asia and dispersed across Beringia to occupy coastal regions of North America during the early part of the PETM (red pathway). A major fall in eustatic sea level later in the PETM allowed Teilhardina to disperse across the North Atlantic land bridge connecting Greenland with northwestern Europe (orange pathway). Increased precipitation in the Rocky Mountain Interior finally allowed Teilhardina to colonize the Bighorn Basin during zone Wa-0 (yellow pathway). (c–e) Correlation among sites in North America and Europe that yield fossils of the early Eocene primate Teilhardina. Lowest stratigraphic record of Teilhardina in each local section is indicated by a red dot. Green lines indicate correlation between Polecat Bench (c) in the Bighorn Basin of Wyoming and Dormaal, Belgium (d), based on stratigraphic position of Teilhardina fossils with respect to the CIE (1). Blue line indicates correlation between Dormaal, Belgium (d), and the Red Hot local fauna of Mississippi (e) based on the type 1 sequence boundary separating sequence TP2.3 from the overlying sequence TE1.1, which was caused by an important fall in eustatic sea level early in the PETM (15, 17, 24, 25).

All T. magnoliana fossils were recovered from the base of the T4 sand, a lenticular fine-grained sand body that locally forms the uppermost part of the Tuscahoma Formation (15, 17). The sediments comprising the T4 sand have been interpreted as estuarine tidal channels that were deposited very close to the ancient shoreline (17). Reflecting this depositional history, the T4 sand preserves a mixture of terrestrial and marine organisms. Dinoflagellate cysts from the T4 sand include abundant Apectodinium augustum and other Apectodinium species (15), indicating that the T4 sand correlates with the Apectodinium acme that coincides with the CIE at sites in Belgium, Austria, and New Zealand (23, 24). The Apectodinium acme is interpreted as an ecological response to the elevated sea surface temperatures that characterized the PETM (23). Accordingly, the cooccurrence of T. magnoliana and abundant Apectodinium in the T4 sand suggests that PETM warming began sometime before the deposition of this unit.

The unique stratigraphic context of T. magnoliana invites comparisons with penecontemporaneous strata in Europe that also yield Teilhardina and the Apectodinium acme. T. belgica, the oldest primate currently documented from Europe, occurs in the Dormaal mammal assemblage, derived from the basal part of the Tienen Formation in Belgium (1, 24, 25). The fluvial strata that yield the Dormaal mammal assemblage accumulated as incised valley-fill deposits, overlying an erosional surface that resulted from a major fall in eustatic sea level (24, 25). In contrast, the T4 sand that yields T. magnoliana was deposited under estuarine conditions, immediately before a major fall in eustatic sea level that caused a regional unconformity across the Gulf Coastal Plain of Mississippi and Alabama (15, 17). Only one significant drop in eustatic sea level is known to have occurred during the PETM, and the type 1 sequence boundary it produced provides an important datum for correlating strata on either side of the North Atlantic at this time (Fig. 2 d and e) (15, 24–26). Correlation on the basis of sequence stratigraphy indicates that T. magnoliana is older than T. belgica, although the precise duration of the temporal gap separating them remains unknown. The geographic and stratigraphic provenance of T. magnoliana demonstrates that Teilhardina inhabited coastal regions of North America before the fall in sea level during the PETM, thereby falsifying the hypothesis that this marine regression allowed Teilhardina to disperse from Europe to North America at this time (1). To the contrary, dispersal of Teilhardina from North America to Europe may have been facilitated by newly emergent land bridges resulting from lower sea level during the PETM regression (Fig. 2b).

Aside from T. magnoliana, the oldest North American primates come from zone Wa-0 in the Bighorn Basin of Wyoming (1). On the basis of their respective positions within the CIE, T. brandti from zone Wa-0 is younger than T. belgica (1). The temporal gap between T. brandti and T. magnoliana must be even greater, although it is probably <25 Kyr, the estimated duration of the interval between the P/E boundary and the first appearance of T. brandti (Fig. 2 c–e) (1). Although much of the mammalian fauna found associated with T. magnoliana in the T4 sand remains undescribed, certain taxa corroborate the correlation shown in Fig. 2. For example, the most common rodent in the T4 sand is the basal cylindrodontid Tuscahomys (16), which is also the most common rodent currently documented from zone Wa-M in the Bighorn Basin (although the specimens in question were identified as “cf. Reithroparamys, new species” in ref. 9). A second distinctive element of the Wa-M faunal zone in the Bighorn Basin is the hyopsodontid Haplomylus zalmouti, which is striking primarily for its diminutive size (9). A tiny species of Haplomylus, equivalent in size to H. zalmouti, is also known from the T4 sand, but minor differences in morphology suggest that the Mississippi Haplomylus specimens may pertain to a different, but closely related species. These data support a tentative correlation between the Red Hot local fauna and zone Wa-M in the Bighorn Basin (Fig. 2 c–e).

Discussion

A previous analysis of mammalian dispersal patterns among the Holarctic continents near the P/E boundary suggested that most, if not all, of the higher-level mammalian taxa that first appear during this interval originated in Asia and dispersed via Beringia to both North America and Europe (2). The geological and phylogenetic data reported here for T. magnoliana are consistent with this larger biogeographic pattern, indicating that primates also initially colonized North America by dispersing across Beringia (Fig. 2). Significantly, the stratigraphic evidence from the Tuscahoma Formation demonstrating that Teilhardina dispersed to North America before the drop in sea level during the PETM further corroborates their initial colonization of North America via Beringia. Evidence from fossil marine invertebrates reveals that Beringia was consistently emergent until the late Miocene, when mollusks that were previously restricted to the Arctic Ocean basin first occur in the North Pacific (27). As such, during the early Cenozoic, Beringia was always a potential avenue for dispersal of terrestrial organisms, regardless of fluctuations in eustatic sea level. In contrast, early primates that attempted to disperse in a progressively westerly direction from Asia to Europe to North America would have faced a series of marine barriers, including the Turgai Straits that then separated eastern Asia from Europe, the North Sea, and the Atlantic Ocean (Fig. 2b). These marine barriers to dispersal would have been even more formidable during intervals of high eustatic sea level, such as that which characterized the uppermost Tuscahoma Formation.

Given the density of paleontological sampling across the P/E boundary in the Bighorn Basin (1, 3, 9, 10), the discovery of North American primates such as T. magnoliana that antedate the first record of this group in the Bighorn Basin is surprising and demands an explanation. Biotic provincialism is perhaps the most compelling potential reason for the apparent diachroneity in the local first appearances of primates along the Gulf Coast and the Bighorn Basin during the PETM. Small-bodied primates such as Teilhardina might be expected to be particularly prone to developing provincial distributions, especially during intervals of dynamic climate change such as the PETM. Tiny primates such as T. magnoliana are invariably arboreal, and primates are among the most thermophilic of all mammals. Additional factors, including specific food requirements or preferences, undoubtedly influenced their geographic distribution as well.

Fossil spores and pollen from the T4 sand indicate a humid, paratropical biome along the Gulf Coastal Plain during the PETM (28) and include distinctive plants that make a transient appearance during zone Wa-0 in the Bighorn Basin (4). Analysis of fossil floras and paleosols in the Bighorn Basin suggests significantly drier conditions during the earliest part of the PETM, followed by a return to higher levels of precipitation later in this interval (4, 5). Greenhouse warming at the onset of the PETM would have allowed Teilhardina to colonize coastal regions of North America, but primates and other Eocene mammalian immigrants apparently failed to extend their ranges into the Rocky Mountain Interior until regional patterns of precipitation ameliorated, and local floral communities adapted to the new climatic regime. The development of significant biotic provincialism within North America during the PETM emphasizes how local records cannot adequately reflect the complexity of evolutionary and ecological changes on a continental scale, even when such local records are exceptionally dense and well documented.

Materials and Methods

Specimens described in this report consist of isolated teeth collected by the author and his assistants during the course of multiple field seasons. For detailed anatomical descriptions of these specimens, see SI Figs. 4–11. The phylogenetic analysis performed here is based on 11 taxa and 26 characters. The character–taxon matrix was assembled in MacClade 3.04 (29), and the data matrix and recovered trees were analyzed by using PAUP* 4.0b10 for Macintosh PPC (30). The character list, the taxon–character matrix, and additional information regarding the phylogenetic analysis are provided in SI Fig. 3.