Fossil Record of the Primates from the Paleocene to the Oligocene

Abstract

The Paleogene primate fossil record is reviewed following higher systematic categories. Among Strepsirhini, Adapiformes underwent Eocene radiations in North America (Notharctinae) and Europe (Cercamoniinae, Adapinae). Several occasional occurrences due to dispersals are found in North America, Europe, and Africa. Asia reveals a limited diversification (Sivaladapidae) and isolated occurrences indicating a central yet poorly understood role. In Africa the origin of living Lemuriformes is documented in the Late Eocene; odd stem lemuriforms occur earlier. The Eocene florescence of Omomyiformes is documented in North America (Anaptomorphinae, Omomyinae) and in Europe (Microchoeridae). Isolated occurrences, including the stem genus Teilhardina, are known in Asia. Two genera of Tarsiidae, known in the Middle Eocene of Asia, lead to a possible character-based definition of Haplorhini. The Asiatic Eosimiidae may belong in this group, and Archicebus may possibly lie on its stem. The Eocene South Asiatic Amphipithecidae are specialized hard-object feeders whose affinities remain enigmatic. Character-based Anthropoidea, or Simiiformes, are documented in the Late Eocene and Oligocene of Africa (Parapithecidae, Proteopithecidae, Oligopithecidae, Propliopithecidae). Toward the end of the Oligocene, the first African proconsuloids and the first South American platyrrhines appear. Anthropoidean origins are still a field of debate and discovery, with unconvincing Asiatic stem simians and a possible role for African Afrotarsiidae. The fossil record is extremely uneven, going from richly documented lineages in the Eocene of North America, to well-delineated radiations in the Eocene of North America and Europe and the Eocene–Oligocene of Africa, to more dispersed occurrences and enormous gaps during the early periods in Africa and Asia. The latters explain persisting controversies. Many aspects of primate evolution are documented over almost 20 million years, including locomotion, diet, vision and other sensory capacities, brain evolution, and one aspect of social structure via sexual dimorphism. The best records allow researchers to approach specific lineages, evolutionary modes, and analysis of faunistic changes.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

When Ludwig Rütimeyer (1862) referred some dental remains from the Swiss locality Egerkingen to an Eocene primate, his contemporaries did not believe that this was possible. Othniel Marsh later recognized primates through some lemur-like limb bones from the Eocene of Wyoming. When skulls with complete postorbital bars were found in the Quercy phosphate quarries in 1873, the scientific community had to admit that lemur-like primates had inhabited North America and Europe in Eocene times, and that Adapis, named much earlier by Georges Cuvier from a partial and crushed skull, was indeed a primate. Since then the discoveries have continued; indeed, so many have been added in the last decades that there are now more than 250 species of fossil primates known between the Late Paleocene and the end of the Oligocene. The record also includes families of Plesiadapiformes that are excluded here (see below).

The Paleogene is the first period of the Tertiary sub-era. It comprises three successive Epochs: the Paleocene, Eocene, and Oligocene. It starts at the famous Cretaceous-Tertiary (K-T) boundary, a geological limit around 65 million years ago (“Ma” will be used in the text for million years ago, and “My” for million-year durations). This limit coincides with an extinction event which affected many groups in the ocean and on land, including the then-dominant dinosaurs. These extinctions allowed the explosive diversification of mammals, among them the primates. However, not all groups of mammals are recognizable in the Early Paleocene. Several modern orders of mammals are well known only 10 Ma later, in the Eocene. Such is the case for primates, despite one exception in the Late Paleocene. Geological epochs, the main stages, some geological formations, stratigraphic scales, and some of the most important localities are shown in Fig. 1.

Fig. 1

Summary chart showing ages, epochs, marine standard stages, and mammal ages or reference-levels for terrestrial deposits on different continents. Some of the main localities or formations containing primates are indicated. Ma is for millions of years, LMA is for land mammal ages, and Ref. Levels is for the European mammalian reference-levels (MP). The North American Wasatchian is recalibrated following recent work on the Wasatchian-Bridgerian transition. The richest primate records are found in the best-documented intervals, especially the Early Eocene in North America, the Late Eocene in Europe (N Quercy Loc refers to the new Quercy localities), and the Eocene-Oligocene in North Africa (BQ2 to M refer to the Fayum fossil localities)

The Order Primates can be defined by a series of derived characters shared by all living members of the order: presence of a postorbital bar (or complete postorbital septum) surrounding large eyes that are directed forward and allow some degree of stereoscopic vision; presence of a maximum of two incisors, indicating loss of one incisor relative to primitive eutherians; a tympanic bulla formed by the petrosal; presence of an opposable hallux (prehensile foot) and typically presence of nails instead of claws (the two living primate groups bearing claws – Callitrichinae among South American monkeys and Daubentonia among Malagasy lemuriforms – are considered to have secondarily evolved claws from a nailed ancestor). These synapomorphies allow a clear delineation of living primates from all other mammals, including tree shrews, which between the 1920s and 1980s were usually considered to be primitive primates. This same definition excludes the Paleocene-Eocene Plesiadapiformes, which lack a postorbital bar and large forward-directed eyes, and also bear claws. If another more inclusive definition of primates is used, subsuming the plesiadapiforms, the above criteria apply at a lower systematic level, defining Euprimates as opposed to Plesiadapiformes (see chapter “Primate Origins and Supraordinal Relationships: Morphological Evidence,” Vol. 2).

The origin of the primates as defined above is still undocumented, because terrestrial faunas from the right time and region are lacking. A dense and continuous fossil record of mammals is known in North America, but primates originated elsewhere. A more discontinuous fossil record of Paleocene mammals is known in South America, in Europe, and in Eastern Asia (Mongolia, China), but again without early primates. Despite recent progress, the Paleocene fossil record of mammals remains very poor in Africa, although one fauna yielded the single Paleocene primate known at the moment, Altiatlasius from Morocco, discussed below. This fossil shows that primates had reached Africa by the Late Paleocene; however, there is strong evidence that they did not originate in Africa but rather in Asia, where their two living sister groups, the tree shrews and flying lemurs, are found today (see chapter “Primate Origins and Supraordinal Relationships: Morphological Evidence,” Vol. 2). A Paleocene fossil record of Southeast or Western Asia should give some definitive clues to this enigma; it would at the same time test a competing scenario, that of an origin on the northward-rafting Indian plate.

The Paleocene-Eocene boundary, 55 Ma ago, witnessed a series of climatic and paleogeographic events that resulted in the first occurrence of primates in North America, Europe, and Asia (China). This occurrence is part of a much broader dispersal event concerning several orders of mammals as well as other animals and plants. This dispersal followed a dramatic episode of global warming, which is one of the most fascinating climatic events discovered in the last 20 years (e.g., Aubry et al. 1998). Starting in the Early Eocene , a tropical climate and environment at high latitudes allowed the radiation of primate groups on northern continents, including Europe and North America. For geological and historical reasons, these radiations have been the best documented, allowing the study of well-circumscribed primate radiations during almost 20 My. However, these relatively well-known groups became extinct around the Eocene-Oligocene boundary, close to 33 Ma. The latter is marked by a global climatic cooling, rendering northern continents inhospitable for primates. Therefore, these Eocene radiations shed no light on the origin of living primate families, the history of which took place in Asia or on southern continents. This explains why, despite a rich Eocene fossil record, there are still controversies concerning the origin of major groups of living primates, among them the anthropoids: they originated in regions and times with a still insufficient fossil record. The northern Eocene radiations also increase the complexity of primate systematics and phylogeny. The phylogeny inferred from living primates is consensual. It divides them into two suborders: the Strepsirhini, which include the Lemuriformes (Malagasy lemuroids and Afro-Asiatic lorisoids), and the Haplorhini, which unite Anthropoidea or Simiiformes with extant Tarsiidae. The introduction of a variety of fossil families renders these notions problematic, as will be seen below. The term “prosimians” will also be used, to designate all non-anthropoidean primates.

Because there are many uncertainties and debates in primate phylogeny, an important conceptual choice is made in this chapter: the adoption of character-based definitions of higher taxa. This choice serves both clarity and stability, as will become apparent on several occasions below. It is usually well recognized that a crown group definition of taxa (all species descended from the common ancestor of living forms) has inconveniences: if many members of a group are extinct, the crown group is too small. Williams and Kay (1995) illustrated this with examples like the genus Homo, which cannot be limited to H. sapiens. One might add that the well-known Aegyptopithecus is usually considered a catarrhine, although it probably does not pertain to the crown group. Concerning the order Primates, a character-based definition is provided above. There might be several clades bearing the five primate synapomorphies listed above which would be branches preceding the strepsirhine/haplorhine dichotomy. It would not be reasonable to exclude such groups from the order Primates because they branch before the dichotomy leading to the living groups. Different but equally unfortunate inconveniences arise from the use of stem-based definitions. Too often, the node separating two higher taxon stem lineages is unknown, so that fossils can be moved from one higher taxon to another and back, depending on cladistic analyses whose results often appear highly unstable. And if the node were known, i.e., the genus in which the initial speciation took place, stem-based definitions would lead to a spread between different higher taxa of species of one and the same stem genus. Any such definition of a higher taxon becomes minuscule in terms of characters, i.e., almost useless. All these inconveniences, bound up with definitions that proceed either from living in-group survivors or from living sister groups, are avoided by the use of character-based definitions. Furthermore, in primate origins and in primate evolution, a number of important anatomical innovations occurred, and instrumental use of these in systematics helps to convey this critical information, instead of ignoring it.

The fossil record is presented here via the well-documented major radiations since the Early Eocene: Adapiformes and Omomyiformes, followed by their close kin Lemuriformes and Tarsiiformes, respectively. Several enigmatic groups are mentioned next, before Simiiformes (=Anthropoidea) and anthropoidean origins are discussed. The following rough size categories will be used: “very small” for primates weighing 100 g or less, “small” for primates between 100 and 500 g, “middle-sized” for primates between 500 g and 2 kg, and “large” above 2 kg. Weight is a very important adaptive and ecological factor; however, the estimation of body weight in fossils is not easy. Body weight estimates will be given only for fossils for which weight indications from dental remains can be supplemented by estimates based on cranial or postcranial remains.

The Adapiformes Radiations

The Adapiformes are an almost cosmopolitan group of fossil primates which diversified during the Eocene and survived in Asia until the Late Miocene. They include around 46 genera and 105 species. They provide some of the best-known fossil primates. In terms of size, dental adaptation, and locomotion, they compare relatively closely with living lemurs. Hence they are usually described as lemur-like, and some cranial and postcranial characters seem to justify their grouping with living Lemuriformes as strepsirhines, a notion analyzed below.

The Adapiformes divide into three families: the Notharctidae, Adapidae and Sivaladapidae.

North American Notharctids

The Notharctidae are the oldest documented adapiforms, being represented in the Earliest Eocene by the genus Cantius in North America and Europe, and by the genus Donrussellia in Europe only. From these two stem genera, two subfamilies diversify: the Notharctinae mainly in North America (plus Cantius surviving several My in Europe), and the Cercamoniinae in Europe. In North America, the history of the subfamily is well known. There are around 10 species of Cantius in North America. The oldest, C. torresi and C. ralstoni, as well as the European C. eppsi, show the typical dental characters of the genus (Fig. 2). The dentition is on the whole moderately bunodont, having somewhat rounded molar cusps. Small incisors, a large canine, and four premolars are typical primitive primate characters. P/1 to P/3 are simple, with one main cusp increasing in size and the posterior part broadening from P/1 to P/3. P/4 is still longer, has a well-formed metaconid, and a broad and short talonid with a small median hypoconid cusp. The lower molars increase in length from M/1 to M/3, whereas the trigonid becomes anteroposteriorly shorter. There is a large lingual paraconid well separated from the metaconid on M/1, and the paraconid merges progressively with the metaconid summit on M/2 and M/3. The trigonid is transversely broad and opens posteriorly on M/3, which is typical of Cantius and advanced over Donrussellia. M/3 has an elongated third lobe, and its triangular outline, narrowing posteriorly, is also typical of the genus. P3/ and P4/ have massive outlines. The protocone lobe is slightly narrower than the labial part on P4/, and much narrower and smaller on P3/. The M2/ is always more transversely elongated than M1/, which is slightly anteroposteriorly longer. Both have small conules, a postmetaconule-crista going toward the base of the metacone, and a distinct protocone fold.

Fig. 2

Upper and lower jugal teeth of Cantius eppsi showing the primitive dental morphology of notharctines. (c), upper series from P3/ to M3/; (d) lower series from P/3 to M/3. The upper M1/ in (a) and the lower M/2 in (b) illustrate most of the terms of dental morphology used in the text. Upper molars have three main cusps and no hypocone. Lower M/1 and M/2 have large paraconids, and P/4 has a main cusp and is relatively elongated (photographs of epoxy casts)

A lineage Cantius–Notharctus is documented in the Bighorn Basin of Wyoming by successive assemblages that can be placed in a synthetic stratigraphic sequence. The assemblages are so close in time and in morphology that this provides one of the most convincing cases of gradual evolution in mammals (Gingerich 1976). This lineage includes the successive species C. ralstoni, C. mckennai, C. trigonodus, C. abditus, and C. nunienus. These species have no precise boundaries, but the distinction between them is necessary to convey the underlying evolutionary information. The lineage is characterized not only by the size increase evident on diagrams, but also by the progressive development of new dental characters, including an entoconid notch on the lower molars, a hypocone on the upper molars, as well as a mesostyle which will progressively lead to W-shaped labial crests as found in several living folivorous primates (Gingerich and Simons 1977). The hypocone grows on the protocone fold, not from the cingulum as in most other primates. It is thus often referred to as a pseudohypocone. Cantius had unfused lower jaws, anteriorly appressed; some later Notharctus have frequently fused lower jaws, with a long horizontal symphysis. Variations in the size and orientation of the anterior lower incisors among Cantius and Notharctus species are illustrated by Rose et al. (1999). The lineage keeps four premolars all along. Among other species, C. simonsi is the largest one, found in Late Wasatchian beds (Wa7) of the Bighorn basin; C. angulatus is found in the San Juan Basin of New Mexico; and C. frugivorus is found in the latter and in several other basins (Beard 1988; Gunnell 2002).

A crushed skull of Cantius abditus appears relatively short and broad in its proportions. It displays characters that are interpreted as primitive in notharctids, such as a ventrally keeled basioccipital, a flat basisphenoid, and proximally broadening nasals (Rose et al. 1999). The auditory region is very similar to that of Notharctus and living lemurs. It shows a large stapedial artery and a much smaller promontory artery, which seems to have followed an open sulcus along the promontorium. This morphology, shared with lemurs, may be primitive in adapiforms, and the bony tubes present in Notharctus and Smilodectes may be derived.

A partial skeleton of Cantius trigonodus includes a fragment of a distal humerus with a prominent brachialis flange, a proximal ulna with a well-developed olecranon process, and a proximal and distal radius showing a strong shaft (Rose and Walker 1985). A partial pelvis shows a relatively long ischium and short ilium compared with living lemurs, probably primitive. The femoral head is less spherical than in living lemurs, and relatively far separated from the greater trochanter. The distum of the femur is anteroposteriorly high, with a prominent lateral ridge, suggesting leaping abilities. The proximal tibia shows retroflexed condyles and a prominent tibial crest. The relatively distal tibial tuberosity may indicate quadrupedalism and also be primitive, associated with the long ischium. An astragalus, a calcaneum and an entocuneiform of C. trigonodus were figured by Matthew in Matthew and Granger (1915) (Fig. 3). Adding characters of the astragalus and the prominent peroneal tubercle of a proximal hallucial metatarsal, Rose and Walker (1985) infer for Cantius the locomotor behavior of an active arboreal quadruped with a propensity for leaping. Additional foot bones, cuboid, navicular, medial cuneiform, and a distal nail-bearing phalanx of C. mckennai were described by Covert (1988), who inferred high foot mobility and powerful grasping in this species. Because tarsals are frequently found in general and the Early Eocene fossil record is rich, more than 150 tarsals of Cantius species, and probably some Pelycodus and Copelemur, were identified (Gebo et al. 1991). They show no significant change, whereas there are some small differences between them and later Notharctus and Smilodectes tarsals.

Fig. 3

Foot bones of Cantius frugivorus. (a) calcaneum; (b) astragalus; (c) entocuneiform; all in dorsal view. The distal articulation of the entocuneiform for the first metatarsal is sellar in relation to the opposability of the first pedal digit (hallux) typical of primates (From Matthew in Matthew and Granger 1915)

Species of Notharctus became particularly well known with the publication of a beautiful monograph by Gregory (1920). The reconstruction of the skull of N. tenebrosus by Gregory, based on distorted specimens, was not exact from today’s perspective, with too long a muzzle. This has since been corrected due to the discovery of more complete skulls (Alexander 1994; Godinot 1998), which show a higher and shorter muzzle (Fig. 4). Partial skeletons allowed Gregory (1920) to describe long bones, a partial scapula and pelvis, and foot and hand bones. He could conclude the analysis of Notharctus’s locomotor behavior by inferring a clear leaping adaptation, albeit one less extreme than in living vertical clingers and leapers (VCL): in Notharctus the forearm still played a more important role in locomotion, muscular insertions on the pelvis were less pronounced, and feet were less specialized than in living indriids. More recent studies have generally concurred with Gregory, with nuances emphasizing either Notharctus’s leaping propensity or, in contrast, its quadrupedal postures. It can be viewed as a frequent leaper, though probably a pronograde (horizontal) leaper and quadruped rather than a VCL. Gregory had reconstructed an incomplete and very long hand. A more complete specimen allowed the description of all carpals, metacarpals, and phalanges, and a better reconstruction of the hand (Hamrick and Alexander 1996). This hand has remarkably long digits due to very long proximal phalanges. The thumb is divergent, and the short second metacarpal appears markedly divergent from the third. The second digit is reconstructed unreduced and placed in the middle of the space between digits one and three. This results in a functionally bizarre hand, which furthermore bears a strange and unique distal phalanx on the second digit. The divergence of the metacarpals and length of the digits give rise to the suspicion that this hand might have used schizodactylous grips, with the support between digits two and three as in living Alouatta.

Fig. 4

Skull of Notharctus tenebrosus in profile view (a), mandibles in occlusal view (b), and cranium in ventral (c) and dorsal (d) views. The dorsal view shows some deformation of the left zygomatic arcade, which may have a slightly exaggerated orbit height on the profile view. The mandibles have a long, inclined, fused symphysis

There are five species of Notharctus. N. venticolus is the earliest one, known from the latest Wasatchian and earliest Bridgerian, and in temporal and morphological continuity with Cantius abditus. An assemblage of this species has clearly shown a strong sexual dimorphism in upper canine size and shape, as admitted by earlier authors from more scattered specimens (Krishtalka et al. 1990). The species is followed by N. robinsoni (BR1B) and N. tenebrosus (BR2). Skull shape differences between males and females in N. tenebrosus have been described by Alexander (1994), and beautiful skull illustrations have been produced (Alexander and Burger 2001). N. robustior is the largest species of the genus, which had a body weight estimated between 6 and 8 kg.

Another well-known Middle Eocene genus is Smilodectes. The best-known species is S. gracilis (Middle Bridgerian), whose skull and endocranial cast were described by Gazin (1958, 1965). It has a slightly shorter and higher cranium than Notharctus. Its skeleton is also well known, but was not studied in as much detail as that of Notharctus. The two are in fact broadly similar. Smilodectes too had a pronograde and leaping adaptation. It also developed the shearing adaptation of the upper molars (W of labial crests), which produced similar upper molar characters. Such similarities can result in the two genera appearing as sister groups in a cladistic analysis (Covert 1990). However, Smilodectes has a long ventrolingually sloping paralophid on the lower molars, which indicates a different lineage, not derivable from C. trigonodus, and convergence with Notharctus in upper molar characters. The same result is obtained after careful evaluation of the characters sustaining different trees obtained by cladistic analysis, leading researchers to favor the close relationship of Smilodectes and Copelemur – included in a tribe Copelemurini — and as a consequence postulating convergences between Smilodectes and Notharctus (Beard 1988; Gunnell 2002). There are two other species of Smilodectes besides S. gracilis: S. mcgrewi is Early Bridgerian, and S. sororis is earliest Bridgerian, both of Wyoming.

Species of Copelemur are characterized by low and anteroposteriorly elongated P/2–4, lower molars with narrow trigonids, a small ventrally placed paraconid or paralophid, and an anteriorly shifted entoconid linked to the trigonid by a long postmetacristid and followed by an entoconid notch; upper molars have a mesostyle, and a protocone fold but no hypocone (Gingerich and Simons 1977; Beard 1988). There are three species of Copelemur, starting with C. praetutus from the Middle and Late Wasatchian (Wa4-6) of Wyoming and Colorado. It is followed by C. australotutus (Wa5-6) and C. tutus (Wa7), the latter of which was the first discovered, by Cope, in the San Juan Basin of New Mexico. Copelemur is considered closely related to Smilodectes; however, the origin of neither taxon is clear, probably involving southern and less well-sampled regions.

Two other genera are recognized in the North American Eocene, Pelycodus and Hesperolemur. Pelycodus is more bunodont than other notharctines. P. jarrovii is found in the late Wasatchian (Wa6) of New Mexico and rarely shows up in Wyoming. P. danielsae, known only by two fragmentary specimens, is a very large species with an estimated weight of above 6 kg (Froehlich and Lucas 1991). The origin of Pelycodus probably lay in the poorly documented southern regions of the United States. A rooting in the earliest Wasatchian Cantius torresi seems possible. The case of Hesperolemur is more complex. It was named based on a cranium and two other specimens from the Uintan of California (Gunnell 1995b), but the interpretation of some of its cranial characters was later criticized by Rose et al. (1999), who proposed to refer it to a species of Cantius. However, it differs from the most derived species of Cantius and appears somewhat reminiscent of Pelycodus, albeit distinct from it due to its weak mesostyle. Gunnell (2002) still considers it to be a valid genus. North American notharctines became extinct in Wyoming at the end of the Bridgerian, but Hesperolemur found refuge in Southern California in the Early Uintan. Their extinction is related to the shrinking of forested areas, due to the surrection of the Rocky Mountain Range and drying of the areas on its eastern side.

European Notharctids

The European notharctid radiation also starts in the Earliest Eocene, represented by a primitive species of Cantius, with survivors during the Early Eocene, and by species of Donrussellia, more primitive than Cantius, that seem to be at the root of the European Cercamoniinae. D. provincialis from Rians, southern France, shows the most primitive adapiform dentition known today (Fig. 5). The anterior part of the dentary is thin and presents a very anteriorly inclined symphyseal region. Alveoli suggest small incisors (probably two), a large canine, a single-rooted P/1, and a double-rooted P/2. P/3 and P/4 are elongated and narrow, P/4 bears an incipient metaconid. A slight crowding of the premolars is shown by some inclination of the roots of P/2 and P/3, the anterior one being more labial than the posterior one. M/1 has a big lingual paraconid well separated from the metaconid; its trigonid is longer than its talonid. The trigonid of the lower molars decreases in size from M/1 to M/3, the paraconid becoming smaller and closer to the metaconid. The talonid basin, in contrast, increases from M/1 to M/3, which has a relatively expanded third lobe. The upper molars are transverse and simple, without any trace of lingual cingulum or hypocone. The postprotocrista starts with a short posterior inclination, being continuous on M2/ and interrupted on some M1/, but there is no real protocone fold. The conules are small and there is no postmetaconule-crista. Other species of Donrussellia are known: D. lusitanica in Silveirinha, Portugal (Estravis 2000), the small and more insectivore-like D. gallica from Avenay, Paris Basin, and the larger D. magna from Palette, southern France, which appears intermediate with Cantius and thus suggests that at least three species dispersed into Europe in the Earliest Eocene.

Fig. 5

Dentition of Donrussellia provincialis. Upper teeth, P4/ to M3/ (a), one mandible showing P/3 to M/1 (b), and one with M/1 to M/3 (c). The upper teeth are among the most primitive known in primates, with three main cusps and no hypocone. Lower molars also show a primitive large trigonid on M/1-2 with a well-formed paraconid. P/3 and P/4 are simple and relatively narrow. Alveoli reveal a relatively large canine followed by a single-rooted P/1 and a double-rooted P/2

In the late Early Eocene , larger and more derived species occur, which can be associated in groups of related genera. The first group includes Protoadapis, Europolemur, and Barnesia. Species of Protoadapis appear less common now than previously thought. The earliest species seems to be P. curvicuspidens, named long ago by Lemoine from late Early Eocene sites which are not precisely located. The species is present but rare in Grauves, the MP 10 reference locality. In Protoadapis, the paraconid is a residual cuspule on a relatively long and subhorizontal paralophid. The genus is also characterized by a P/3 higher than P/4. Other similar-sized species of Protoadapis are P. angustidens, from an unknown level in the Quercy, and P. ignoratus and P. muechelnensis, from the Middle Eocene lignite mines of the Geiseltal, Germany (MP 12, Thalmann 1994). Two larger species of Protoadapis are also described from lower jaws alone. P. (Cercamonius) brachyrhynchus is also from an unknown level of the Quercy fissure fillings, whereas P. weigelti is from the same level in the Geiseltal as the above-mentioned species. These species have more robust teeth and jaws. Robustness of the jaw and possible anterior premolar reduction and more vertical symphysis led Gingerich (1975) to erect the new genus “Cercamonius” for the species brachyrhynchus, considered to be evolving toward anthropoid characters. However, Thalmann (1994) followed Szalay and Delson (1979) in rejecting the new genus, noting only small differences between P. brachyrhynchus and P. weigelti. More material is necessary to ascertain the differences between these species, and Cercamonius can be retained as a subgenus to mark its distinctness. Species of Protoadapis remain poorly known, with only a few upper teeth referred to P. curvicuspidens, among which one figured specimen is misidentified (Russell et al. 1967).

Species of Europolemur are known through partial skeletons found in the Middle Eocene localities of Messel and the Geiseltal in Germany, and dental material from Eckfeld, Bouxwiller, and probably also the Paris Basin in France. The Messel Oil Shale deposited in a volcanic lake dated close to 47 Ma. The skeleton of Europolemur koenigswaldi was found there in two parts: the anterior part, crushed skull, forelimb, and trunk were found first (Franzen 1987); the rear part was found later, allowing the reconstruction of the entire animal (Franzen and Frey 1993). It was relatively small. The first weight estimate from dental regressions was close to 2 kg; however, a more reliable estimate from trunk length suggests a much lower weight, around 300 g, underlining the uncertainties associated with weight estimates from tooth dimensions alone. Crushed skeletal parts are difficult to study, but they show parts that are most often unknown in fossils. The anterior part is that of a young adult, with erupting M/3 and still functional DP/3–4/. The muzzle appears elongated, and the small orbits suggest a diurnal way of life. Incisors are in place, recalling those of Notharctus, with I1/ two times broader than I2/. From the breakage not being in midline, Franzen (1987) deduced that the two mandibles were fused. X-ray images show the outline of middle ear ossicles within the right tympanic bulla, as well as parts of a tympanic ring. The vertebral column is preserved from the axis to the os sacrum, the latter consisting only of two fused sacral vertebrae. Ribs 1–13 can be seen. A right forearm is there, from scapula to hand. The scapula, rarely found in fossils, has a shape close to that of Galago and Eulemur macaco. The broad humerus is broken. The hand shows flexed digits, digit one isolated from the others on the palmar side, as well as a large pisiform. Two distal phalanges appear broad, flat, and scutiform. The rear part of another individual, ascribed to the same species, shows a long tail with 30 vertebrae. Proportions of femur and tibia led to a crural index of 83, and an estimation of the intermembral index of 72.6. Among the tarsals, the astragalus shows a wedge-shaped trochlea, and a neck and a posterior trochlear shelf more extended than in Notharctus. The calcaneum is slender, elongated (index of anterior part 45.2 %), and with a short posterior astragalar facet – characters which suggest leaping propensities. The two feet are complete but crushed. Their metatarsals are short, the first one being very robust. The terminal phalanges are quite narrow with a flat plantar side, suggesting the presence of narrow nails (this difference with the anterior part is bizarre, raising a question about their association). There is no special toilet claw. Franzen and Frey (1993) reconstruct E. koenigswaldi as an arboreal quadruped, climbing above branches and having leaping capacities.

Another somewhat larger species is E. kelleri, named from a crushed skull. Several earlier finds were referred to this species: a pelvis, baculum, and hindlimbs, and a forearm with hand (Franzen 1988, 2000a; von Koenigswald 1979). E. klatti from the Geiseltal is the type species of the genus. It was described from a crushed cranium bei Weigelt (1933). Its dentition includes a large laterally compressed upper canine, a small P2/ and a single-cusped P3/, upper molars with complete lingual cingulum and large hypocone, but no metaconule. Details on the variation in dental characters are given by Thalmann (1994), who tentatively ascribed to the species some isolated postcranials, an atlas, an astragalus, and a calcaneum. E. dunaifi from Bouxwiller, known only through isolated teeth, appears very close to E. klatti. A crushed cranium from the highest level of the Geiseltal (MP 14) shows upper teeth close to those of Europolemur, though different enough for Thalmann (1994) to erect a new genus and species, Barnesia hauboldi. Its M2/ is enlarged in comparison with M1/ and M3/. It is transversely elongated, and on its lingual cingulum a pericone is added to a moderately large-sized hypocone. The species is possibly present in Bouxwiller.

The genera Pronycticebus and Godinotia constitute an isolated lineage rooted in primitive forms. It has been known for a long time through the description by Grandidier (1904) of a beautiful cranium and jaw coming from the old phosphate exploitations of the Quercy region, southern France, which he named P. gaudryi. Its age is not precisely known; however, one tooth attributed to a close species was found in a new Quercy locality, dated MP 10–11. This suggests that Pronycticebus is probably a Lutetian genus. The skull of P. gaudryi is important because it is the only cercamoniine skull that is three-dimensionally well preserved. Its tympanic bulla and basicranial foramina are similar to those of lemurs and adapids, and partial preparation of one bulla also shows a free tympanic ring, which is often considered typical of strepsirhines (Le Gros Clark 1934; Simons 1962). A CT-scan analysis revealed the morphology of its bony labyrinth (cochlea and semi-circular canals), which appears closest to that of Adapinae, both having similarities with lemuroids (Lebrun et al. 2012). A relatively complete and somewhat crushed skeleton from an MP 12 level in the Geiseltal lignite series, Germany, was described as “P.” neglectus (Thalmann et al. 1989). A mandible previously ascribed to Europolemur klatti was also referred to this species by Thalmann (1994). The species was made the type of a new genus Godinotia by Franzen (2000b); however, this was based on the misallocation to this species of the first slab of Darwinius (see below). The species seems to be dentally close to P. gaudryi, but it differs from it by the loss of P1 and reduction of P2, which are single-rooted above and below, versus double-rooted in P. gaudryi. This may justify a generic (or subgeneric) distinction. Among the best preserved skeletal elements are a left humerus, a right forearm and hand, with some carpals, five metacarpals, and some very elongated proximal phalanges (Thalmann 1994). Godinotia appears to have more gracile limb bones than Europolemur and Darwinius (Franzen et al. 2009).

A third group of cercamoniines consists of Agerinia, Periconodon, and Darwinius. New faunas under study reveal an abundant new species in the Paris Basin that is close to Agerinia and Periconodon, and also appears closely related to Darwinius (Herbomel and Godinot 2011). Species of Periconodon are named after the presence of a supplementary cusp on the upper molars called a pericone (Stehlin 1916). The taxonomy is uncertain and provisionally follows Godinot (1988), with P. helveticus from Egerkingen as type species of the genus (Stehlin 1916). Its type specimen is a maxilla bearing P3/ and M1–2/. These molars have a transversely short trigon basin, a long lingual slope of the protocone, a well-developed pericone that is lingual or slightly anterolingual to the protocone summit, and a large hypocone. P3/ is high and pointed. Its lingual border is underlined by a moderate bulging, without any cusp or protocone lobe. A very close species is P. huerzeleri from Bouxwiller, the type specimen of which is a mandible (Gingerich 1977a). The lower molars of this species show a high intraspecific variability, e.g., M/1 with or without a tiny paraconid; lower molars with a clear separation between the premetacristid and the curved paralophid, or a paralophid continuous until the metaconid on M/3; and M/3 with a more or less extended talonid basin, with or without distinct entoconid. Upper molars vary in their transverse extension, and in the variable presence of a crest linking hypocone and pericone (Godinot 1988). Similar high variations occur on the teeth of P. jaegeri from a lower level at Bouxwiller, with hypocone and pericone varying from small to very large and lingually bulging. This species is typified by crenulated enamel and lower molars with a supplementary anterolingual crest issued from the metaconid (entometacristid). P. helleri is known through a maxilla from the Geiseltal series (MP 13–14). Its teeth are worn and chemically eroded. It shows a P4/ having a relatively short protocone lobe. A dentary of Periconodon sp. is reported from the rich locality Eckfeld Maar (MP 13). It preserves P/4, P/3, an incomplete and apparently two-rooted P/2, and the alveolus for a P/1 (Franzen 2004). The older P. lemoinei proposed by Gingerich (1977a) needs further study. Until now, all species of Periconodon are known dentally only.

Agerinia is known through its type species A. roselli from Les Saleres, Spain (MP 10), and through another species from southern France. Only lower teeth have been described, which include narrow P/3 and P/4 with curved preprotocristid and lingual cingulum. The lower molars have no more paraconid, and a paralophid joining a premetacristid, thus realizing a complete crest anteriorly joining protoconid and metaconid. The included trigonid basin (or fovea) becomes anteroposteriorly shorter and labiolingually longer from M/1 to M/3. The talonid basin is especially broad, with rounded outline, on M/1 and M/2. It is this distinctive morphology of M/2, with very short trigonid and broad talonid, which makes Darwinius remarkably similar to Agerinia in these rare derived traits. This suggests a close phylogenetic relationship between them, also shared with a new species from the Paris Basin (Herbomel and Godinot 2011).

D. masillae is the name given to the most complete Messel skeleton, which has an unusual history. After it was found by amateurs in the form of two opposite slabs, the first of these, which was incomplete, was supplemented by fake parts and described for the first time as the sixth Messel primate by Franzen (1994). It was at that time referred to “Caenopithecus neglectus.” The preserved parts were already remarkable, including a crushed skull, the anterior part of the trunk and forelimbs up to the middle of the forearms, the rear part of the trunk, anterior part of the tail, and a hindlimb almost to the extremity of tibia and fibula. Outlines of the soft part of the body were visible. The specimen was described in detail. It is a juvenile individual with milk dentition and just erupted M/1, with a reduced DP/2, a short cranium, large orbits, and proportions of humerus, ulna, and tibia different from those of Europolemur. Further dental comparisons led Franzen to refer the species to a new genus as “Godinotia neglecta” (Franzen 2000b). Study of the dark remnants corresponding to gut content allowed the identification of a 3–4 mm seed coat and scattered leaf particles. These, together with the absence of insect cuticle, which is normally well preserved in insectivorous Messel mammals, suggest that the species had a frugivorous-folivorous diet (Franzen and Wilde 2003). Several years later, the second slab from the same individual was made available to science, leading to the in-depth analysis of the most complete fossil primate skeleton, now referred to Darwinius as D. masillae (Franzen et al. 2009) and nicknamed “Ida” (Fig. 6).

Fig. 6

The skeleton of Darwinius masillae from Messel, Germany. (a) the best-preserved plate; (b) detail of the same showing forelimbs and part of the vertebral column and rib cage; (c) detail of right hindlimb with complete foot. Note the presence of fur impressions on the complete skeleton, heavily crushed skull, a callus on the right wrist due to reaction after an accident, probably falling from a tree. The right foot shows a strongly opposable first toe and the dorsal side of the astragalus (Courtesy of J. L. Franzen)

The complete skeleton confirms Darwinius to have possessed a relatively short rostrum, a steep face, large orbits, and a rather large braincase. Among its noticeable characters are the unusually short forelimb, the proximally curved ulna, mesaxonic hand (third ray the longest, primitive) with relatively small and short pollex, short metacarpals and long proximal phalanges, and scutiform nail-bearing distal phalanges. Analysis of its tarsus appears insufficient. Franzen et al. (2009) write that the fibular facet of the astragalus would be steep as in primitive primates (“haplorhines” for them); however, their best figure (idem Fig. 10) seems to expose the astragalus in dorsal view, showing mainly its tibial trochlea. A short ribbon of fibular facet is insufficient to describe its slope. The other parts of the foot – navicular, cuneiforms, and enormous pollex – are lemur-like, and it would be very surprising to find them associated with a laterally primitive astragalus. The fibular facet is steep but strepsirhine-like in closely related species from the Paris Basin (Godinot et al. 2011). Future micro-CT-scan studies will complete the first analyses, which raise some questions (e.g., the articulations between trapezoid and MC-II, and hamate and MC-V, are said to be saddle-shaped, which is bizarre). A radiographic study added to the CT-scan revals the pattern of dental replacement in Ida. The specimen is a juvenile with fully erupted M1, M2 probably erupted but with incomplete roots, M3 not erupted, DP3 and DP4 functional. Comparison with living primates suggests rapid growth, similar to that of “medium fast” primates with a maximum life span of 12–20 years (Schultz 1960; Franzen et al. 2009). It suggests that Ida was a juvenile, weaned and independently feeding, which during its life suffered an accident causing a severe trauma of its right wrist, probably a fall from a tree (Franzen et al. 2012). Ida has no baculum, whereas this bone is large and conspicuous in some Europolemur rear skeletons interpreted as males (von Koenigswald 1979); so Ida probably is a female. Its weight is estimated at 650–900 g, depending on the estimators. Its locomotion is reconstructed from a multivariate analysis as quadrupedal without specialization for climbing or leaping. However, such multivariate approach is likely to be influenced by evolutionary trends (e.g., the long lumbar region is likely primitive). The intermembral index (IMI) is more directly linked to locomotor adaptation. The IMI calculated for Darwinius from its measurements is 63–64, relatively low, indicating that its locomotion included frequent leaping. It is thus more likely to have been an arboreal quadruped leaper. A haplorhine status has been advocated for Darwinius, but only on weak grounds (Franzen et al. 2009). For example, a fused mandibular symphysis evolved many times in primates; loss of a grooming claw is uncertain in Darwinius but such a grooming claw exists in Europolemur and, moreover, reveals a complex pattern of evolution (von Koenigswald et al. 2012; Maiolino et al. 2012); see above concerning the fibular facet of the astragalus. Darwinius is a cercamoniine showing similarities with Pronycticebus (large orbit, large upper canine), as well as with Agerinia and Periconodon in dental morphology, and the placement of adapiforms with the strepsirhines appears well founded (see below).

A series of smaller species and genera of cercamoniines are known, mainly dentally. They include three genera close to Anchomomys, which will be grouped in a restricted tribe Anchomomyini. They lived during a large part of the Middle Eocene. The smallest of them is the type species of Anchomomys, A. gaillardi, known by a maxilla and a mandible from Lissieu (MP 14), described by Stehlin (1916). The upper teeth are very simple, lingually narrow with the three main cusps, a continuous crista obliqua without metaconule, a paraconule, a very small incipient hypocone on M1–2/, and a non-reduced M3/. The lower molars are elongated and narrow. Typical of Anchomomys species is their almost rectilinear anterior paralophid, which in occlusal and in anterior view is at a roughly right angle with the strongly sloping preprotocristid. There is no trace of a paraconid. This thin paralophid, always well separated from the base of the metaconid, is reminiscent of adapines (and of Pronycticebus M/3). The trigonid becomes shorter from M/1 to M/3. Two dentaries from Egerkingen referred to A. cf pygmaeus (taxonomy uncertain) add some information. One shows a P/4 which is anteroposteriorly elongated, narrow, has no trace of a metaconid, and bears only one posterior median crest descending from the protoconid summit, and a well-formed talonid with hypoconid and lingually inclined basin surrounded by a lingual cingulid. The other dentary shows anterior alveoli, which suggest the presence of two small incisors, a relatively large canine, and a two-rooted P/2 (Stehlin 1916; Szalay and Delson 1979).

Other species of Anchomomys include the slightly larger A. (Huerzeleris) quercyi, which has a larger but still small hypocone. The type specimen is from old Quercy collections, without precise location; however, very close specimens have been found in new Quercy faunas from reference levels 16 and 17a. A. frontanyensis from Sant Jaume de Frontanyà (MP 14/15, Spain) completes our knowledge of the genus. Described are a robust, slightly recurved upper canine, small P1/, P2/, and larger P3/ which are subtriangular in outline, surrounded by an almost continuous cingulum. P4/ is transversely elongated with a well-formed protocone. M1–2/ have very small cingular hypocones (Marigo et al. 2011). Among the lower teeth are a robust canine with cingulids descending from the apex, a surprisingly large P/1 (P/2?), P/2, and P/3 oval in outline and surrounded by a continuous cingulid, and an elongated P/4 close to that of A. cf pygmaeus. Among the lower molars, several M/3 show an unusually high variability of their talonids, broad and short or elongated, with variable supplementary cuspules on their periphery. Postcranials of this species are known and allow a weight estimate of 120 g. The astragalus is close to those of small notharctids (Moyà-Solà and Köhler 1993). The first metatarsal has a moderate-sized peroneal process (Roig and Moyà-Solà 2011). The calcaneum has remarkable proportions. Its anterior part is more elongated than in any other known adapiform, rather resembling omomyids in this proportion. However, the authors conclude that such proportions are a compensatory effect in grasping foot postures, and not an indication of leaping propensities (Moyà-Solà et al. 2012). Other postcranials are said to confirm a cheirogaleid-like, generalized type of arboreal locomotion for A. frontanyensis.

Among the species closely related to Anchomomys are two species of Buxella, B. prisca and B. magna, described from the locality of Bouxwiller and known only by isolated teeth (Godinot 1988); Nievesia sossisensis from Sossis, Spain, known by a maxilla with M2–3/ and isolated teeth, is distinct due to the presence on the lower molars of a premetacristid joining the paralophid, closing a sloping trigonid basin (Marigo et al. 2013); and Mazateronodon endemicus. The latter is from the Middle Eocene Spanish locality Mazateron (MP 15–16), in the Almazan Basin. This basin is further west in comparion with other Pyrenean/Catalan basins, and its fauna, including Mazateronodon, reveals some endemism relative to other European localities (Marigo et al. 2010). Like Nievesia, M. endemicus has lower molars with a closed trigonid basin. Two upper incisors referred to this species show an elongated cutting edge reminiscent of adapines. In contrast, a referred lower incisor seems not to fit well with the uppers and would suggest a great amount of evolution in comparison with the alveoli of A. cf pygmaeus. A referred P4/ has a very narrow lingual lobe. One mandible fragment with P/3–4 and the canine shows the latter to be robust and of triangular outline. P/3 and P/4 on this specimen and on another one are noteworthy: elongated, blade-like, slightly crowded with the anterior root more labial than the posterior one. P/3 is longer and higher than P/4, somewhat posterodorsally inclined; P/3 is slightly narrower, more elongated and more inclined on the specimen bearing the canine than on the other. On both, P/3 displays a long anterior blade, suggesting a possible honing mechanism for the robust upper canine (which is also described). This should be explored: it would be remarkable for such a small species. In any case, M. endemicus shows an interesting evolution of the lower premolars, possibly convergent with catarrhines. Middle Eocene anchomomyins survived longer than other cercamoniines, probably because their small size induced them to avoid competition with the then-dominating adapines. They must have been partly insectivorous.

Adapids in Europe and Three Other Continents

The family Adapidae is known by a classical group of European forms, the Adapinae, and by a series of more primitive and almost cosmopolitan forms (Fig. 7), provisionally grouped in the Caenopithecinae. The adapines arrive as several lineages, Microadapis and Leptadapis (including Paradapis), between reference levels 13 and 14. Until now, it has not been possible to root them convincingly in European cercamoniines, despite some similarities. Surprisingly, a species with marked similarities with adapids has been found in China, suggesting an origin of the subfamily in some intermediate area and strengthening the notion of a dispersal into Europe. Adapines make a small radiation in the Late Eocene of Europe, documented by the beautiful skulls found in the Quercy fissure fillings in the nineteenth century. A detailed description and analysis of these skulls was provided by Stehlin (1912). Information on one beautifully preserved cranium of Adapis and its endocast is given by Gingerich and Martin (1981). A careful revision of the large-sized adapines more recently has led to the distinction of two genera: Leptadapis, with a narrow interorbital breadth and muzzle (three species), and Magnadapis, with broader interorbital breadth and muzzle (four species) (Fig. 8). In both genera, some species show large or extreme cranial superstructures (sagittal and nuchal crests, thickened zygomatic arcades), while others show small ones or none. Whereas one study had proposed to interpret these differences as a marked sexual dimorphism (Gingerich 1981), restudy of more material, which underlined modest differences in canine sizes and morphological changes in dental morphology, led to the hypothesis of different lineages, some of them characterized now by a marked increase in cranial superstructures (Godinot and Couette 2008). A similar systematic revision is needed for the smaller Adapis-sized species, which are also numerous (Lanèque 1992, 1993). The postcranials of adapines are distinctive, showing no hindlimb lengthening and calcanea which have a very short anterior part, shorter than in any living strepsirhine (Dagosto 1983). The sole known complete femur has a relatively broad distal end (Fig. 9). These characters are interpreted as indicative of climbing and a relatively slow locomotion (Dagosto 1983). However, some Adapis-sized species appear to have varied in locomotor modes, one being more walking and running than the other (Godinot 1992a; Fig. 9). The variety of femora that can be attributed to Adapis-sized species show that they underwent a diversification in their locomotor adaptations, resulting in five different types, some of which were more climbing and others more walking and running forms (Bacon and Godinot 1998).

All of these adapines have highly crested molars and molarized P/4. Their phylogeny is rather complex and not yet understood. Species of Cryptadapis had a large hypocone, whereas large adapines show a progressive reduction of this cusp. Several lineages further increase their shearing adaptation in developing a metastylid on their lower molars. They appear adapted to shearing food, which is suggestive of degrees of folivory; however, they varied in their diets. Microwear analysis suggests that Cryptadapis tertius was strictly folivorous, whereas one Adapis species added more fruits to its diet than two other adapines (Ramdarshan et al. 2011). Adapine remains from well-dated faunas reveal that the large genera, Leptadapis and Magnadapis, preceded the mid-sized ones, and that the latter became extinct in the Quercy region at the end of the Eocene. The adapines apparently did not survive the Terminal Eocene Event, or “Grande Coupure,” in Europe, which involved an invasion of Asiatic mammals better adapted to more open environments, as well as to cooler and more seasonal climates. However, it seems that one species of a relatively large adapine briefly survived the dispersal event in England (Hooker 2010).

Fig. 7

Known stratigraphic ranges and suggested phylogenetic relationships among genera of Adapiformes. Dashed bars indicate genera including more than one specific lineage; black bars indicate that only one specific lineage is known; simple vertical lines indicate uncertainty in the stratigraphic range. The global picture of this record is a good reflection of the whole Eocene primate record: relatively well known in Europe and North America, much less in Africa. In Asia, which played a central role, several groups are documented; however, the record is too scanty to allow even a vague schema of possible relationships. Dashed lines indicate reasonable phylogenetic hypotheses; dotted lines indicate unknown origins and affinities for Asiatic groups. Arrows indicate dispersals between continents (or the continuation of sivaladapines in the Miocene above 33). 1, new species close to Periconodon and Agerinia; 2, Periconodon; 3, Agerinia; 4, Darwinius; 5, Pronycticebus; 6, Godinotia; 7, Donrussellia; 8, Anchomomys; 9, Buxella; 10, Mazateronodon; 11, Nievesia; 12, Protoadapis; 13, Europolemur; 14, Barnesia; 15, Cantius; 16, Caenopithecus; 17, Leptadapis; 18, Magnadapis; 19, Cryptadapis; 20, Microadapis; 21, Adapis; 22, Paleolemur; 23, Aframonius; 24, Afradapis; 25, Adapoides; 26, Marcgodinotius; 27, Asiadapis; 28, Panobius; 29, Lushius; 30, Rencunius; 31, Hoanghonius; 32, Wailekia; 33, Guangxilemur; 34, Bugtilemur; 35, Muangthanhinius; 36, Paukkaungia; 37, Kyitchaungia; 38, Pelycodus; 39, Mescalerolemur; 40, Mahgarita; 41, Hesperolemur; 42, Notharctus; 43, Smilodectes; 44, Copelemur

Fig. 8

Three crania of Late Eocene large adapines coming from old Quercy collections. (a) is the type specimen of Magnadapis intermedius, (b) is the type specimen of Leptadapis magnus, and (c) is the type specimen of L. filholi. Note that the interorbital breadth and the muzzle are larger in (a) than in (b) and (c). Cranial superstructures are also more developed in (a) in comparison with the two others; however, specimens with very low or no sagittal crest exist in both genera

Fig. 9

Limb bones of Adapis-sized adapines from the old and new Quercy collections. The humerus (a), femur (b), and tibia (c) may come from the same fissure : They have the same type of preservation and were described together by Filhol. The calcaneum (d) and the first metatarsal (g) also come from old Quercy collections, whereas the two astragali come from two new localities, Rosières 2 (e) and Escamps (f). Proportions of humerus and femur suggest a not-elongated hind limb. The anterior part of the calcaneum is particularly short, and the peroneal tubercle of the first metatarsal is small. The two astragali are quite different, (e) with a much longer trochlea and (f) with a shorter, broader, and flatter one reflecting different locomotor adaptations in different Adapis-sized species

The more primitive adapids provisionally grouped in the subfamily Caenopithecinae are found in Europe, North America, Africa, and possibly Asia. The Asiatic Adapoides troglodytes, from the Middle Eocene Shanghuang fissure fillings, China, is known by a lower dentary with M/2–3 and isolated upper molars and DP4/ (Beard et al. 1994). Its lower molars are so derived in an adapid direction, with anteroposteriorly compressed trigonid, long ventrolingually sloping paracristid, and deep talonid notch, that they could point toward adapine affinities. However, the upper molars (including the “Europolemur-like” molar of Beard et al. 1994) are transversely elongated and show no hypocone or only a tiny incipient one, characters markedly more primitive than in the adapines. Without knowledge of the P4/4 and their degree of molarization, it will be difficult to determine the place of this genus relative to the adapines. A small adapiform astragalus from Shanghuang appears already adapine-like, suggesting climbing propensities (Gebo et al. 2001). The European primitive adapids include Caenopithecus and Microadapis, which arrive in Europe at the same time (Egerkingen fissure fillings, Middle Eocene, Switzerland). C. lemuroides shows a number of similarities with later adapines, but it also has non-molarized P/4/ –a primitive trait – and a metastylid on the lower molars and a mesostyle on the uppers, the latter being a derived character unknown in adapines and otherwise found, among adapids, only in the African Afradapis. Microadapis is small, has no metastylid, a simple premolariform P/4 and relatively narrow and elongated P/4–2, and a moderate-sized P/1. A referred upper molar bears a very large hypocone, as well as a metaconule – a small cusp usually lost in other genera, including Adapoides. The most primitive adapids probably differentiated somewhere on the Eurasiatic landmass between Europe and China. Marcgodinotius from India (see below) might have some relevance.

The two African genera Aframonius and Afradapis from the Late Eocene deposits of the Fayum, Egypt, also appear to show affinities with this group, revealing its dispersal into Africa. Afradapis longicristatus is from the Late Eocene BQ-2 locality (Seiffert et al. 2009), whereas Aframonius dieides is from the Latest Eocene locality L-41 (Simons et al. 1995; Simons and Miller 1997). The two genera have upper molars with a broad trigon basin, high crests, and a large hypocone linked to the posterior cingulum (Fig. 10). The long and sharp centrocrista is slightly labially deflected on the M2/ and M3/ of Aframonius, and even more deflected and joining a mesostyle on the M1–2/of Afradapis. Their P4/ are simple with a broad protocone lobe, and their P3/ are triangular due to a narrower protocone lobe. Both have lower molars without paraconids. Those of Aframonius have shorter paracristids, and metaconids posterior relative to the protoconid. Those of Afradapis have a more anterior metaconid, a longer protocristid, and longer paracristids which increasingly curve toward the metaconid summit from M/1 to M/3. The P/3–4 of Aframonius are relatively short, posteriorly broad, and only incipiently molarized. The P/3 and P/4 of Afradapis are remarkable and unique among adapiforms. P/4 is elongated, molarized through a high metaconid almost as anterior as the protoconid, has a well-formed and narrow talonid basin, and is unusual because of its anteriorly elongated protocristid. P/3 is even larger than P/4, higher and much longer. Its anteriorly elongated protocristid served as a honing device for the large upper canine. Lower dentaries show that there was no P/2 in Afradapis, which appears convergent with catarrhines in the possession of only two premolars and an enlarged P/3 honing with the upper canine. The dentaries of Afradapis were fused, whereas there was variability in this character in Aframonius, which retains a moderate-sized P/2. Both upper and lower isolated incisors have been ascribed to Afradapis. The lowers make a transverse cropping mechanism analogous to that of Adapis. A large phylogenetic analysis proposed a placement of the two African genera close to the European Caenopithecus (Seiffert et al. 2009). Given that Aframonius had some more primitive characters, the two groups likely share a close common ancestor, whose descendents were able to disperse to Europe and to North Africa in the Middle Eocene.

Fig. 10

The dentition of Aframonius dieides from the Fayum, Egypt. (a) upper dentition from the large M3/to the small P2/in occlusal view; P3/and P4/are in a more labial orientation, with the protocone partly hidden, due to specimen deformation; (b) lower dentition from M/3 to the lower canine in occlusal view. Scanning electron micrographs (SEM) of casts

Finally, there are two North American genera, Mahgarita and Mescalerolemur, which are of uncertain affinities. Mahgarita stevensi, from the Latest Eocene of Texas, has long been recognized to have no affinity with the North American notharctines (Wilson and Szalay 1976). Similarities with primitive adapines led to its inclusion in that group (Godinot 1998). However, the description of the older and more primitive Mescalerolemur horneri and more detailed phylogenetic analyses showed the adapid features of Mahgarita to be probably convergent, and both these genera to be probably rooted in more primitive adapiforms (Kirk and Williams 2011). A dispersal from Asia, as for many other Eocene forms, appears likely. Both genera have upper molars with a well-formed crestiform hypocone and complete lingual cingulum on M1–2/. M2/ is transversely broader than M1/. The M1/ of Mescalerolemur is more triangular, with a strong prehypocone-crista and an expanded basin posterior to the postprotocrista. Whereas Mahgarita has a P4/ with a broad protocone recalling adapines, that of Mescalerolemur has a much narrower protocone lobe. P3/ is triangular, dissymmetrical due to its posterior protocone lobe, in Mahgarita. It is even more unusual in Mescalerolemur, with a pinched posterior protocone lobe underlined by a strong posterior ectoflexus. Both taxa have lost P1 and have reduced P2 above and below. The lower molars of both genera have very reduced paracristids. Mescalerolemur has a small anterior paraconid on M/1 only. The posteriorly broad and elongated M/3 of Mahgarita could be reminiscent of adapines; however, similar morphologies occur in cercamoniines as well, and Mescalerolemur has a shorter and posteriorly narrower M/3. P/3 and P/4 appear particularly simple in both genera, transversely narrow and with no metaconid or paraconid. In Mahgarita, where the large canines are known, the lower canine is posteriorly recurved, but the upper canine – straight, with a strong basal cingulum, a sharp posterior crest, and vertical grooves – is somewhat reminiscent of Leptadapis. Mescalerolemur has an unfused symphysis, whereas Mahgarita has a fused one – evidently one more convergence with adapines and with many other primates. These two genera have an overall similarity and very peculiar P3–4/ which demonstrate their close affinity. Mescalerolemur is lacking clear derived similarities with adapids, and its origin will be in more primitive Asiatic adapiforms.

The holotype of Mahgarita stevensi is a crushed cranium which shows a number of features: a high maxilla, a pronounced posterior palatal spine, a very low occipital height, and a posteriorly directed foramen magnum. The basicranial morphology was earlier considered essentially lemur-like (Wilson and Szalay 1976). Further study of this crushed cranium and two other partial crania led Rasmussen (1990) to add a number of observations. For example, there is noticeable petromastoid pneumatization; one specimen has a much reduced stapedial canal and the other has none at all. Rasmussen also suggested that Mahgarita had a lateral transverse septum resembling that of Aegyptopithecus, that there was an ectotympanic band fused to the petrosal, and probably not enough space for a free ectotympanic ring. However, restudy of these structures led others to refute these suggestions and reaffirm that Mahgarita was adapid-like in tympanic position and morphology (Ross 1994).

Asiatic Sivaladapids and Primitive Adapiforms

The family Sivaladapidae is restricted to Asia, where the sivaladapines survived until the Late Miocene. The subfamily Hoanghoniinae is naturally placed near them. Here will also be placed the recently described Early Eocene Asiadapini, which are probably related to them, and other more divergent Asiatic adapiforms. This radiation is, on the whole, poorly documented: no skull has yet been described. The best-known members of the group were until recently the Miocene sivaladapines, for which upper and lower jaws allow a description of their dental adaptations. Very high-crested molars and highly molarized P4/4 indicate a folivorous adaptation. A similar adaptation seems to be present in the Late Eocene–Early Oligocene genus Guangxilemur. First described through G. tongi, known by one M2/ and one upper canine from the Gongkang Formation of Guangxi Province, China (Qi and Beard 1998), its analysis was completed by the description of the Early Oligocene G. singsilai from Pakistan (Marivaux et al. 2002). The M2/ of G. tongi has high crests, a mesostyle linked to the centrocrista contributing to W-shaped labial crests, and pre- and postprotocristae as divergent as in the Miocene sivaladapines. The main difference with the latter is its possession of large hypocone and pericone. G. singsilai differs by a smaller hypocone and pericone, and a straight centrocrista on M2/. It also shows a highly molarized P4/, a DP4/, a simple P/3 with very small and narrow talonid, and an M/1–2 with salient crests, broad paracristid not joining the metaconid summit, and big, close, and deeply separated entoconid and hypoconulid. This morphology relates Guangxilemur to Hoanghonius stehlini and its close relative Rencunius zhoui, both from Late Eocene beds of the Heti Formation, Shanxi Province, China (Gingerich et al. 1994). Both species bear the hallmarks of this group: lower molars with twinned entoconid and hypoconulid, and upper molars with a continuous lingual cingulum bearing hypocone and pericone and a wide trigon basin limited by very divergent protocristae. On the mandible of Hoanghonius, which bears M/2 and M/3, a continuous paracristid joins the metaconid anteriorly. On the mandible of Rencunius, which bears M/1–2 and P/4, the molars are more bunodont, M/1 bears a well-formed paraconid, and P/4 is simple with a small metaconid, very short rounded talonid shelf, and continuous cingulids. The M1/ of Rencunius resembles the isolated upper molar of Hoanghonius but differs in some details, such as a more lingually bulging pericone, larger conules, and less waisting of the posterior border. The associated P4/ of Rencunius is very simple, with a large paracone, smaller protocone, and well-formed preprotocrista. Another sivaladapid very close to the above-mentioned genera is Wailaikia, from the Late Eocene Krabi mine of Thailand. W. orientale is represented by a mandible bearing M/2 and M/3 (Ducrocq et al. 1995). Alveoli for the anterior teeth show that it possessed a long premolar series with an unreduced two-rooted P/2 and a relatively large canine. The continuous and transversely long paracristid and the twinned entoconid and hypoconulid leave no doubt that this species is a sivaladapid close to Hoanghonius. It differs from the latter by its broader lower molars and much shorter M/3 with a barely salient third lobe. The dentary is low and elongated, and it preserves a high coronoid process and the articular condyle.

The bizarre maxilla known under the name of Lushius quinlinensis, from the late Middle Eocene of Shanxi, might pertain to the same group. Considered an adapiform incertae sedis by Szalay and Delson (1979), its M2/ bears similarities with that of Hoanghonius, both in outline – the posterior half being narrower than the anterior one – and in trajectory of postprotocrista. It is more primitive with respect to its incomplete lingual cingulum, and more derived with respect to its extremely high paracone, metacone, and crests joining them. These constitute a high ectoloph recalling some ungulates and a very unusual morphology for primates. However, this specialization goes in a direction analogous to many high-crested sivaladapids, and thus its least unlikely affinities are probably with sivaladapids.

The Early Eocene beds of the Vastan Mine, Cambay Formation, Gujarat, India, are dated around 53 Ma. They have been yielding new adapiforms in the last years. Marcgodinotius indicus was named by Bajpai et al. (2005), Asiadapis cambayensis by Rose et al. (2007), and a synthesis on these taxa is provided by Rose et al. (2009). Both are small and primitive adapiforms, with global phenetic similarity with early European cercamoniines, but they also have interesting differences. The dentaries of M. indicus are elongated and slender and show a large canine, small P/1, two-rooted P/2, and some compression of the premolar series leading to an anterolingual shift of their root pairs. P/3–4 are simple; P/3 is higher than P/4, and the latter bears a small, low, little-differentiated metaconid. The lower molars have simple talonid basins. M/1 has a large paraconid in anterior (not lingual) position. M/2 has a shelf-like paracristid, and M/3 has a short trigonid with narrow and rounded paracristid, and also a very small and narrow third lobe. M1/ is remarkable for its overall triangular outline, with a lingual part much narrower anteroposteriorly than its labial part. It also has a strong parastyle, a posterolabially directed postmetacrista, a well-expressed posterior cingulum making a small hypocone shelf without cusp, and a very marked waisting of its posterior border. These characters differ markedly from those of Donrussellia, and several of them might be primitive, which would lead to a reappraisal of the adapiform ancestral morphotype. It would be important to find some M2/. Asiadapis cambayensis is larger than M. indicus. It differs from it by a single-rooted P/2, several details of the lower teeth (small paraconid present on some M/2–3, larger third lobe on M/3), and the upper teeth showing more massive outlines and proportions as well as a well-formed crista obliqua almost continuous to the tip of the metacone. Some characters of this species are again quite different from those of cercamoniines; for example, some M/2–3 have a deep groove between protoconid and metaconid, and an M1/ has a preprotocrista leading to a paraconule and continuing toward the tip of the paracone. The differences noted between these two genera and European cercamoniines underline a marked systematic separation between the two groups.

Several limb bones of these asiadapines have been described (Rose et al. 2009). A beautiful complete humerus shows a rounded head, a prominent and low greater tubercle, a salient deltopectoral crest slightly overhanging the bicipital groove, and a proximally extended brachialis flange. Its distal extremity shows a spherical capitulum projecting distally beyond the trochlea, a well-formed intercondylar groove, a conical trochlea projecting only very slightly distally (more than in Cantius and Notharctus, less than in the omomyiforms Microchoerus and Shoshonius). Its characters are in overall agreement with a generalized arboreal quadruped. Proximal radii show a slightly ovoid head (slightly rounder than in Cantius, but less than in living galagos). Two femora have a subspherical head and a long neck making an angle relative to the shaft near 57°, as in adapoids and living lemurs. The greater trochanter does not project as far proximally as the head. The trochanteric fossa is deep, narrow, and bordered by a salient paratrochanteric crest. A small third trochanter is distal, opposite to the distal part of the lesser trochanter. The femoral shaft is elongate and comparable to that of Cantius, Notharctus, and living lemurs. However, the distal extremity is not as high anteroposteriorly as in these fossils or in leaping prosimians. Again, the femora fit with the notion of an active arboreal quadruped. Tarsals have also been found, 11 calcanea and 5 astragali. These fall into two size classes and probably pertain to more than two species. The three larger partial calcanea, probably belonging to A. cambayensis, appear very similar to, but smaller than, those of Cantius. Among the smaller ones, four appear again similar (probably M. indicus) and four differ slightly in proportions, having a shorter proximal and a longer distal part, indicative of increased leaping propensities. It is difficult to establish whether these differences reflect a different species or merely extensive intraspecific variability (Rose et al. 2009). One large and four smaller astragali are very similar to those of notharctids. Only one of them has a lower neck angle, closer to those of omomyids and eosimiids. However, relative neck length is more discriminant of primate groups, and by that measure these astragali are like those of notharctids. The postcranials as a whole reflect the adaptations of small active arboreal quadrupeds. In conjunction with tooth dimensions, the postcranials permit relatively good weight estimates for these species, which are around 100–120 g for M. indicus and 250–300 g for A. cambayensis (Rose et al. 2009).

Asiatic adapiforms of late Early and early Middle Eocene age are restricted to small species from Pakistan. The Early Eocene Gandhara Quarry yielded an assemblage of Panobius russelli, which shows interesting characters (Gunnell et al. 2008). M/2 in the holotype dentary is remarkably primitive due to its large lingual paraconid, well separated from the metaconid, and a P/4 that is more elongated and posteriorly narrower than in Donrussellia. The sole illustrated M1or 2/ is very transversely elongated, due to the extended lingual slope of the protocone. Panobius might have some dental characters more primitive than in Marcgodinotius, and it probably would add to a reconstruction of the primitive adapiform morphotype. Two other species, P. afridi and P. amplior, are from the younger locality of Chorlakki, early Middle Eocene, and are more fragmentarily known (Russell and Gingerich 1987; Gunnell et al. 2008). Sulaimania arifi from the Gandhera Quarry is known only through an isolated M/2, revealing a small adapiform with a continuous anterior paralophid, reminiscent of some anchomomyini.

Four genera of small Asiatic adapiforms have been described from Late Eocene and Early Oligocene beds of Thailand, Burma, and Pakistan. They are all known by very fragmentary material. The first described, Bugtilemur mathesoni from the Early Oligocene of Pakistan, is also the best known, being represented by some upper teeth in addition to lowers (Marivaux et al. 2001). Its upper molars are simple, with a continuous lingual cingulum, no hypocone and conules, a postprotocrista posteriorly directed, and a posteriorly opened trigon basin. A high centrocrista is reported as shared-derived with living Cheirogaleus. The lower molars have a short trigonid with metaconid posterior to the protoconid, a broad talonid basin rounded posteriorly, and an anteriorly directed cristid obliqua. The P/4 is elongated, molarized with a narrow trigonid and a broad talonid basin with a small entoconid. Many dental similarities with Cheirogaleus led a parsimony analysis to place Bugtilemur as a sister group of Cheirogaleus, nested within the Malagasy lemuriform radiation, despite its lower canine not having the morphology of a tooth comb canine. The subsequent description of a small mandible from the Late Eocene of Thailand, named Muangthanhinius siami, revealed important features: an elongated and low dentary, a two-rooted, high and unreduced P/2, a partial canine root that is large and relatively vertically implanted – all suggestive of adapiform affinities. Some differences notwithstanding, P/3 to M/1 show a remarkable general similarity with those of Bugtilemur (Marivaux et al. 2006), which established adapiform status for both genera.

Four isolated teeth from two different localities of the Pondaung Formation, late Middle Eocene of Myanmar, were described as the smallest sivaladapid species, Paukkaungia parva (Beard et al. 2007). The M/1 is primitive by its broad trigonid with large paraconid and extended trigonid basin. Its talonid is broad and rounded in outline. It is possible to identify a hypoconulid and an entoconid among the poorly differentiated posterolingual cusps, making this species reminiscent of other sivaladapids. The P/3 and P/4 attributed to this species have a protoconid which is very extended anteroposteriorly and at the same time low. They have a sloping talonid basin restricted lingually by the posterior extension of the postprotocristid. P/3 has a low crestiform hypoconid, P/4 had a better differentiated one (worn) and a small entoconid. The roots of P/3–4 show a partial coalescence, suggesting a trend toward premolar compaction (Beard et al. 2007). Their unusually low relief might have been linked to some anteroposterior overlapping. They are intriguing in any event. The other species, Kyitchaungia takaii, is based on one isolated M/2 that is partially eroded. It is larger than Paukkaungia parva, and its entoconid and hypoconulid are better differentiated. Some postcranials are referred to this species. A calcaneum is relatively similar to those of the notharctids, although it differs from them by a broader proximal facet and a deep medially offset cuboid pivot. The curvature of its proximal part is marked. An astragalus without head also resembles those of small notharctids. A proximal femur from the same locality, attributed to the same individual as the tarsals, has been described in detail (Marivaux et al. 2008). It reflects good hip mobility. All of these elements in Kyitchaungia suggest a broad locomotor repertoire with quadrupedalism, including some leaping, climbing, and possible suspensory activities.

On the whole, the Asiatic adapiforms remain poorly known, with no skull described, many taxa known only by fragmentary dental remains, and a recently expanded but still limited postcranial record. Various recent discoveries have substantially increased our knowledge, confirming that the Asiatic adapiform radiation probably played a central role as the source of dispersals into Europe, North America, and Africa. However, much more remains to be discovered.

Eocene Lemuriformes, Stem Lemuriforms, and the Concept of Strepsirhini

Eocene Lemuriformes

Living lemuriforms can be easily distinguished from fossil adapiforms by the possession of a dental complex, the tooth comb, formed at the anterior extremity of the lower jaws by the closely appressed and proclive lower incisors and canines. Recurrent speculations about the affinity of some adapiform genera, lately anchomomyins, with the Lemuriformes were put to rest by the discovery in Africa of Eocene lemuriforms bearing a tooth comb (Seiffert et al. 2003). Found in the Egyptian Fayum stratigraphic sequence, they come from the two localities seen above for the two adapids. Saharagalago and Karanisia come from BQ–2, early Late Eocene (around 37 Ma), and Wadilemur from L-41, a Latest Eocene locality (Seiffert et al. 2005a; Seiffert 2006). They are all small primates whose upper molars have a well-formed cingular hypocone, protocone crests surrounding an anteroposteriorly broad trigon basin, and a sharply defined postprotocrista joining the metacone on Karanisia and Saharagalago, but not reaching its summit in Wadilemur. The lower molars have an anteroposteriorly compressed trigonid without paraconid and with a long paracristid often joining the metaconid summit. Karanisia is peculiar by its continuous lingual cingulum and extensive hypocone shelf, with small crestiform hypocone. Its P4/ and P3/ have a broad protocone lobe and cusp, with P4/ also showing a continuous lingual cingulum. A lower dentary of W. elegans shows procumbent incisor and canine alveoli, and P/2–4. P/2 and P/3 have crowns which project anteriorly with some overlap of successive teeth. P/4 has a well-formed talonid basin, with a tall labial hypoconid and a long acute cristid obliqua. This morphology is reminiscent of later galagids. A partial femur ascribed to Wadilemur presents characters reminiscent of those of living galagids, e.g., a cylindrical femoral head, suggesting frequent leaping in its repertoire, but less specialized than in the vertical clinging and leaping galagos. Wadilemur and Saharagalago are interpreted as stem Galagidae, and Karanisia as a stem lorisoid (Seiffert et al. 2003, 2005a). This implies that the split between lorisoids and the Malagasy lemuroids would be older than Late Eocene; however, the lemuroid morphotype is unknown. Two genera found in the Early Oligocene of Oman, Omanodon and Shizarodon, are known only by isolated teeth and have remained relatively enigmatic (Gheerbrant et al. 1993). Dental similarities between O. minor and S. misrensis suggest that they probably belong to the same group of early lemuriforms (Godinot 2006).

Stem Lemuriformes

Two distinct genera can be placed in a family named the Djebelemuridae, which are now assumed to be stem lemuriforms – which means, closely related to Lemuriformes but not possessing their defining apomorphy, the tooth comb. Djebelemur martinezi occurs in the late Early or early Middle Eocene locality of Chambi, Tunisia (Hartenberger and Marandat 1992; Marivaux et al. 2013). The type mandible preserves P/3–M/3, and alveoli for a moderate-sized canine and P/2. P/3 and P/4 are very simple, elongated and narrow, relatively low, bearing a continuous lingual cingulid and a simple talonid cusp. The lower molars have long paracristids joining the metaconid summit. An isolated lower canine and P/2, and a maxilla bearing P3–M3/, were added recently. P/2 is single-rooted, moderate in size, slightly procumbent. The canine is surprisingly small, and also somewhat procumbent. The upper molars have a trigon resembling that of Eocene lemuriforms, but there is no hypocone, whereas there is an almost complete lingual cingulum, interrupted only lingually to the protocone summit. P3/ and P4/ are triangular in outline, P4/ having a small and low protocone, and P3/ only a cingular cuspule. The polarity of these premolar characters is intriguing: primitive or secondarily simplified? An astragalus from Chambi shows the typical characters of small strepsirhines, with a strongly sloping fibular facet and a posterior trochlear shelf with a laterally offset groove for the flexor fibularis tendon. The elongated neck and tightly curved profile of the trochlea suggest that leaping was a part of its locomotor repertoire (Marivaux et al. 2013). Isolated petrosals from Chambi also pertain to Djebelemur or to a small azibiid present in the fauna. Their detailed study with micro-CT-scan reveals interesting characters (Benoit et al. 2013). Among them are a stapedial and a promontory artery that are very small and suggest that the internal carotid was supplemented by another vessel.

Another more recent species, awaiting a new generic name, is “Anchomomys” milleri from the L-41 locality, Latest Eocene of the Fayum, Egypt (Simons 1997). A mandible bearing all teeth from canine to M/2 shows M/1–2 similar to those of Djebelemur, and broader P/3 and P/4, the latter having a second crest descending posteriorly from the protoconid summit (Fig. 11). P/2 is somewhat smaller than P/3, not reduced. The canine is larger than P/4, low with a curved anterior border and a relatively low and rounded summit. Its crown is slightly procumbent, but the root is rather vertical. Its continuous lingual cingulid becomes posteriorly higher above the crown base, all this giving it a premolariform appearance. Both Djebelemur and “A.” milleri clearly did not possess a tooth comb, yet their canines are no longer the high pointed canines of primitive adapiforms; they are low and slightly procumbent, probably illustrating a step in the transformation of a primitive canine into a procumbent tooth comb canine. Phylogenetic analyses recover the place of djebelemurids as stem lemuriforms, along with Plesiopithecus (Seiffert et al. 2005; Marivaux et al. 2013).

Fig. 11

Lower teeth of the stem lemuriform “Anchomomys” milleri from the Fayum in occlusal (a) and lingual (b) views. The anterior tooth on the left is a canine, which is relatively low and premolariform. It is followed by long and narrow P/2 and P/3, a larger P/4, and two molars, which have a short trigonid without a paraconid. SEM of casts

Plesiopithecus teras is a very unusual primate found in Quarry L-41 of the Fayum. Its lower mandible bears an enormous, procumbent anterior tooth which is a lower incisor or a canine, followed by a very reduced canine or P/1. The lower premolars are simple; they are broader, lower, and more anteroposteriorly compressed than in djebelemurids. The lower molars likewise have anteroposteriorly compressed trigonids with long paracristids, and they also are much broader and lower than in djebelemurids. There is sometimes a metastylid, and M/3 is much shorter than in Djebelemur, having a very abbreviated third lobe. A distorted cranium shows upper molars decreasing in size from M1/ to M3/ (Simons and Rasmussen 1994). They are relatively simple, with a continuous lingual cingulum but without hypocones. Whereas P4/ is transversely broad, P3/ is single-cusped and the small P2/ is bilaterally compressed. There is a very large vertical canine, inserted in a high muzzle. The roof of the cranium is anteroposteriorly arched. The orbits were large and indicate nocturnal habits. Plesiopithecus is highly autapomorphic. Its lower jugal teeth recall djebelemurids and lorisoids, and the very large lower tooth could indicate an enlarged tooth comb canine accompanied by incisor loss (Simons and Rasmussen 1994). This scenario or possible ties with Daubentonia (Godinot 2006) would imply a lemuriform status, whereas scenarios from a more primitive djebelemurid state imply a stem lemuriform status as found in the cladistic analyses.

Azibiidae

It took recent recovery of sufficient dental remains for researchers to better realize what these bizarre primates probably are. They occur in late Early or early Middle Eocene localities of North Africa. After the initial description of Azibius trerki from the Gour Lazib, Algeria, by Sudre (1975) as a probable prosimian, the dentary of this genus was later disputed. Its high and roughly blade-like P/4 was so unusual that the primate status of the genus was doubted (Szalay and Delson 1979). Isolated molars of the smaller Algeripithecus minutus, found in the close Glib Zegdou, Algeria, are extremely bunodont, and its M2/ is so similar to that of some parapithecids that its anthropoid status received general agreement (Godinot and Mahboubi 1992, 1994). However, more complete material recovered later revealed that the two genera were close to each other and led researchers to synonymize two other genera (Tabuce et al. 2009). A dentary of Algeripithecus shows a tooth row increasing in height from a low M/3 to an M/1 with elongated and high trigonid, to long, high and blade-like P/3–4, P/4 with two successive summits, P/3 with a second summit, presumably a metaconid, smaller and lower than the protoconid. P/3 and P/4 have complete labial and lingual cingulids showing an anterior elevation. Anterior alveoli are interpreted as those of a small single-rooted and procumbent P/2 and of a large canine, procumbent and with a long posterior root lingual to the roots of P/3. The upper teeth are all isolated. The upper molars are extremely bunodont. The transversely elongated M2/ has a large hypocone, almost as high as the protocone, a metaconule, and an extended labial slope of the paracone. The anteroposterior crests on the paracone and metacone are salient. M1/ is less transversely broad, and its paracone is higher, with steeper slopes. P4/ has two labial cusps, the larger one being a high peak, in profile view, with a high labial slope suggesting some exodaenodonty (enamel extended beyond the alveolar margin, along the roots). The tooth is transversely short, with the lingual lobe bearing two small cusps, a small pointed protocone and a smaller hypocone. The high P3/ seems to have one large cusp. A smaller transversely compressed P2/ is associated with the others.

The dentition of Azibius trerki is roughly similar but also shows some differences. Aside from its larger size, it is even more bunodont, and its P3–4/are more transversely extended and more molarized: P4/ has a recognizable trigon basin and paraconule, and P3/ a small protocone lobe. A small portion of maxilla bearing P3/ and P4/ revealed a large infraorbital foramen, the trace of a lacrimal canal oblique anteroventrally, and a part of the orbital floor suggesting that the orbits were large, as in nocturnal primates. All these characters are more typical of strepsirhines (or omomyiforms), and the phylogenetic analysis of Tabuce et al. (2009) placed azibiids as primitive sister group of a clade (dejbelemurids + lemuriforms). In the future, such a position as stem lemuriforms might even be shifted within lemuriforms, if the long procumbent lower canine of Algeripithecus turns out to be as reminiscent of a tooth comb canine as the authors suggest. Strepsirhine affinity is confirmed by the description of an astragalus ascribed to Azibius and a larger to a larger undescribed species of the same group. These astragali show the laterally sloping fibular facet typical of strepsirhines (Marivaux et al. 2011). Researchers have inferred quadrupedal and climbing abilities for this species that are similar to those of living cheirogaleids. One of the most intriguing aspects of these azibiids has received little attention until now, namely the adaptive significance of their dental specialization. Most living and fossil lemuriforms of small size are highly insectivorous. What may have been the diet of Algeripithecus, whose body weight is estimated between 65 and 85 g?

The Concept of Strepsirhini

A number of shared derived characters, which will be detailed below, unite tarsiids and simians in a monophyletic clade, the Haplorhini. By contrast, living Lemuriformes and Strepsirhini appear basically primitive. This is not a problem for living lemuriforms, which can easily be diagnosed by their possession of the tooth comb, a derived character. However, it becomes critical when one considers the place of fossil adapiforms, which are most often considered strepsirhines. Such an affinity has long been assumed, based on their possession of middle ear characters extremely similar to those of living lemurs. The bulla, the free tympanic ring inside of it (formed by the ectotympanic), the carotid entry into the bulla, and the arterial circulation inside it are extremely similar in adapid and living lemur crania (Stehlin 1912). However, inasmuch as these characters have long been considered primitive in primates, they do not prove close affinity. Nevertheless, detailed morphological and embryological studies enable us to identify possible synapomorphies. The truly annular ectotympanic is a specialized retention of a fetal character limited to adapiforms and lemuriforms, and rarely found elsewhere among mammals (among which most “ring-like” ectotympanics are in fact slightly expanded; MacPhee 1981, 1987; MacPhee and Cartmill 1986). Another potential synapomorphy is a gap existing in the annular bridge connecting the ectotympanic and the bulla wall (“recessus dehiscence”; Beard and MacPhee 1994). Furthermore, clearly shared-derived between adapiforms and lemuriforms are a suite of characters issued from studies of postcranial anatomy . Several characters of the astragalus – a laterally sloping fibular facet, a position of the groove for the flexor fibularis tendon offset from the posterior trochlear facet, a long posterior trochlear shelf – are derived and found in both adapiforms and lemuriforms. They are complemented by characters of the distal tibia and the navicular (Dagosto 1985; Gebo 1988; Covert 1988). This suite of characters indicates increased hallucial opposability, more habitually inverted foot postures, and foot flexion/extension accompanied by some conjunct rotation at the upper ankle joint. So far, there is no evidence of convergent acquisition of these characters, some of which are found in stem lemuriforms (Marivaux et al. 2011, 2013). These characters are probably strengthened by hand characters: adapiform hands have a structurally divergent thumb, as do those of lemuriforms, probably derived relative to simian hands (Godinot 1992b). However, this depends on the reconstruction of the primitive primate morphotype, which is under debate (see below in anthropoid origins part). Early omomyiform and simiiform hands are almost unknown. In any case, regardless of possible hand characters, strepsirhine monophyly appears well established.

The Omomyiformes Radiations

Besides adapiformes, many smaller fossils have been described in Eocene times; these are either united in one family, the Omomyidae, or spread out over two – the Omomyidae for the North American forms and Microchoeridae for the European ones. For a long time, these small forms were found to be “tarsier-like” and referred to an infra-order Tarsiiformes. They have also been suspected to possibly include anthropoid ancestors. However, the so-called similarities with Tarsius were exaggerated. Most omomyids have large eyes due to their small size and nocturnal adaptation; yet these eyes are no larger than in living small nocturnal strepsirhines. Despite new discoveries, recurrent re-analyses of fossils, and increasingly sophisticated phylogenetic analyses, it has proven impossible until now to identify among the Omomyidae a group which would be a consensual sister group of Tarsiidae. Moreover, most of these fossils appear not to possess the characters indicating anatomical haplorhinism (explained below for Tarsiidae). Thus, for the sake of clarity and consistency, these omomyids should not be included in Tarsiiformes. A number of omomyid groups had their own history and became extinct without showing any trace of evolution toward tarsiid or simian characters. The best choice is to place them in the taxon Omomyiformes , proposed by Schmid (1982), which has been adopted by a growing number of specialists (e.g., Ross et al. 1998; Williams et al. 2010). That is probably a paraphyletic grouping, but it is not appropriate to transfer some of them to a useful Tarsiiformes until it can definitively be shown that the group is a sister group to tarsiids. This is an ongoing debate, to which we will return below after discussion of the Tarsiidae. There are more than a hundred species of omomyiforms, found on three continents, which indicates a complex history for this broad group.

The Stem Genus Teilhardina

The genus Teilhardina is exceptional in many respects. Found on three different continents in the Earliest Eocene, it appears at the base of the later diversification of several subfamilies. As such it is paraphyletic and represents one of the goals of paleontologists: to identify an ancestral genus, a true stem genus at the base of the evolutionary diversification of entire families or subfamilies. This genus was first described from dental remains from the Belgian locality of Dormaal, and was named “Omomys” belgicus (Teilhard de Chardin 1927). The species was removed from Omomys and placed in a new genus Teilhardina, as T. belgica, by Simpson (1940). The same genus was later identified in an Early Eocene locality from the Willwood Formation, Wyoming, by Bown (1976), who named it T. americana. Several other North American species were subsequently named (see below). Eventually, the genus was also discovered in China. A cranium and associated mandibles, named T. asiatica, increased our knowledge of the genus substantially (Ni et al. 2004). One species has also been named from the Tuscahoma Formation of Mississippi (T. magnoliana, Beard 2008). Linked to the rapid warming of Paleocene–Eocene boundary events, Teilhardina spread from Asia to the two other continents – probably through Europe to North America (T. belgica and Earliest Eocene American T. brandti, Smith et al. 2006). Species of Teilhardina are very small, with dentally estimated body weights around 30 g, similar to the smallest living primates. They have a very primitive dentition, with four premolars, a tiny metaconid on P/4, lower molars with a large paraconid on M/1, reducing on M/2 and M/3, and upper molars without hypocone (Fig. 12). The cranium is the oldest known for primates. It shows a broad and rounded braincase and probably featured a somewhat shortened snout. Its orbits are smaller than in any other omomyid, probably reflecting diurnal habits. In contrast, most later omomyiforms are considered to have been nocturnal. These orbits are convergent (angle estimated at 51°) and the interorbital breadth is narrow (Ni et al. 2004). The infraorbital foramen is relatively large. In the case of T. belgica, postranials are known (Szalay 1976). A moderate elongation of its calcaneum suggests leaping abilities and a locomotor repertoire close to that of living cheirogaleids.

Fig. 12

Isolated teeth of Teilhardina belgica from Dormaal, Begium, all in occlusal view. The M2/ (a) is very simple and primitive, transversely elongated without hypocone or protocone fold; however, there is a slight lingual thickening of the posterior cingulum. Lower molars are M/1 (b), M/2 (c), and M/3 (d); M/1 and M/3 appear primitive through their large trigonid with a large paraconid. SEM of casts

North American Anaptomorphinae

Among the North American Omomyidae two subfamilies are recognized, succeeding each other in abundance through time: species of Anaptomorphinae are the most abundant during the Early Eocene, whereas Omomyinae become dominant during the Middle Eocene. Several clades of anaptomorphines can be recognized in the Early Eocene (Bown and Rose 1987). Some of the Middle Eocene followers can be linked to these clades whereas others are more difficult to root in the earlier forms.

The first clade, that of Teilhardina-Anemorhysis, starts from T. americana. In the exceptional fossil record of the Bighorn Basin, Wyoming, a series of assemblages links the species T. americana and T. crassidens through intermediate assemblages (Bown and Rose 1987). The whole lineage has an I/1 relatively enlarged in comparison with T. belgica or T. brandti (Fig. 13). This anagenetic lineage shows a slight diminution in size, broadening of the cheek teeth, lowering of P/3 and P/4 which become more molarized (P/4 with larger paraconid and metaconid, P/3 with small metaconid), a small mesostyle present on M1/, variable on M2/. A very rare smaller species T. tenuicula occurs later. In the Bighorn Basin, species of Anemorhysis show up later as punctuated occurrences, showing that the lineage was evolving elsewhere and species were entering from time to time in the Basin. The five species of Anemorhysis have a wide geographic distribution in the Early Eocene of Wyoming, Utah, and North Dakota, surviving in the earliest Middle Eocene of Wyoming. They differ from Teilhardina and Tetonius species by their more molarized P/4, lower molars with sharp cusps less basally inflated than in Tetonius, anteriorly broad talonid basin and straight postcristid (Fig. 13). P/4 has a well-developed metaconid, a prominent paraconid, and a well-developed talonid basin with hypoconid and small entoconid. A. savagei, from the Washakie and Wind River Basins (Lysite, Wa6), Wyoming, is considered a structural intermediate between Teilhardina and later Anemorhysis species (Williams and Covert 1994). It is small, retains a P/2 and has relatively simple P/4. A. wortmani and A. sublettensis have P/4 with a large paraconid close to the metaconid, and the second is further derived by its long and broad talonid basin. A. pattersoni is larger and has a low and weak P/4 paraconid. A. natronensis, from the earliest Bridger, is distinct through narrow lower molars, a very large entoconid on P/4, and I/1 only slightly larger than I/2.

Fig. 13

Dentitions of primitive North American anaptomorphines, Teilhardina and Anemorhysis. Maxillae of Teilhardina intermediate between T. americana and T. crasssidens in occlusal view (a) and of T. crassidens (e); mandibles of T. americana in lingual view (b), T. crassidens in occlusal view (c), and intermediate between the two species in lingual view (f). Left mandible of Anemorhysis sublettensis in occlusal view (d). The anterior incisor is enlarged in the lineage (f), and the canine in the intermediate specimens (f) is more reduced than in Teilhardina americana (b: from right to left canine, small P/2, P/3, P/4, M/1-2-3). Photographs of white-coated epoxy casts, all at the same scale

It is possible that this clade gave rise to Arapahovius. The difficulty of making systematic decisions involving morphologically close species, even ones represented by abundant dental material, is illustrated by Tetonoides pearcei. Since its creation by Gazin (1962) this species has many times been placed in Anemorhysis, then subsequently removed and placed again in a valid genus Tetonoides. Lately, Gunnell and Rose (2002) listed it as an Anemorhysis species. Cuozzo (2002) showed that Late Graybullian material assigned to T. pearcei differs only in very subtle details from Lysitean specimens referred to A. savagei, implying that both be placed in Anemorhysis. However, Tornow (2008) determined T. pearcei to be in the position of a separate genus related to Arapahovius, and listed a second species, T. coverti, that was named in a dissertation. The message is that these species are very close to each other. It is possible that artifacts of cladistics and nomenclature are blurring the incipient divergence of lineages.

Contrary to Tetonoides, Arapahovius differs by marked characters. Found in the upper part of the Wasatch Formation (Lysite equivalent), Wyoming, A. gazini is characterized by crenulated enamel on all molars and upper premolars (Savage and Waters 1978). The upper molars are transversely elongated, bear a well-formed protocone fold, conules with marked pre- and postcristae, and an incomplete lingual cingulum without hypocone. On the dentaries, alveoli show its I/1 to have been moderately enlarged, with I/2 and C being smaller. P/3 and P/4 are moderately molarized, and P/2 is reduced. M/2–3 have broad and anteroposteriorly short trigonids. Tarsals of Arapahovius include an astragalus which has an elongated neck, a tibial trochlea with lateral ridge higher than the medial one, a posterior trochlear shelf, and a very rounded head. Partial calcanea present a short anterior part and a distal elongation indicating leaping propensities. A navicular and a cuboid are both moderately elongated, contributing to foot elongation as in Hemiacodon (see below). A smaller and more primitive species, A. advena, has been found in the Bighorn Basin (Bown and Rose 1991).

A second clade that is well identified in the Early Eocene is represented by the Tetonius-Pseudotetonius group. At the beginning, it is still very close to some Teilhardina. It contains the remarkable series of assemblages linking T. matthewi to P. ambiguus through intermediate assemblages that have been difficult to name (Fig. 14). This series illustrates one of the most beautiful anagenetic lineages showing progressive morphological change through time, in the ideal context of regional stratigraphic superposition (Bown and Rose 1987). This lineage displays a progressive increase in size of I/1 and P/4, and reduction of the teeth which are between them. In T. matthewi, I/1 is large and P/3–4 have a normal size relative to M/1. At the end of the lineage, I/1 has become enormous, P/4 is strongly enlarged relative to M/1, one tooth has been lost, and the others are crowded and reduced, justifying a different generic name. Intermediate assemblages show some reduction of I/2 and C, the loss of P/2, and a progressive reduction of P/3, which becomes single-rooted and later very small. Changes occur not abruptly but through displacements in variations. The process of anterior incisor increase linked to a reduction of the teeth between it and P/4 or P/3 is a common evolutionary trend, usually explained by selection on anterior incisor function. The small Tatmanius szalayi, which has a high pointed P/4 without metaconid, is considered a likely descendant of Pseudotetonius (Bown and Rose 1991).

Tetonius is a well-known genus. The skull of T. homunculus is known since its description by Cope in 1884, and it has been restudied since (e.g., Szalay 1976). The cranium is broad and short in dorsal view. A large orbit is circumscribed by a complete postorbital bar. The cranium is incomplete and crushed. A remnant of bulla wall shows that the bulla was large. In its accessible details, such as the promontory canal included in a septum, Szalay (1976) found it very similar to Necrolemur. The dental formula was discussed by earlier authors. Szalay (1976) identified isolated teeth and reconstructed T. homunculus as having enlarged anterior incisors, smaller roughly similar-sized I2 and canine, and very reduced P2 above and below. The P4 are slightly enlarged above and below, the lower one being higher than the molars and P/3. P/3 and P/4 are simple, P/4 being broad with reduced or absent metaconid and paraconid. M2/ is transversely elongated. M1/ and M2/ have a protocone fold, small conules, and a complete lingual cingulum (Fig. 14). The lower molars have a lingually placed paraconid that is large on M/1, smaller and closer to the metaconid on M/2–3. The M3s are reduced.

Fig. 14

The lineage Tetonius-Pseudotetonius. Maxilla (a) and mandible (b) of Tetonius; mandible of an intermediate, stage 3 of Bown and Rose (1987), in occlusal (c) and lingual (d) views (d inverted for comparison); two mandibles of Pseudotetonius in lingual (e) and labial (f) views. There is a marked difference between the mandible of Tetonius, with a large P/3 and three teeth between P/4 and the large incisor (b), and that of Pseudotetonius, with a very small P/3 and alveoli for two other small teeth between the large anterior incisor and P/4 (f). The intermediate stage (c and d) shows a moderately reduced P/3. Photographs of white-coated epoxy casts, all at the same scale

The third clade of anaptomorphines which is differentiated in the Early Eocene is that represented by species of Absarokius, its close derivatives Strigorhysis and Artimonius, and Middle Eocene genera considered to be descendants of this group, namely Aycrossia and Gazinius (Fig. 15), as well as possibly Anaptomorphus. The genus Absarokius is characterized by enlarged upper and lower P4, associated with incisors much smaller and M3 more reduced than in Tetonius. In the rich and detailed record of the Bighorn Basin, Bown and Rose (1987) distinguish two divergent specific lineages, A. metoecus and A. abbotti, both showing anagenetic change through time and giving rise to two different genera. The lineage of A. metoecus shows trends toward transverse narrowing and trigon basin broadening on M/1–2, some ridulation of upper molar enamel, and labiolingual narrowing of the M/1 trigonid. These characters are found in the Late Wastachian, briefly occurring A. gazini, which has narrow lower molars and somewhat enlarged incisors. The A. metoecus specific lineage is giving rise directly to species of Strigorhysis, which split in the Early Bridgerian into S. bridgerensis and S. huerfanensis. There are three species of Strigorhysis, which differ from those of Absarokius by rugose enamel on all molars and upper molars in which the protocone fold has increased and joined the posterior cingulum to realize one strong posterior crest (postprotocingulum). These species are well represented in the Aycross Formation, Wyoming, and also the Willwood Formation of Wyoming and the Huerfano Formation of Colorado. The second lineage of Absarokius, A. abbotti, shows a tendency toward size increase, as seen in M/1–2 size, and increased hypertrophy and exodaenodonty of P/4 (ventral expansion of enamel on its labial side). These trends continue in the latest Wasatchian and earliest Bridgerian in three species formerly included in Absarokius and now included in the genus Artimonius (Muldoon and Gunnell 2002). In addition to the increase in P/4 hypertrophy, these species show different degrees of lower premolar crowding. They all have lost P/2. P/3 becomes single-rooted in Artimonius nocerai and A. australis, and P/3 is very reduced in A. australis and A. witteri. Several limb bones of Absarokius have been described (Covert and Hamrick 1993). A distal humerus shows the trochlea to be well separated from the capitulum by a groove. A distal tibia exhibits a proximally long facet for the fibula, reflecting close appression between the two bones. The omomyid-like astragalus shows a well-grooved trochlea, and the calaneum has an elongated anterior part (54 % of total length). These characters indicate a small quadrupedal and leaping primate (weight estimated around 200 g).

Fig. 15

Absarokius, other derived anaptomorphines, and Trogolemur. Maxilla of Absarokius abbotti (a) and mandible of A. nocerai (c). Maxilla of Gazinius amplus (b), mandible of Trogolemur myodes (d), and maxilla of Strigorhysis bridgerensis (e). Upper dentition (f and g) and mandible (h and i) of Aycrossia lovei. (f) shows P2/, P3/, and the lingual part of P4/. All are occlusal views, except for (h), which is lingual, and (i), which is labial. Photographs of white-coated epoxy casts, all at the same scale

Aycrossia and Gazinius include three rare species, known by fragmentary material (Fig. 15). The first species were discovered in the Aycross Formation of Wyoming, which samples basin margin, upland areas (Bown 1979). A. lovei has a tall P/4, small two-rooted P/3, M/1 with large paraconid, transversely elongated M1–2/, M3 not much reduced. Gazinius amplus is a large anaptomorphine which has a transversely very elongated M2/ lacking conules, postparacrista, and protocone fold, and with a lingual part so expanded that the protocone is almost centrally placed. G. bowni, found in the Green River Basin, is represented only by one M2/, which is smaller than that of G. amplus and has a well-formed postprotocrista and protocone fold (Gunnell 1995a).

Chlororhysis and Anaptomorphus are also genera known by fragmentary material and difficult to relate closely to the other, better documented anaptomorphine lineages. C. knightensis, known by four specimens (Early Eocene), is similar to Teilhardina in retaining unreduced canine and P/2, but it has more crowded anterior teeth and P/3–4 with more developed lingual cingulid. Gunnell and Rose (2002) note that it is similar to the omomyine Loveina. C. incomptus differs by small details. Two or three species of Anaptomorphus are known later, during the Middle Eocene, separated from potential ancestral forms by a long gap. They remain very small species, generalized, which have lost P/2 and have small M3. The lower molars are slightly bunodont, with bulbous cusps. M2/ is transversely elongated with expanded lingual slope. M1–2/ have well-developed metaconules and protocone fold. A. westi differs in being larger; sometimes a third species, A. wortmani, is also distinguished by its smaller size (or it is lumped into A. aemulus). These species remain poorly known.

The genus Trogolemur is so distinctive that it has often been placed in a special tribe, the Trogolemurini. However, the content of the tribe is in debate. Trogolemur represents the extreme in the trend toward I/1 size increase common in anaptomorphines. Its lower incisor is so large that its root is posteriorly extended below the molars. T. myodes is known through a number of specimens from the Bridgerian (Br2 and Br3) of the Bridger Formation, southern Green River Basin, Wyoming (Gunnell 1995a), and from Nevada (Emry 1990) (Fig. 15). Slightly older species, T. amplior and T. fragilis, have been described from the earliest Bridgerian (Br1) of the Wind River Basin; however, these are very fragmentary (Beard et al. 1992). The genus Sphacorhysis has been erected for a species which shows a morphology plesiomorph in comparison with Trogolemur but is advanced in its direction in comparison with other genera. Several phylogenetic analyses have rooted Trogolemur (+Sphacorhysis) near Anemorhysis, postulating a sister group relationship of this clade with Tetonoides and Arapahovius. However, the content of such an extended tribe of Trogolemurini is not consensual among specialists (e.g., Arapahovius + Tetonoides are rooted in Teilhardina crassidens according to Tornow 2008).

Omomyinae

The subfamily Omomyinae is a beautiful example of mosaic evolution during a phase of diversification. Starting with the Early Eocene generalized genus Steinius, a rapid diversification leads to a large number of genera in the Middle Eocene, well recorded in Wyoming and surrounding basins, which decreases again in the Late Eocene, during which members of this group find refuge in California and Texas. Dental specializations allow the recognition of a number of tribes or subtribes, but a precise resolution of their phylogenetic relationships is difficult to achieve due to the large number of convergences in their dental characters. Rose et al. (1994, p. 20) summarized the problem thus: “Particular derived characters that must have evolved independently in two or more lineages include enlargement of I/1 (usually associated with crowding of anterior teeth), loss of one or more lower premolars, hypertrophy of P/4, molarization of P/4 (involving more distinct metaconid and paraconid or development of a talonid basin), reduction or enlargement of third molars, crenulation of enamel, and presence of a mesostyle”. In spite of these difficulties, added to an incomplete documentation of their early phases of divergence, successive phylogenetic analyses have recovered certain relationships, some of which are becoming consensual and will be mentioned below (Szalay 1976; Honey 1990; Gunnell 1995a; Muldoon and Gunnell 2002; Tornow 2008). The record is extremely irregular, including taxa represented by one specimen as well as taxa represented by hundreds of them, and covering long time spans and large geographic areas.

Steinius, Omomys, Diablomomys, and Chumashius

These taxa constitute a first group. Steinius verspertinus is found in the Early Eocene of Wyoming. It is very primitive in retaining four premolars, a canine relatively as large as in Teilhardina, tall P/3 and P/4, and an unreduced P/3. Its sole clearly derived feature is a moderately enlarged I/1. If its unreduced M/3 and the more peripheral cusps on its molars were considered primitive relative to Teilhardina, it would imply the existence of an unknown lineage in the Earliest Eocene (Rose et al. 1994). The two latter characters make it a good candidate as an ancestral omomyine. In fact, a second species S. annectens is closer to Omomys and confirms the proximity of the two genera (Bown and Rose 1991).

The genus Omomys was described by Leidy in 1869. The species O. carteri is very abundant in Middle Eocene beds of the western regions of North America, accounting for 64–90 % of all omomyid specimens through the Early and Middle Bridgerian (Muldoon and Gunnell 2002). Its dentition is remarkably generalized in comparison with all other omomyines (Fig. 16). It has a moderately enlarged I/1, a relatively small canine (smaller than I/1 and P/3), and a very small single-rooted P/2. P/3 and P/4 are relatively high and elongated; P/3 is simple and its short talonid is somewhat crowded below the anterior part of P/4. P/4 has a small metaconid. The three lower molars are simple with peripheral cusps and sharp crests. M/3 is unreduced. The paraconid is large and lingual on M/1, smaller and more labial on M/2 and M/3. P2/ is also very reduced. P3/ and P4/ have a large paracone, a parastyle, a lower protocone lobe with a preprotocrista, and a postprotocrista continuous with the complete posterior cingulum. The upper molars have a relatively broad (anteroposteriorly) trigon basin and no protocone fold. Both conules have their pre- and postcristae, and there are a hypoparacrista and a hypometacrista. The lingual cingulum is complete; a hypocone is present posteriorly, and a small pericone anteriorly on M2/. Detailed analysis of large assemblages of O. carteri from the Bridger Basin revealed aspects of intraspecific variability in dental traits (e.g., P/4 metaconid present in 91 % of individuals, M2/ pericones present in 80 %) and an increase in the frequency of several premolar features, suggesting anagenetic change through time (Cuozzo 2008). Three petrosals of O. carteri were analyzed in detail, revealing a series of characters of the otic capsule and middle ear cavity (Ross and Covert 2000). Most characters conform to an omomyiform model as documented in Necrolemur and Shoshonius, with minor differences between them. Postcranials of O. carteri, which are analyzed below, allowed an estimation of its body weight at 230 g (between 170 and 290 g; Anemone and Covert 2000).

Fig. 16

Dentitions of omomyines, Omomys, Uintanius, and two utahiini, Utahia and Stockia. Maxilla (a) and mandible (b) of Omomys carteri showing a relatively primitive dentition. The right mandible of Uintanius shows a very enlarged P/4 (c). The molars of Utahia kayi (d) and Stockia powayensis (e) are derived through their short trigonid with a reduced paraconid and large crenulated talonid basins. Photographs of white-coated epoxy casts, all at the same scale

A smaller species O. lloydi is documented in the early Middle Eocene. Diablomomys dalquesti is based on a maxilla from the Middle Eocene (Late Unintan) from Texas. Its M1/ is narrower lingually and has larger conules than in Omomys, and it has no lingual cingulum (Williams and Kirk 2008). Chumashius balchi, represented by a small number of specimens from California, is very close to Omomys. It differs from the latter only by its relatively larger canine, lower P/3 and P/4, and lack of distinct pericones and hypocones on the upper molars.

Uintanius and Jemezius

These form a small group of three species, sometimes considered close to Omomys, found to be a primitive sister group of the washakiins by Tornow (2008). Uintanius is characterized by enlarged P/3 and very enlarged P/4, which are exodaenodont: their enamel is ventrally extended below the alveolar margin on the labial side (Fig. 16). The upper premolars are also enlarged and have a reduced protocone lobe. The molars are relatively simple, the uppers having a small protocone fold. Alveoli of the anterior teeth preserved on a dentary of U. cf rutherfurdi show that Uintanius had two small subequal incisors, and a somewhat larger canine and P/2 (Gunnell 1995a). Szalay (1976) postulated that Uintanius specialized in food items which required great force to open but not much mastication. Gunnell found confirmation of this claim in a high proportion of dental specimens that showed heavy wear or had been broken and polished during life. Jemezius szalayi, found in the Early Eocene of New Mexico, has a lower and more complex P/4, relatively larger P3/ and P4/ protocones, and a relatively larger M/3 with a less compressed trigonid – characters which put it closer to Steinius.

Utahiini

A tribe Utahiini can be used to unite Utahia, Stockia, Ourayia, Chipetaia, Asiomomys, Wyomomys, and Ageitodendron (equivalent to the Ourayiini of Gunnell 1995a; Gunnell and Rose 2002). With the exception of Ourayia, known by two species (three specimens for each), all of these genera are monospecific and known essentially by (sometimes very limited) dental remains. They are all found in Middle Eocene beds, and Utahia kayi starts in the late Early Eocene of Wyoming. Utahia, Stockia, and Chipetaia have lower molars with large talonid basins, as well as compressed trigonids with reduced paraconids on M/2–3 (Fig. 16). The lower molars of Chipetaia are low and heavily crenulated (convergent with Microchoerus), reminiscent of frugivores with emphasis on seeds (Rasmussen 1996). Asiomomys changbaicus, from the Middle Eocene of northeastern China, known by one mandible bearing P/3 and M/2–3, is believed to be close enough to Stockia to testify to an utahiin dispersal from North America to China during the Middle Eocene (Beard and Wang 1991). The molars of Asiomomys are also partially convergent with those of the European Nannopithex, which does not imply close affinity but recalls how pervasive convergent characters can be. Gunnell (1995a) proposed a morphocline from Wyomomys bridgeri, which shows rounded simple talonids with robust labially bulging ectocingulid on M/2–3, without a talonid notch as found in Utahia, and with paraconid and metaconid isolated by a deep fissure; to Ageitodendron matthewi, which has a more reduced paraconid; to Ourayia uintensis, which has a more compressed trigonid without paraconid. More complete material from all these taxa is needed to more securely establish their relationships. Fragmentary hindlimb bones of Ouraya and Chipetaia are relatively similar to those of Omomys and Hemiacodon, which are detailed below (Dunn et al. 2006). Features such as the cylindrical shape of the femoral head in Chipetaia and details of the tibial plateau in Ourayia reflect a relatively large amount of leaping in these mid-sized primates. Body weight estimates are 500–700 g for C. lamporea, and 1,500–2,000 g for O. uintensis (Dunn et al. 2006).

Macrotarsiini

A tribe Macrotarsiini is used sensu Gunnell and Rose (2002) to unite Macrotarsius, Hemiacodon, and Yaquius, and to also include Tarka, Tarkadectes, and Tarkops, which have recently been shown to be closely related (Ni et al. 2010). Hemiacodon is a well-known genus described by Marsh, known by postcranials (Simpson 1940) and beautifully described by Szalay (1976, including a frontal). Gunnell mentioned 370 specimens of H. gracilis, from 50 different localities, in the collections of the University of Michigan. Yet its phylogenetic affinities have been debated (Fig. 17). It shares many dental characters with the washakiins, with which it was classified, with doubt, by Szalay (1976). It was taken out of the washakiins by Honey (1990), who emphasized that Hemiacodon has a greatly enlarged I/1 relative to I/2, more elongated and narrow (primitive) P/3–4, and a P/3 higher than P/4, whereas the known I/1–2 in washakiins are subequal in size and small, and their P/3 is slightly molarized, bearing incipient paraconid and metaconid. A different classification implies that Hemiacodon evolved enlarged conules, hypocones, and pericones in parallel with some washakiins. Honey hypothesized a sister group relationship with an extended concept of omomyini, including Macrotarsius, and Gunnell (1995a) restricted this close relationship to Macrotarsius only, a choice preserved by Gunnell and Rose (2002). Hemiacodon is found sister to a clade (Macrotarsius + Utahiins) by Tornow (2008). The upper molars of H. gracilis are more transversely elongated than in Macrotarsius; they have a protocone fold, large conules, rugose enamel, and a hypocone more distinct than in Macrotarsius. In the lower dentition, the P/4 is molarized, with a lingually arching paracristid and a well-formed and short talonid. The lower molars have peripheral cusps and high crests, and M/1–2 have a very wide talonid basin and distinct hypoconulids. The putative earlier species H. casamissus, represented by dentary fragments and worn molars, is considered doubtful. A new late Middle Eocene species, H. engardae, is larger and shows more acute crests on its lower molars and P/4, suggesting a probable increase in folivory (Murphey and Dunn 2009). Hemiacodon gracilis has long been the postcranially best-known omomyid. Its bones are analyzed below.

Fig. 17

Macrotarsius and Hemiacodon. Upper dentition of Macrotarsius showing the prominent molar mesostyle (a) and right mandible of M. montanus (b). Left mandible of Hemiacodon gracilis showing its P/4 with a well-formed trigonid (c). Photographs of white-coated epoxy casts, all at the same scale

Five species of Marcrotarsius have been named. This genus has a wide geographic distribution, being found in Middle and Late Eocene beds of the western interior basins, Texas, California, Canada, and possibly also in China. The most primitive species, M. jepseni, is known by two dentaries and the palate of one individual. M. montanus has reduced lower premolars and especially small canines relative to its molars. Upper molars have an anteroposteriorly wide trigon, no hypocone or protocone fold, big mesostyles linked to the centrocrista, big parastyles, and a thick crenulated labial cingulum (Fig. 17). The lower molars are unusual in possessing a prominent crest posteriorly joining the metaconid to the entoconid. The species M. macrorhysis is based on two isolated teeth from the Middle Eocene Shanghuang fissure fillings of Jiangsu Province, China (Beard et al. 1994). These P/4 and M/1 are similar to those of Macrotarsius species; however, given the high number of dental convergences among omomyiforms, it would be important to have more complete evidence to confirm this generic attribution and its consequences for dispersals.

Three dentally specialized Middle Eocene genera are united in a subtribe or tribe Tarkadectini. Because they appear firmly rooted in species of Macrotarsius in the phylogenetic analysis of Ni et al. (2010), they are included here in the Marcrotarsiini. Tarka stylifera (Ui1) and Tarkadectes montanensis (Ui) were enigmatic species from the late Middle Eocene of Wyoming and Montana, often referred to the Plagiomenidae (Dermoptera) due to their broad lower molars bearing supplementary cuspules and upper molars with a wide stylar shelf and complex stylar cusps. The discovery of a more complete dentary in a Middle Eocene locality of Inner Mongolia, pertaining to a slightly less derived species named Tarkops mckennai, showed the anterior dentition to be typical of omomyids and unlike plagiomenids (Ni et al. 2010). At the same time, it illustrated another example of Middle Eocene dispersal between North America and Asia. The alveoli in front of the dentary of T. mckennai show that there was a large anteriorly inclined I/1, a small I/2, a larger vertical canine, a tiny P/2, and a two-rooted P/3. P/4 has a large metaconid, almost as high and anterior as the protoconid, and a smaller and lower paraconid. The lower molars are very bunodont; they have a shallow crenulated talonid basin, a long postmetacristid, and a peculiar cingular cusp at the labial base of the protoconid. Tarka and Tarkadectes have exaggerated the transversal breadth of their lower molars, which are endowed with supplementary cuspules.

Washakiini

This is one of the most interesting tribes of the omomyines. It includes 10 species, contained in the genera Loveina, Basius, Shoshonius, Washakius (4 species) and Dyseolemur (Fig. 18). They are known from the late Early and Middle Eocene, and Dyseolemur survives in the Late Eocene of California. They have transversely elongated upper molars with conules and a protocone fold. They tend to develop a moderately sized hypocone (Washakius) or a big mesostyle (Shoshonius). Their P/3–4 are moderate in size and slightly molarized. The P/4 paraconid and metaconid are well developed in Washakius. Their lower molars have a relatively transversely narrow trigonid. They develop a metastylid on the lower molars, considered to be homologous between Shoshonius, Washakius, and Dyseolemur. They all have an elongated M/3. Loveina is the most primitive genus, starting with the rare and small L. minuta (Lisyte). Honey (1990) described two primitive species which extended the range of two separate lineages. Shoshonius bowni, more primitive than S. cooperi, was included in Shoshonius because it possesses enlarged mesostyles on its upper molars (convergent with Ourayia and Macrotarsius). Washakius izetti completed a morphocline W. izetti – woodringi – insignis, tracking a progressive enlargement of pericone and hypocone, as well as an increase in size of the metastylid. This morphocline in fact includes two lineages, and W. woodringi is clearly ancestral to Dyseolemur.

Fig. 18

Dentitions of Washakiini. Maxilla with upper teeth (a) and left mandible with P/4-M/2 (b) of Washakius insignis and fragmentary maxilla with M1-2/of W. woodringi (c), all at the same scale. Left mandible of Shoshonius cooperi (d) and right mandible of Dyseolemur pacificus (e), both at a similar scale, larger than for (a–c). Increase in cusp number is easily seen on the upper molars (a and c), and a metastylid behind the metaconid can be seen on the lower molars in (d) and (e). Photographs of white-coated epoxy casts

The washakiins have received renewed attention since the discovery of several partial crania and postcranials of Shoshonius cooperi in the late Early Eocene of the Wind River Basin, Wyoming (Beard et al. 1991; Beard and MacPhee 1994; Dagosto et al. 1999). The crania possess very large orbits, which not only indicate probable nocturnality but also raise the issue of a possible relationship with tarsiids. A detailed analysis of available cranial characters led to the conclusion that Shoshonius, Tetonius, Necrolemur, and Tarsius pertained to a monophyletic taxon, Tarsiiformes (Beard and MacPhee 1994). They share a series of derived characters: posteromedial and anterolateral bony flanges that overlap the bulla, narrow, peaked choanae, a narrow central stem of basicranium, reduced snout, parotic fissure, and suprameatal foramen. The authors add that, among the fossil tarsiiforms known by relatively complete cranial remains, Shoshonius appears to share the most recent common ancestry with Tarsius. They refrain from referring the washakiins to Tarsiidae because the phylogenetic relationships between washakiins and other omomyine tribes are not clearly resolved. In fact, several of the listed characters might be consequences of small size and large orbits as much as common heritage. Information from more omomyid genera is needed to further evaluate these characters. In the recent description of Archicebus, Necrolemur is said to have a non-reduced snout (Ni et al. 2013). Furthermore, Necrolemur has also been cited as possessing anatomical strepsirhinism. The monophyly of a broad tarsiiform clade comprising tarsiids, omomyids, and michrochoerids implies the convergent evolution of anatomical haplorhinism in tarsiiforms and simians, which would destroy the basis for the concept of Haplorhini (based on tarsiids + anthropoideans, see below). Such a far-reaching conclusion needs to be carefully evaluated. The postcranials of Shoshonius are analyzed below with those of other omomyines. When the postcranial characters were added to the cranial characters used in the phylogenetic analysis described above, tarsiiform monophyly was again recovered, but this time with Necrolemur as the sister group of Tarsius instead of Shoshonius (Dagosto et al. 1999). Adding the dental evidence would not clarify the issue, as Tarsius can hardly be rooted in the microchoerids, most of which have a very distinctive dental formula, and it is not any easier to root the dentition of Tarsius in the washakiin dentition (or the reverse). If there are phylogenetic relationships, they are not close. The phylogenetic relationships of Tarsius are further considered below.

The postcranial anatomy of the omomyines is progressively better documented, and its interpretation in terms of locomotor behavior has been correlatively enhanced. For a long time the limb bones of Hemiacodon gracilis were the best documents. Several bones were found together, including a partial pelvis, femoral and tibial extremities, and a partial right foot with astragalus, calcaneum, cuboid, navicular, entocuneiform, and first metatarsal (Simpson 1940; Szalay 1976). Their functional interpretation progressively improved (e.g., Dagosto 1985, 1993; Gebo 1988). Postcranials of two other omomyines were subsequently found and described at almost the same time: those of Shoshonius cooperi (Dagosto et al. 1999) and those of Omomys carteri (Anemone and Covert 2000). Because the bones in common between these taxa show an overall similarity, they are described and analyzed jointly here. The lower limb is the best documented, and it shows clear signs of leaping adaptation. A complete femur is known only for Shoshonius, in which it appears relatively short and robust. The proximal femur has an overall similarity in the three genera, bearing a semicylindrical articular surface on its head and a short neck forming an angle of more than 90° with the shaft. The articulation of the head is more cylindrical and the neck angle close to 90° in living specialized leapers. The morphology of the knee joint particularly well reflects its function. The three genera have a distal femur with an anteroposteriorly very high distum, a deep patellar groove bordered by a prominent lateral ridge. The femoral condylar index of 111 in Omomys and 119 in Hemiacodon (111.5 in Shoshonius for a close index) is similar to that of specialized leapers. Omomys has a retroflexed tibial plateau and a mediolaterally compressed proximal shaft. All of these characters are typical of specialized leapers, implying that powerful leaping was a component of the locomotor repertoire in these genera (Anemone and Covert 2000). The distal tibia shows an extensive distal articulation between tibia and fibula, which prevents rotation and is found in leaping primates. It seems that this articular surface is more extensive in Shoshonius (around 25–35 % of tibial length; Dagosto et al. 1999) than in Omomys. Concerning foot bones, the three genera have very similar calcanea (also quite similar to those of Teilhardina) (Fig. 19). They all show a moderate elongation of their anterior part (anterior length/total length is 51 % in Shoshonius, 51.4 % in Omomys, 52 % in Hemiacodon; also 53 % in Washakius insignis, Gebo 1988). Their astragali as well are similar to each other and to those of other omomyines: high and narrow trochlea, long neck, spherical head (not in Hemiacodon), no posterior trochlear shelf. The navicular is moderately elongated, as is the calcaneum, which is similarly the case for both in living cheirogaleids. However, in Hemiacodon and Omomys there are also a cuboid and an entocuneiform which are more elongated than in living primates. Elongation of the foot is another indication of leaping; however, the foot of these omomyines is not as extremely elongated as in the living specialized leapers. There is a screw-like articulation between astragalus and calcaneum, and a pivot joint between calcaneum and cuboid, allowing for some rotation of the foot. Both the entocuneiform and its sellar articular facet, and the first metatarsal with its enormous peroneal tubercle (long, tall, mediolaterally narrow), reflect a foot with powerful hallucial grasping. The pelvis, partially known in Hemiacodon and Omomys, reveals in the latter a shorter ilium and a longer ischium than in living prosimians. The ischium is also less dorsally expanded than in leaping prosimians. These proportions might be more primitive (they recall tree shrews) and/or indicative of a more generalized form of leaping than in living leapers, using horizontal and oblique supports more than vertical ones (Anemone and Covert 2000).

Fig. 19

Three foot bones of Omomys carteri. Calcaneum (a), astragalus (b), and navicular (c), all in dorsal view. The calcaneum shows the two articular facets on which the astragalus lies and a marked distal elongation (and small artifacts of casting). The astragalus shows the long and salient lateral rim of the tibial trochlea. The well-grooved trochlea and calcaneal and navicular elongation are interpreted as reflecting a moderate leaping adaptation

The forelimb is less well documented. A complete humerus is known only in Shoshonius. It has a relatively round head, strongly developed attachment areas for the shoulder muscles, and a prominent brachial flange. The trochlea is long and low. The elbow joint, like the shoulder joint, is that of a quadruped and leaper (Dagosto et al. 1999). The presence of a complete humerus and femur allows an estimation of the humero-femoral index at 64.6 for Shoshonius (and of an intermembral index probably between 64 and 68). Most of these anatomical characters suggest that these omomyines were similar in their locomotor repertoire to cheirogaleids or to the more frequently quadrupedal galagos (Otolemur, Galagoides). They probably used powerful leaping, although quadrupedalism and climbing were also important parts of their locomotor repertoire. There are certainly more small differences between them than the few mentioned above, and more precise interpretations in terms of frequencies of behavior or support use are difficult to obtain (Dagosto 1993). The deep slope on the fibular side of the astragalus and the moderate medial rotation of the tibial malleolus indicate that dorsiflexion of the foot was accompanied by only a slight degree of abduction of the foot, contrary to what is observed in strepsirhines (Dagosto 1985).

European Microchoeridae

The European Microchoeridae are numerous enough to deserve recognition as a family. They had long been suspected to be rooted in the Earliest Eocene Teilhardina belgica; this hypothesis received a recent confirmation through the discovery of two Early Eocene intermediate species, of the genus Melaneremia. M. schrevei is the older and more primitive (Hooker 2012). Compared to T. belgica, it shows a lowering of the protoconid of P/4, more accentuated on P/3, a broadening of P/4, especially of its posterior part, and a more developed paraconid and metaconid on that tooth, more developed cingulids on the lower molars, a marked broadening of the M/3 talonid basin and third lobe, and incipient modifications of the M/2–3 trigonid. The younger M. bryanti has a more reduced P/3 (Hooker 2007). These species provide a link with the genus Nannopithex, which is known through many species in the late Early and Middle Eocene. Two crushed crania and several almost complete lower jaws are known in species found in the Lutetian lignite mines of the Geiseltal, Germany. They show that these species had a very enlarged lower anterior incisor, followed by a reduced I/2, a canine, and two premolars. Compression of the teeth located between I/1 and P/4 led to the loss of P/2 and the reduction of P/3, which is single-rooted. The resulting dental formula, 2123 for lower teeth, will remain stable in later microchoerids. In the upper dentition, I1/ is enlarged, followed by a smaller I2/ and a larger canine; in some of these Nannopithex P2/ is lost, whereas it will still be present in later necrolemurines. The oldest species, the late Early Eocene N. zuccolae, shows a number of characters typical of Nannopithex: enlarged P/4 with ventrally expanded enamel on its labial side, very small metaconid and curved paracristid without paraconid; lower molars with trigonid becoming narrower on M/2–3, smaller paraconid slightly labial and joined to the metaconid summit by a crest; large M/3 talonid basin posteriorly extended in the broad third lobe, and some enamel wrinkling. On the upper teeth, P4/ appears transversely extensive (as much as M1/), with a protocone lobe slightly narrower than the labial part and a low protocone. The upper molars have small conules, with a postmetaconule-crista joining the metaconid summit, whereas a postparaconule-crista is only variably present; there is no lingual cingulum or hypocone, and a protocone fold is variably present. Among evolutionary tendencies observed in Nannopithex species are a progressive broadening of the protocone lobe of P3/, on upper molars an increase in the size of the conules and development of supplementary crests on the walls of the trigon basin, strengthening and isolation of the protocone fold (“Nannopithex-fold”), and development of a small cingular hypocone on the posterior cingulum; on M/2, the paraconid continues to shrink into a continuous paralophid. These characters are best expressed in N. filholi from Lissieu, an MP 14 locality (Godinot et al. 1992). Three other species, N. raabi, N. humilidens, and N. barnesi, are known in the Geiseltal sequence (Thalmann 1994). The genus Nannopithex is probably a paraphyletic stem genus, in which most later clades of microchoerids originate.

The genus Vectipithex has been erected for three species known almost exclusively by isolated teeth (Hooker and Harrison 2008). They show that species phenetically close to Nannopithex survived in northern European localities until the Late Eocene, at a time when Pseudoloris and microchoerines were abundant in more southern regions. The oldest one, V. quaylei from Creechbarrow (Bartonian, MP 16), retains very large anterior incisors, a transversely broad P4/, and a P3/ with a small protocone lobe. It is advanced over Nannopithex through its M/1 without paraconid and its upper molars having an almost complete lingual cingulum with crestiform hypocone. M2/ is lingually narrow, whereas M1/ is broader in its lingual part. The Late Eocene V. smithorum is larger and has a lingually narrower P4/ and somewhat transversely shorter upper molars with slightly larger hypocones. The Late Eocene V. ulmensis from Ehrenstein in Germany is known by a few isolated teeth, which are again larger and have more massive proportions. In their phylogenetic analysis, Hooker and Harrison (2008) find Vectipithex species to be rooted close to N. raabi and thus transfer the latter species into Vectipithex.

Species of Necrolemur and Microchoerus form a morphologically tight group, abundant in the late Middle and Late Eocene. The species Necrolemur antiquus has long been known by beautiful crania found in Quercy fissure fillings (Filhol 1874; Stehlin 1916). Its orbits are relatively large, an observation which was used to support tarsiiform affinities; however, these orbits are like those of living strepsirhines, and are interpreted as reflecting nocturnal habits. A posterior expansion of the auditory chambers is so large that it forms a “mastoid bulla” which is salient on each side on the back of the cranium (Fig. 20). The basicranium of N. antiquus was studied in detail by Szalay (1975). It has since served as one of the best-known references for omomyiform basicranial characters (see below on Haplorhini and tarsiid sister groups). Necrolemur differs from species of Nannopithex by more squared upper molars, especially M1/, linked to a larger cuspidate hypocone. It has crenulated molars and a duplicated metaconule. Lower molars have a trigonid which becomes shorter and transversely broader from M/1 to M/3. Contrary to Nannopithex, the M/3 is short with a reduced third lobe, and almost rectangular with trigonid and talonid of similar breadth. M/1 retains a large paraconid. A progressive increase in crenulation and upper molar lingual breadth can be observed from the primitive unnamed species of Egerkingen to intermediate forms from La Bouffie, to Late Eocene forms closer to the N. antiquus type specimen, which has stronger crenulation and a larger hypocone on M1/ and M2/ (Godinot 2003). Such a crenulated dentition has no analogue in living prosimians; it presumably suggests an adaptation to some kind of abrasive food. Postcranials of Quercy N. antiquus include an elongated femur with high distal extremity, a fused distal tibia and fibula and an extremely elongated calcaneum, such as occur only in Tarsius among living primates, and an astragalus with a deeply grooved trochlea which was tightly maintained between the tibial and peroneal malleolae (Schlosser 1907; Godinot and Dagosto 1983). All of these characters typify an extreme leaping adaptation, unique until now in the Eocene fossil record, as is found in living forms which essentially move by long jumps between vertical supports (VCL). Vertical clinging is confirmed by the position of the foramen magnum below the cranium, visible in a larger Necrolemur skull.

Fig. 20

One of the best-preserved crania of Necrolemur antiquus from the old Quercy collections (Montauban) in lateral (a) and ventral (b) views at two different scales. On the ventral side, the tympanic bullae have been prepared; their posterior extensions, or mastoid bullae, are preserved; anteriorly, alveoli show the presence of two well-separated and large anterior incisors. The lateral view shows that the muzzle of Necrolemur is anteriorly elongated

Species of Microchoerus are essentially large Necrolemur with more crenulated enamel, which develop a mesostyle on their upper molars. Two species are known in the Bartonian locality of Creechbarrow, England (MP 16), M. wardi and M. creechbarrowensis. Hooker (1986) related these two species to the later occurring M. erinaceus (Hordle, MP 17a, England) and M. edwardsi from an unknown level in the Quercy region, which are also large, heavily crenulated, and display the most strongly molarized p/4 of all, with big and high metaconid and paraconid. However, the study of assemblages from Quercy in biochronological order revealed a continuous lineage, which increases in size and molar crenulation and develops a mesostyle. This lineage links a species of Necrolemur cf antiquus (stage of La Bouffie) to assemblages extremely close to M. erinaceus (Godinot 1985). These hypotheses will have to be restudied with quantitative approaches. There are convergences in the increasing molar crenulation of different lineages. This process culminates with the extraordinary M. ornatus described by Stehlin (1916), which has highly tuberculated upper teeth.

Species of Pseudoloris are small and developed pointed teeth, adapted to insectivory, which are convergent with those of Tarsius. However, the anterior dentition retains the stamp of the family, with moderately enlarged anterior incisors of typical morphology and single-rooted P/3 somewhat crowded under the P/4 (it probably lost the reduced I/2 of others). The muzzle of Pseudoloris shows proportionately large orbits, due to its small size, and this makes it even more similar to Tarsius (see discussion of tarsiid sister groups below). The best-known species is P. parvulus from the Quercy fissure fillings, which appears long-lived. Early assemblages from Le Bretou show lower molars with a relatively open trigonid, whereas later assemblages usually show them with a longer paracristid more or less closing the trigonid anteriorly. M/3 have a narrow third lobe. P/4 remains very simple, with only an incipient metaconid. Upper molars have a small paraconule and a well formed metaconule with well formed pre- and postmetaconule-cristae. There are small variations in the size of the hypocone, which in any case remains small. Larger species known by fragmentary remains are documented: P. crusafonti from Grisolles, France, and P. reguanti from Sant Cugat de Gavadons, Spain (Louis and Sudre 1975; Minwer-Barakat et al. 2013). New species of Pseudoloris have been described from Spain: P. cuestai and P. pyrenaicus (Minwer-Barakat et al. 2010, 2012). P. cuestai is typified by I/1 quite unlike those of other microchoerids, which calls for further analysis. One species, P. godinoti, has been demonstrated to have survived the Late Eocene climatic deterioration, being present in the lower Oligocene of Spain (Köhler and Moyà-Solà 1999). The oldest species sometimes referred to Pseudoloris deserve to be placed in a different genus, Pivetonia, with P. isabenae known in Spain and P. saalae known in the Geiseltal (Thalmann 1994). These very small species are phenetically closer to Nannopithex species, suggesting a rooting of the Pseudoloris group as well in the stem genus Nannopithex. Pseudoloris has been mentioned in the Late Eocene of Nei Mongol (Wang 2008), but the lower molar referred to “P. erenensis” is too different from the European genus. It belongs to another taxon.

The species Paraloris bavaricus has been found in Late Eocene marine sediments. Its lower jaw with four teeth shows a unique combination of characters: posteriorly narrowing P/4 without metaconid, smooth enamel on lower molars, M/1 with moderate-sized paraconid, relatively short paracristid on M/2–3 (Fahlbusch 1995). It is unlike all other genera, possibly due to branching very early (before Melaneremia for Hooker and Harrison 2008). It shows that undocumented lineages existed in unsampled regions of Europe, e.g., towards the east.

Possible Asiatic Omomyiforms

Other omomyiforms have been described in Asia. The existence of Teilhardina asiatica, described above with other species of Teilhardina, underlines the fact that a diversification of omomyiforms in Asia can be expected. However, several species are known only by very fragmentary remains. These were first compared with European and North American taxa, but they most probably document Asiatic groups. Baataromomys is known by one lower molar from the Early Eocene of Inner Mongolia, close to that of a Teilhardina (Ni et al. 2007). As mentioned above, “Pseudoloris” erenensis from younger Eocene strata of the same country is not a microchoerid (Wang 2008). Is it an omomyiform, or an eosimiid? Vastanomys, from the Early Eocene Vastan mine in India, has an omomyid-like M/2; but a referred upper molar, transversely elongated, is also reminiscent of eosimiids (Bajpai et al. 2005). More material is eagerly awaited. Among several species from the late Early Eocene of Pakistan, Kohatius is known by isolated M/1 and P/4, and Indusius by isolated M2/ and M/2 which are somewhat reminiscent of microchoerids (Russell and Gingerich 1987; Gunnell et al. 2008). These fragmentary fossils confirm that a vast proportion of Eocene Asiatic primates, particularly the omomyiforms, are still unknown.

The Early Eocene Altanius from Mongolia is better documented. It is an extremely small species, with the primitive dental formula of 2143 below and probably above (Fig. 21). Alveoli show a canine larger than the two small incisors below, and larger than the anterior premolars above and below. P/4 through M/3 are superficially omomyid-like, but their protoconids have extended labial slopes, especially on M/1, which recall plesiadapiforms. Its upper molars are transversely extended and bear well-formed para- and metaconules. Their crown is very high in labial view, and a strong protocone fold is added to the protocristae (i.e., not derived from the postprotocrista as in omomyids), contributing to its plesiadapiform stamp. Despite being described as a primitive omomyid (Dashzeveg and McKenna 1977) and considered an anaptomorphine by Szalay and Delson (1979), its affinities are probably closer to plesiadapiforms than to the earliest primates (Rose and Krause 1984). If it were a primate, it would contribute critical information to the reconstruction of their primitive dental morphotype. A better understanding of its dental adaptation and affinities would be welcome.

Fig. 21

Altanius orlovi from the Early Eocene of Mongolia, upper (a) and lower dentition (b): it is a very tiny species. Photographs of white-coated epoxy casts, both at the same scale

Tarsiidae, Possible Tarsiiformes, and the Concept of Haplorhini

The Concept of Haplorhini

Among living primates, shared derived characters of soft anatomy, reproductive organs, and genetic markers unite tarsiids and simians in a monophyletic clade Haplorhini. The name comes from characters of the nose and face: tarsiers and simians have lost the moist rhinarium typical of strepsirhines and other mammals, and replaced it with dry and hairy skin. Anatomists have looked for osteological characters, which would allow the recognition of fossil haplorhines. The most attractive of them, the presence of a postorbital septum isolating the eye from the temporal cavity, has turned out to be problematic. The orbital wall of simians consists of the frontal, alisphenoid, and a large zygomatic, whereas that of tarsiers consists of the frontal, alisphenoid, a small zygomatic, and a large amount of maxillary bone. Because the constituents are not the same, some researchers have argued that the two septa must have evolved convergently (e.g., Simons and Russell 1960; Simons and Rasmussen 1989), whereas others have argued that sutures can move through time, and that a zygomatic-alisphenoid contact is so exceptional that it must be at least partly homologous (e.g., Cartmill 1980). Recent embryological work concerning the septum shows that of tarsiers to be formed by a processus different from that of simians: no frontal spur, and ossification of a membrane (Smith et al. 2013). This seems to support the convergence hypothesis, even though some paleontological confirmation would be welcome to eliminate any kind of common precursor morphology. In any case, detailed anatomical work has found other derived anatomical characters supporting a clade (Tarsius + simians) in the basicranium, a region which has provided characters linked to higher systematic categories. MacPhee and Cartmill (1986) provided six such characters from the otic region, which subsequently have been elaborated (Ross 1994; Ross and Covert 2000). Agreed upon are the presence of a septum isolating an anterior tympanic accessory cavity, the anteromedial location of the posterior carotid foramen, the perbullar carotid pathway (through the petrosal septum), the extrabullar ventral edge of the tympanic bone (no annular bridge), and a highly reduced or absent stapedial artery.

To the above list can be added characters more or less directly dependent of anatomical haplorhinism. The absence of an interincisor diastema, between the roots of the two I1/, is correlated with the loss of the rhinarium, which requires such a space for its philtrum (Beard 1988). Haplorhinism had been correlated with a smaller infraorbital foramen indicative of a reduction of the sensory organs of the muzzle, the vibrissae (Kay and Cartmill 1977). However, this has recently been refuted in favor of a different adaptive meaning, related to diet (Muchlinski 2010a). Haplorhinism is likely correlated with an abbreviated snout, and some reduction of the nasal cavity and olfactory organs. Linked to this is a reorientation of the nasolacrimal canal, which is anteriorly inclined in strepsirhines, and shorter and vertical in haplorhines. A detailed embryological study showed that the vertical canal and duct in tarsiers and simians is acquired by exactly similar embryological processes, giving a strong argument in favor of their homology, and of a monophyletic Haplorhini (Rossie and Smith 2007).

The whole process of the acquisition of anatomical haplorhinism may have been driven by an emphasis on vision in the lineage leading to haplorhines. Tarsiers and simians possess a fovea in their retina, which considerably increases visual acuity. They also lack the tapetum lucidum, a reflecting membrane behind the retina which enhances vision in low levels of light and is present in living nocturnal strepsirhines. This lack explains why tarsiers possess such enormous eyes, larger than those of other nocturnal prosimians. Some of the peculiar cranial characters of tarsiers are probably related to these enormous eyes, starting with their postorbital septum. The fovea in the eye was probably acquired to enhance vision in diurnal species. Hence the early haplorhines can be reconstructed as diurnal, possessing relatively small orbits (Cartmill 1980). Tarsius became secondarily nocturnal and in the process acquired its enormous eyes and other autapomorphies.

The preceding list of characters strongly supports a clade Haplorhini, based on the derived characters shared by tarsiers and anthropoids, even if some of them turn out to be difficult to use (e.g., the quantitative ones) or include partial homoplasy. However, these numerous characters were acquired sequentially, and fossil evidence will be needed to decipher in what order. Some characters used to define haplorhines in the past now appear problematic; for example, the apical passage of the olfactory nerves above an interorbital septum was thought to characterize a more inclusive group (omomyiforms + crown haplorhines) until it was found lacking in the simiiform Aegyptopithecus (Simons and Rasmussen 1989). Because known omomyids lack most of the above-listed characters, it is unwise to extend to them a definition of haplorhines, which would thereby lose its content. Alternatively, if the washakiins can be shown to share enough of these characters, they might be included in tarsiiforms and character-based haplorhines. An alternative view is supported by those who propose a sister group relationship between some omomyiforms and Tarsius. Such a view implies that anatomical haplorhinism was acquired convergently in tarsiids and anthropoideans – an unparsimonious assumption, which would require grounding in a convincing fossil record, or else refutation of the homologies listed above. This discussion will be picked up again in section “Proposed Sister Groups for Tarsiids” below.

Fossil Tarsiidae

The family Tarsiidae has long been absent from the Paleogene fossil record. However, the discovery of teeth extremely similar to those of the living tarsier led to the description of Tarsius eocaenus in the Middle Eocene fissure fillings from Shanghuang, China (Beard et al. 1994). These five isolated cheek teeth were completed by the description of a piece of maxilla with P3/, coming from a different fissure but also very similar to the living Tarsius (Rossie et al. 2006). Specifically, this small specimen shows two similarities with Tarsius: a relatively reduced infraorbital foramen and a short vertical nasolacrimal duct. As seen above, the former is not as diagnostic as formerly believed; the latter, however, really seems to turn the specimen into an “anatomical haplorhine” (Rossie et al. 2006), i.e., a haplorhine in the sense used here. By contrast, contemporary omomyiforms appear markedly different.

Tarsal bones found in the Shanghuang fissure fillings have also been ascribed to tarsiids (Gebo et al. 2001). The calcaneum is distally incomplete; but it appears elongated in its distal part, and also progressively narrower distally, which makes it more similar to tarsiers than other Shanghuang calcanea (see Fig. 35). An astragalus is similar to that of Tarsius in its wide body bearing a wedge-shaped trochlea, a short neck, and a low neck angle. Many of these characters suggest a marked leaping adaptation. However, the astragalus also has a small trochlear shelf, whereas Tarsius is lacking such a shelf. Some other differences with Tarsius led the authors to conclude that these tarsals belong to a genus other than Tarsius (Gebo et al. 2001). Possibly Shanghuang dental tarsiids shared the nocturnal adaptation of Tarsius and associated anatomical characters, but not yet the full vertical clinging and leaping (VCL) specialization of living tarsiers.

Another Middle Eocene genus found in the Heti Formation of the Yuanqu Basin (Shanxi Province, China) has been ascribed to the tarsiids. Xanthorhysis tabrumi is represented by a tiny lower jaw bearing P/3–M/3 and alveoli for a P/2 and a larger canine (Beard 1998). P/3 and P/4 have a pointed protoconid and a very simple and reduced talonid. P/4 has a small metaconid, a weak and almost complete labial cingulid, and a thicker anterolingual cingulid. The lower molars are primitive in possessing a trigonid with strong paraconid. The trigonid is similar in M/1–M/3, as is typical of Tarsius. The talonid basin is broader than the trigonid, and the entoconid is well formed. M/3 has a very short third lobe, much smaller than in Tarsius, similar to that of Eosimias but more pinched. The dentition shows a striking overall similarity to that of Eosimias, a genus studied below.

Proposed Sister Groups for Tarsiids

Comparisons of small Eocene fossil primates with Tarsius have been conducted for a long time. At the beginning of the twentieth century several authors held the view that omomyids were not distinct enough to be placed in a different family, and they were thus ascribed to Tarsiidae (e.g. Matthew in Matthew and Granger 1915). However, there were dissenting views as well. Central to these discussions were the crania of Tetonius and Necrolemur. For example, Stehlin (1916) gives a detailed explanation of why he does not follow Gregory in placing Necrolemur in Tarsiidae, and why he prefers to maintain a family of “Necrolemuridae” (=Microchoeridae); he argues that Necrolemur and Microchoerus are dentally too specialized and thus follow a divergent evolutionary line (also shown by the mastoid inflation) – a point made earlier by Schlosser (1907). A revived defense of the close affinity of Necrolemur and Tarsius was put forth by Simons (1961), who nevertheless used the family “Necrolemuridae.” As seen above, the matter has become more complex in recent decades, with more omomyid crania discovered, tarsiids identified by the Middle Eocene, Eosimiidae entering these discussions (see below), and renewed anatomical research. Clearly, however, the view that tarsiids could be part of a monophyletic Tarsiiformes that includes a number of omomyiforms (and particularly microchoerids) stands in sharp opposition with the concept of Haplorhini as endorsed here. As seen above, it would imply the convergent acquisition of haplorhinism in tarsiiforms and simians, a far-reaching conclusion.

The two groups which continue to be central to this debate are the microchoerids with Necrolemur and Pseudoloris, and the washakiins with Shoshonius. The large orbits of Necrolemur have been emphasized, but they appear to be as in nocturnal strepsirhines of similar size, not tarsier-like. The characters linked to an extreme leaping specialization, reaching the stage of tibio-fibular fusion, also produce a list of shared derived characters that can lead to Necrolemur being the sister group of Tarsius in some parsimony analyses (Dagosto et al. 1999). However, as discussed long ago by Stehlin (1916), leaping specializations are numerous in primates, so that convergence is easily possible. The most powerful arguments in favor of the Necrolemur-tarsier hypothesis were reassessed by Rosenberger (1985), who added the peaked shape of the choanae, the pterygoid fossa encroached by the bulla, and recalled the significance of the extensive laminar contact between pterygoid and bulla, the shape of the guttered temporo-mandibular joint, and some less striking characters. These shared derived characters have been given as evidence in favor of close Necrolemur-tarsier phylogenetic relationships. However, as seen above, Necrolemur has anatomical strepsirhinism, a relatively long snout, and orbits of “normal” nocturnal size. Also, as emphasized by earlier authors, Tarsius is very unlikely to be derived from a species having the typical microchoerid dental formula; it possesses an anterior accessory auditory chamber, whereas the microchoerids developed their posterior mastoid inflation, demonstrating that they belong to two divergent phylogenetic lineages.

Pseudoloris plays a role in the debate because it shows additional similarities with Tarsius in dental traits linked to insectivory. Its position as sister group of Tarsius sometimes reappears in parsimony analyses of very large data sets (e.g., Ni et al. 2013). However, inasmuch as a position of Tarsius nested within microchoerids is impossible, this result shows that parsimony analyses can be driven by overall phenetic similarity, to the detriment of real phylogenetic affinities. Taxon sampling to date has not been complete enough to allow resolution of the debate. Necrolemur and Pseudoloris appear well inserted in the European microchoerid radiation, which has an almost continuous history going back to Teilhardina belgica, and in which the distinctive dental formula was acquired very early. They are unlikely to be sister group of tarsiids, but raise good questions concerning the acquisition of the above-mentioned synapomorphies: were these inherited from Teilhardina on, or did they converge under similar adaptive pressures (especially on vision)?

Another candidate for the sister taxon of Tarsius is the North American omomyine Shoshonius (Beard et al. 1991). After careful reevaluation of many cranial characters, Beard and MacPhee (1994) find a number of similarities shared with Necrolemur (presence of a parotic fissure) and other characters more specifically shared with Tarsius (e.g., enlarged orbits, abbreviated snout, narrower annular bridge). Furthermore, Rossie et al. (2006) briefly mention that Shoshonius possessed a vertical nasolacrimal canal, like Tarsius. This might make Shoshonius a real tarsiiform and character-based haplorhine, or it might be an argument in favor of convergent acquisition of haplorhinism. Then again, Shoshonius does not show the beginning of a postorbital closure similar to that of Tarsius, its dentition is not especially similar to that of tarsiids, and its postcranial characters are much less specialized for leaping than in tarsiids: their phylogenetic relationship cannot be very close. Also, such a relationship would call for an analysis of the washakiin tribe, in which all specialists place Shoshonius: are the washakiins well nested within the omomyine radiation? The omomyines seem to represent a relatively well-documented North American diversification, rooted in Steinius; the Tarsius–Shoshonius relationship would thus imply a North American origin for the tarsiids, which would have to be descended from a washakiin dispersal into Asia. Other omomyin tribes, when known, appear not to have acquired hypertrophied orbits and other characters linking them so closely to Tarsius. Another possible scenario is that the washakiins represent immigrants from Asia, pertaining to a group closely related to tarsiids. These scenarios need to be further explored and tested.

To conclude, if Tarsius were shown to be the sister group of more primitive omomyiforms, and not of early anthropoideans, this would imply the convergent acquisition of haplorhinism in the two groups. Haplorhinism would become an adaptive grade, reached by two lineages and more or less approached by others, under the adaptive emphasis on vision in small species (snout and nose reduction). Until now, the evidence does not appear sufficient to adopt this gradistic view, to the detriment of the haplorhine clade.

Archicebidae

For the recently described Archicebus achilles, from the Earliest Eocene of China, a new family Archicebidae has been erected, placed by the authors in their extended concept of Tarsiiformes (Ni et al. 2013). Whereas the placement of this fossil in early tarsiiforms or early omomyiforms or something else is debatable, its description adds considerably to our knowledge of the earliest primate radiation. It is represented by a crushed skull, isolated elements of the forelimb, and the entire rear part consisting of lumbar vertebrae, pelvis, tail, and both posterior limbs. Its skull length is approximately 2.5 cm, its long tail more than 13 cm, and its body weight is estimated to be 20–30 g. The crushed skull shows orbits of a relatively small size, indicative of likely diurnal habits. The skull shape is described as close to that of Teilhardina asiatica. The snout is said to be very short, as in Tarsius and some omomyids. The dentition is close to that of the other primitive primates; however, it shows a derived single-rooted P/2, and P1 and P2 appear reduced in comparison with adapiforms. The skeletal elements show elongated hindlimbs, long tibiae and long metatarsals, and other characters linked to a moderate leaping adaptation. Remarkably, the calcaneum of A. achilles is broad in its distal half, and is described as anthropoidean-like. Several of its tarsal characters are admitted as primitive in primates. The importance of this fossil lies, among other aspects, in the documentation of a calcaneum strongly different from that of Teilhardina–more primitive and associated with a moderate leaping adaptation – which substantially adds to our knowledge of early primate locomotor adaptations.

Eosimiidae

The Asiatic family Eosimiidae has provided crucial information and challenge concerning the early primate radiation (Beard et al. 1994, 1996). This family now includes three genera from the Middle Eocene of China and Thailand – Eosimias (three species), Phenacopithecus (two species), and Bahinia – and possibly a fourth one from the lower Oligocene of Pakistan – Phileosimias. The upper molars of Eosimias and Phenacopithecus are extremely primitive, transversely elongated with a marked waisting lingual to the labial cusps (Fig. 22); this waisting, very marked on the posterior border of M1/, is also present on M2/, and is slightly marked on the anterior border of both teeth, which is exceptional (Beard and Wang 2004). On the labial side, there is a well-formed parastyle, as well as some labial extension of the crown labially to M2/, showing the remnant of a stylar shelf. These proportions give to these upper molars an extraordinarily primitive stamp. The complete lingual cingulum is derived in comparison with Teilhardina and Donrussellia; however, the absence of a hypocone by Middle Eocene time is very archaic in comparison with most other fossil primates. The upper P3/and P4/have narrow protocone lobes, presumably also primitive. The lower dentition as well shows many primitive characters. It is best documented on the two lower jaws of Eosimias centennicus, which show the complete lower dentition and aspects of the posterior part of the mandible (Beard et al. 1996). The incisors are small; I/2 shows one pointed cusp and I/1 was probably similar. Both incisors have a complete lingual cingulid. The canine is large, robust, and recurved anteriorly. A cingulid starts at the apex, descends posteriorly, and terminates as a posterior basal cingulid. There is a small, single-rooted P/2 bearing a single cusp and a posterior and lingual cingulid. P/3 and P/4, quite similar to each other, are massive, broad, and long premolars. P/3 is simple. Its massive protoconid has a curved preprotocristid in profile view and is surrounded by an almost complete cingulid. In occlusal view, its base shows an anterolabial bulge, associated with a labially shifted anterior root. P/4 is similarly massive and is more molarized, bearing a metaconid, a paraconid linked to the latter, and a more extended talonid. P/3 and P/4 have an almost complete labial cingulid, and on the labial side the enamel is ventrally extensive below the cingulid (exodaenodonty, easy to see on the anterior view of the isolated P/4 of Phenacopithecus xueshii; Beard and Wang 2004, Fig 16). Intraspecific variations in dental detail are described by Beard and Wang (2004). One of their specimens illustrates very well a degree of premolar crowding, by which P/3 is canted anterodorsally and its base is crowded below the anterior part of P/4, exactly as in the tarsiid Xanthorhysis. The main cusps of P/3 and P/4 are aligned with molar cusps as usual, but the bases of these cusps are anterolabially expanded, linked to a labial shift of their anterior root (Fig. 22). The three lower molars of Eosimias display a combination of high cusps recalling insectivorous species and rounded summits giving them a slight bunodont touch. They have a broad and elongated trigonid, bearing a big paraconid well separated from the metaconid. In occlusal view the trigonid is almost exactly the same from M/1 to M/3, a situation typical of tarsiids and exceptional in other Eocene prosimians. On the type specimen the notch on the protocristid is deeper (a deep V) on M/3 than on M/2. There is some waisting on the labial side of M/2 (ectoflexus) between the trigonid and the talonid. The talonid basin is deep, transversely broad, and anteroposteriorly short. A hypoconulid is posteriorly salient on M/2, being slightly closer to the hypoconid than to the entoconid. M/3 is noteworthy for the small size of its talonid, which is narrower than the trigonid, and for the small size of its third lobe, which is reduced to a very moderate hypoconulid (slightly less salient than in Xanthorhysis). The dentaries of Eosimias are relatively high and robust, with a vertical and anteroposteriorly short symphyseal region. The posterior part of the dentary of E. centennicus shows a rounded angular region and a small coronoid process.

Fig. 22

Dentition of Eosimiidae. Mandibles of Eosimias centennicus in occlusal (a and b) and lingual (d and e) views. Maxilla of Phenacopithecus krishtalkai in occlusal view (c). Upper and lower molars are remarkably primitive, the two mandibles illustrate an interesting intraspecific dental variability, and premolars in (e) are reminiscent of Xanthorhysis (Courtesy of K. C. Beard)

Some important facial characters of eosimiids were revealed by a species of Phenacopithecus, another genus found in the Middle Eocene beds of the Heti Formation of Shanxi (Beard and Wang 2004). These species are close to Eosimias: they are larger and differ by some details of P/4 and other teeth. A maxilla of P. krishtalkai shows that this bone is high between the orbital margin and the alveolar rim. The orbits seem to have been small, suggesting diurnal activity. The infraorbital foramen is also relatively small; however, this is no more proof of haplorhinism.

Another eosimiid has been found in the late Middle Eocene Pondaung Formation of Myanmar, Bahinia pondaungensis (Jaeger et al. 1999). Two associated maxillae and a partial mandible revealed many dental characters of this species, which is larger than Eosimias and Phenacopithecus. The upper canine is especially large, with a rounded outline, and roots show that the upper incisors were vertically implanted, I1/being slightly larger than I2/. This morphology is said to resemble Tarsius more than the omomyids (Jaeger et al. 1999). The upper premolars have a lingual lobe but no differentiated protocone. The upper molars show well-developed protocristae, one joining the metacone, the crista obliqua, and another quite rare in primates joining almost directly the paracone (Fig. 23). These upper molars also have a thick labial and a continuous, strong lingual cingulum, with M2/ showing a slight thickening of the cingulum that represents an incipient hypocone. The P/2 is larger than in Eosimias. P/3 and P/4 are large, having roughly the same size. P/4 has no metaconid or paraconid, in contrast to Eosimias. The right maxilla allows the delineation of an orbit of small size, a relatively high distance between the ventral orbital rim (with exposed maxilla) and the alveolar margin, and a small zygomaticofacial foramen (Kay et al. 2004b).

Fig. 23

Left part of the palate of Bahinia pondaungensis showing the broken canine, very small P2/, P3/, damaged P4/, M1/, and M2 (Courtesy of R. L. Ciochon)

Another eosimiid genus has been described from Early Oligocene beds of Pakistan (Marivaux et al. 2005). Phileosimias kamali is represented by eight isolated teeth (and the very close, possibly conspecific, P. brahuiorum by two upper molars). The upper molars differ from other eosimiids by well-formed conules, lingually rounded paracone and metacone (no crista obliqua), and a cuspidate hypocone on the M2/ of P. brahuiorum. The lower molars seem to have a more bunodont appearance and a somewhat reduced paraconid, slightly shifted lingually in comparison with Eosimias. The M/3 has a larger talonid with a broader third lobe than in Eosimias. The P/4 has a small crestiform metaconid and is narrow in its posterior part. P/3 and P/4 are slightly exodaenodont. If the affinities with more typical eosimiids are confirmed, this later taxon would illustrate a trend opposite to that seen in Bahinia: it increased bunodonty and the development of conules, whereas Bahinia increased its crests. Its smaller P/3 and P/4 also appear to represent a trend opposite to what is observed in Eosimias. The mention of an Early Eocene eosimiid “Anthrasimias gujaratensis,” in the Vastan Mine, India, is a misidentification (Bajpai et al. 2008; Rose et al. 2009).

Among the tarsals found in the Shanghuang fissure fills, one group is attributed to the eosimiids (Gebo et al. 2000, 2001). These tarsals present calcanea with whole proportions intermediate between those of the omomyids and tarsiids, which are narrower, and those of the anthropoideans, which are broader (see Fig. 35). They present a moderately elongated distal part (45–62 %), a long heel (28–36 %) and a relatively long and narrow posterior calcaneal facet. These characters suggest a mixture of quadrupedalism and leaping. Distally, these calcanea present a cuboid facet that is relatively flat, with a wedge removed from the medioplantar region. The astragali ascribed to eosimiids are moderately high, and also have a moderate neck angle. They possess a posterior trochlear shelf. Astragalar characters concur with those of the calcanea to suggest quadrupedalism, and more leaping than in platyrrhines (Gebo et al. 2001).

The phylogenetic affinities of Eosimiidae have been debated since their discovery. Many authors have followed Beard and collaborators, who defended close affinity with anthropoideans (an anthropoidean status with a stem-based definition of that group, or a stem anthropoidean or protoanthropoidean status based on our conventions), e.g., Kay et al. (1997), Marivaux et al. (2005), Williams et al. (2010). However, others have disagreed (Godinot 1994; Gunnell and Miller 2001; Simons 2003). The noteworthy aspects of the eosimiid dentition that need to be evaluated foremost are some of its extraordinarily primitive characters. Most authors agree that the transversely elongated upper molars, with the remnant of a stylar shelf labial to the metacone, marked waisting in the median part, and complete absence of a hypocone, are primitive characters – with the first two of these being more primitive than in Teilhardina, and thereby modifying the concept of the primitive primate dental morphotype. Such is also the case for the very large paraconid of the lower molars, especially on M/3. These observations raise further questions: Could the short M/3 talonid also be primitive in primates? Could the lower metaconid relative to the protoconid also be primitive by comparison with a primitive out-group, contrary to Gunnell and Miller (2001), who consider an equal height of these cusps likely to be primitive? These and other important questions will have to be scrutinized in the future. In any case, eosimiid dental features are leading researchers to modify the primitive primate dental morphotype, which plays a critical role in primate phylogenetic analyses.

Other characters that have been considered shared-derived with anthropoideans include P/3 and P/4 that are slightly exodaenodont and show an anterior root more labial than the posterior one. These features are shared with several early African simians; an oblique though not exodaenodont P/4 is considered part of the African simian morphotype by Gunnell and Miller (2001). Furthermore, these characters appear on P/3 and P/4 of very large size relative to the molars, larger than in any early African simian. Whereas the latter have a broad and short P/4 and a relatively large P/2, Eosimias has long, anterobasally inflated P/3–4 reflecting a peculiar adaptation emphasizing these big premolars, associated with a reduction of its P/2. This suggests the possibility of a partial functional convergence leading to labial reinforcement and some exodaenodonty of P/3–4 (probably also to a high and robust dentary). The vertically implanted lower incisors and the short vertical symphyseal region are shared with anthropoideans, but they can also be expected in haplorhines having acquired an abbreviated snout. As these incisors are very small and pointed, whereas those of anthropoideans are spatulate and usually larger, there is little evidence of characters exclusively shared with anthropoideans; these observations more probably indicate a haplorhine grade. It is difficult to find convincing phylogenetic signals in the very primitive molars of Eosimias. There is some convergence between more crested genera, as e.g., in the eosimiid Bahinia and the anthropoidean Catopithecus, which acquired a complete crista obliqua and an almost continuous crest linking protocone and paracone. However, known lower teeth of Bahinia are very different from those of oligopithecids, making a close link between them unlikely. The bunodont trend in eosimiids possibly led to the Oligocene Phileosimias, which is far from any known African simian. In the absence of indisputably derived characters homologous with those of African simians, are there other possible phylogenetic links? A sister group relationship with Amphipithecidae was advocated by Jaeger et al. (1999), and thus is examined below with that group.

A probable close affinity between Eosimias and the tarsiid Xanthorhysis has been mentioned above. The lower molars are higher and more pointed in Xanthorhysis. The P/3 and P/4 share a similar global pattern, including even the peculiar crowding between P/3 and P/4 described above. Eosimias appears as a more bunodont version of Xanthorhysis, having increased the size of its P/3 and the anterobasal robustness of its P/3–4. Found in omomyids, such differences would not prevent the placement of two genera in the same subfamily, because the overall pattern is so similar. This is significant and should not be ignored. The resemblance is not a proof of direct tarsiid affinities, however, because Xanthorhysis might turn out to be the most primitive eosimiid. Not enough is known about Tarsius eocaenus and its possible close relatives to decide on this. However, the similarity clearly does point towards close affinities between the two families. In this context, the recently introduced concept of Eosimiiformes appears problematic (Chaimanee et al. 2012).

Certain facial characters mentioned above show that the eosimiids were probably diurnal. Being diurnal is sometimes considered a challenge for possible tarsiiform affinities (Beard and Wang 2004); however, as seen above, early haplorhines are in fact expected to be diurnal (Cartmill 1980; Ross 1996). Thus Tarsiiformes other than tarsiids or tarsiines are expected to be diurnal, as eosimiids are. The simplest scenario, backed up by the strong dental similarities between Tarsius, Xanthorhysis, and Eosimias, is that eosimiids are tarsiiforms, and are character-based haplorhines. An isolated petrosal from the Shanghuang fissure fillings, attributed to an eosimiid, may add to our knowledge of eosimiid cranial anatomy (MacPhee et al. 1995). This petrosal exhibits a large canal reflecting a fully functional stapedial artery. Large and subequal stapedial and promontorial canals are also found in Shoshonius and Omomys; they contrast with another omomyid (Necrolemur) and Rooneyia, in which the stapedial canal is consistently smaller, and even more with the anthropoideans and Tarsius, in which it is further reduced or lost (probably convergently). The authors conclude that for this character, eosimiids probably retained a primitive state. They also conclude that eosimiids preceded the reorganization of the anthropoidean ear region (MacPhee et al. 1995), which agrees with a tarsiiform or an omomyiform view. The postcranial evidence for eosimiids has also been used to argue in favor of anthropoidean affinities. However, there are a number of caveats to be noted. Thus, a wedge in the plantar region of the cuboid facet developed in several primate groups. Astragali of eosimiids have a posterior trochlear shelf, which is absent in anthropoideans. They share with anthropoideans a reduced medial tibial facet, but this character is likely primitive in primates. In other respects they are, in the words of Gebo et al. (2001, p. 96) “unlike anthropoids and resemble omomyids.” Postcranial characters are further discussed below in the anthropoidean origins part. The extreme leaping specialization of the living Tarsius leads one to suspect that primitive tarsiids were less specialized, as seen in the Shanghuang tarsiids; a fortiori, other early tarsiiforms are expected to be differently adapted, more quadrupedal. In conclusion, eosimiids probably fit in among early tarsiiforms (unless the petrosal, if well attributed, indicates omomyiforms).

North American and Asiatic Enigmas: Rooneyiidae, Ekgmowechashala, and Amphipithecidae

Rooneyiidae

The two North American genera Rooneyia and Ekgmowechashala occurred relatively late, in the Latest Eocene and in the Oligo-Miocene, respectively. They were both included in the systematic revision of the Omomyidae by Szalay (1976) and continue to be considered a possible tribe of omomyines, in the case of Rooneyia, and a problematic omomyid, in the case of Ekgmowechashala (Gunnell and Rose 2002). However, their status is questioned in several recent analyses. Rooneyia viejaensis is known by its type cranium from the Late Eocene of Texas, which is well preserved (Wilson 1966). Its upper dentition is bunodont and anteriorly abbreviated. The upper molars bear large conules (and P4/ a large paraconule) and a hypocone almost as large as the protocone on M1/. There are only two upper premolars and a very small canine. The bunodont molars bear similarities with those of anthropoideans, whereas the two premolars resemble only those of catarrhines. However, the small canine together with other cranial characters exclude a close affinity with the latter. Nevertheless the cranium of Rooneyia shows a number of characters which make this genus unique among primates. It shares with omomyids the presence of a tubular ectotympanic. Differences with omomyids include the lack of any inflation of the petromastoid region and relatively small orbits which suggest diurnal habits, contrary to other omomyids (with the exception of Teilhardina). Characters unique to Rooneyia concern its orbits. The large frontal with fused metopic suture extends above the orbit as a partial roof; the lateral frontal process is high and constitutes an incipient postorbital closure; the orbital fossa is funnel-shaped and pushed slightly below the forebrain (Rosenberger 2006). In addition, the orbits of Rooneyia show high degrees of convergence and frontation which position this taxon closer to living anthropoideans than to living prosimians (Ross 1995). These characters led Rosenberger (2006) to propose for Rooneyia a formal status of Protoanthropoidea. The unique combination of characters found in Rooneyia resulted in various placements. The large cladistic analysis of Kay et al. (2004b) led to a preferred tree in which Rooneyia appears nested within the adapiform radiation, as a primitive sister group of a clade (amphipithecids + adapids). However, in the phylogenetic analysis of Ni et al. (2010), Rooneyia appears nested within the washakiins, which seems hard to reconcile with the cranial characters of Shoshonius. In view of these contradictory results, it seems important to acknowledge the uniqueness of Rooneyia by using a family Rooneyiidae, and to keep in mind its important protoanthropoidean-like orbital characters. Because it is known that faunal exchanges between North America and Asia occurred repeatedly, it would be fruitful to pursue comparisons of Rooneyia with bunodont Asiatic primates.

Ekgmowechashala

Ekgmowechashala philotau occurs in the Late Oligocene to Early Miocene of South Dakota and Oregon, more than 10 My later than Rooneyia. It is known by a dentary with P/3–M/3 and upper P4/–M1/ and half of M2/. The dentary is elongated. Upper and lower P4 and molars have a low relief, heavily crenulated enamel, and supplementary cuspules (Fig. 24). The upper molars have large conules and a large hypocone shifted anteriorly, which is linked to the protocone and sometimes interpreted as arising from a protocone fold. The lower molars have a large cuspidate metastylid and a low posteriorly opened trigonid without paraconid. Both upper and lower P4 have large anterior cusps and a posterior basin. P/2 and P/3 are simple. Alveoli suggest the presence of two small incisors and a larger canine. P/2 is double-rooted, which is otherwise unknown in any omomyid. These dental details have often been compared with those of Rooneyia, and Ekgmowechashala was at one point thought to be derivable from a form close to Rooneyia (Rose and Rensberger 1983). However, the two roots of the P/2 are problematic, suggesting primitively little reduced premolars, contrary to what is seen in the uppers of Rooneyia. The two-rooted P/2 is probably among the characters which led to a placement close to Adapis in the phylogenetic analysis of Ni et al. (2010). On the whole, the phylogenetic place of Ekgmowechashala appears unresolved. Is it possible that there was an unusual, atavistic reversal from one to two roots in P/2? A putative rooting close to the highly bunodont Chipetaia might be further explored, as well as possible Asiatic clues.

Fig. 24

Lower dentition of Ekgmowechashala philotau. Supplementary cusps are present on P/4 and molars, and the very low relief can be seen on M/3, which is less worn than the other teeth. Photograph of a white-coated epoxy cast

Amphipithecidae

The Amphipithecidae are a group of Asiatic species found mainly in late Middle Eocene beds of Myanmar; one species comes from the Late Eocene of Thailand, and a small genus possibly survived in the Oligocene of Pakistan. Since the discovery of the fragmentary jaws of Pondaungia cotteri (Pilgrim 1927) and Amphipithecus mogaungensis (Colbert 1937), the overall similarity of this group with the anthropoideans has been recognized. However, they also show peculiar characters which prevent their placement within the African simian radiation, possibly suggesting affinities with a prosimian group (Szalay and Delson 1979). Associated postcranials have increased the controversy. Despite a growing fossil record, no cranial specimen is complete enough to prove the anthropoidean status of amphipithecids.

The Pondaung Formation of central Myanmar yielded a series of specimens of Pondaungia and Amphipithecus, the two best documented genera, as well as rare specimens of two other species, Myanmarpithecus yarshensis (Takai et al. 2001) and Ganlea megacanina (Beard et al. 2009). Species of Pondaungia and Amphipithecus are close in morphology, showing short, high, and heavy jaws with vertical symphyseal region, parabolic tooth rows in occulsal view, and very bunodont teeth. Importantly, their symphyses are not fused. Many specimens show heavily worn teeth. There are two different sizes in specimens of Pondaungia from Pangan and Mogaung, suggesting either a high sexual dimorphism or the presence of two species. Due to the larger-sized specimens being more abundant, Takai and Shigehara (2004) consider the assumption of two species more parsimonious, in which case the large one should be called P. savagei (Gunnell et al. 2002). Details of dental morphology seen on Amphipithecus mogaungensis include short, wide, posterolingually extended P/3 and P/4 (Fig. 25). These teeth have no well-formed metaconid, only irregularities (cuspules?) on a postero-lingual crest descending from the protoconid summit. A small hypoconid is present on the summit of the posterior cingulid. Its lingual crest joins the preceding crest, or the lingual cingulid on P/3, to surround a talonid basin reduced to a narrow and inclined gully. The lower molars have a long, broad, and shallow talonid basin. Their outlines show a median waisting. The cusps of trigonid and talonid have the same height; they are rounded, robust, low, and highly bunodont. The trigonid is antero-posteriorly short, longer on M/1. There is no hypoconulid. The enamel is smooth on most molars of Amphipithecus, which has a small M/3. Pondaungia has crenulated enamel and a large M/3. The jaws of the large Pondaungia are especially high and longitudinally curved (Jeager et al. 1998), whereas jaws of the smaller Amphipithecus are less extreme. Unworn molars of Pondaungia show interesting details, among them M/1 having a paraconid, a continuous anterior paracristid joining the metaconid, a bicuspid metaconid, and a small hypoconulid arising from the postcingulum as in many primitive primate groups.

Fig. 25

Dentition of Amphipithecus and Pondaungia. Left mandibles of A. mogaungensis showing P/3, P/4, and M/1 (a) and M/1 and M/2 (c). Mandible of P. savagei showing unworn M/2 and M/3 (b) and P4/ of Pondaungia, possibly P. cotteri (d). All are in occlusal views (Courtesy of R. L. Ciochon)

The upper teeth are correctly documented since the discovery of a maxilla and associated I1/ and canine of Pondaungia (Shigehara et al. 2002), and other specimens including a complete unworn P4/ (Gunnell et al. 2002). The I1/ is large, robust, and spatulate. The canine is very large and oval in cross-section. It bears a lingual cingulum and a shallow vertical ridge on its lingual side. P3/ is transversely elongated with rounded extremities and has a protocone only slightly narrower than the paracone. A beautiful P4/ is more transversely elongated than P3/, and shows an anteriorly located protocone issuing three crests (pre-, postprotocrista, postprotocingulum) and a large paraconule. Both protocone and paracone are massive and inflated, yet there is no labial or lingual cingulum as in bunodont simians (Fig. 25). The upper molars are rectangular in outline, with some narrowing of the posterior half of the M2/ in both Amphipithecus and Pondaungia, and some narrowing of the lingual part of the M1/ in Amphipithecus only (Takai and Shigehara 2004). The upper molars are better preserved on specimens of Pondaungia, in which they are particularly low-crowned and crenulated. The four major cusps have a rectangular disposition, the hypocone clearly having a strong link to the protocone. Due to this position, isolated from a lingual cingulum on M2/, the hypocone has been interpreted by some authors as a pseudohypocone issued from the posterior flank of the protocone, which would be a rare similarity shared with notharctine adapiforms. However, it is difficult to distinguish those aspects of dental traits which are due to an extreme bunodont adaptation and those which reflect ancestral features of the group. A strong link between hypocone and protocone can develop on a cingular hypocone, e.g., in Necrolemur. A pseudohypocone homologous with notharctines is unlikely (Shigehara et al. 2002), and interpretation of this cusp as a displaced metaconule as in artiodactyls (Coster et al. 2013) is also unlikely. The less derived morphology of Myanmarpithecus and Ganlea leads to the suspicion that the hypocone of amphipithecids is a cingular one, as in most primates.

Maxillae are known from both Amphipithecus and Pondaungia. In dorsal view both show an unusual groove on the lateral side of the orbital floor, just medial to the zygomatic wall (Shigehara and Takai 2004). The orbital floor is very flat in Amphipithecus. Its posterior border does not show any posterior protrusion as seen in taxa with postorbital closure, indicating that this taxon had an incomplete postorbital closure at the most. However, there is no evidence against or in favor of any development of a posterior part of postorbital closure (Shigehara and Takai 2004). In lateral view, the distance between the base of the orbit and the alveolar rim is moderate: higher than in lemuriforms (nocturnal, with large orbits) but lower than in anthropoideans of comparable size (Shigehara et al. 2002). From the relative size of the upper premolars and the shape of the zygomatic root, Shigehara and Takai (2004) deduce that Pondaungia was probably a diurnal primate with a relatively short muzzle. An isolated bone found in close proximity with a maxilla of Amphipithecus was initially interpreted as a frontal of this taxon (Gunnell et al. 2002; Takai and Shigehara 2004). It figured in discussions until it was shown to be misidentified, and certainly not to be a primate frontal (Beard et al. 2005).

Myanmarpithecus yarshensis is a small species that also presents highly bunodont teeth and crenulated enamel (Takai et al. 2001). Its upper teeth have more rounded lingual outlines than in the above-mentioned species (Fig. 26). Despite the fact that the specimen is worn and some details are invisible, the small hypocone on M1/ and M2/ is seen to be peripheral and seems placed on a broad posterolingual cingulum. It would thus seem to be a cingular hypocone. The P4/ is somewhat transversely elongated and has an anteriorly placed protocone without lingual cingulum. A partial left mandible shows C, P/2, and P/3 all with postero-lingual expansion and cingulid, bearing resemblance to platyrrhine teeth. This expansion on P/3 is also reminiscent of Amphipithecus, even if the tooth is more elongated. M/2 and M/3 have anteroposteriorly compressed trigonids, with an apparently complete paralophid joining the summit of the metaconid. Low relief and reduced labial cingulid are as in other amphipithecids; the M/3 third lobe is elongated. The compressed trigonid and extended M/3 lobe are reminiscent of some fossil prosimians, but no close tie to any one of them has been suggested.

Fig. 26

Dentition of Myanmarpithecus yarshensis. Upper teeth P4/ to M3/ (a) and lower molars M/2 and M/3 (c) at the same scale. The fragment of anterior jaw with canine, P/2, and P/3 in lingual view (b) is at a different scale (Courtesy of R. L. Ciochon)

The most recently described amphipithecid Ganlea megacanina (Beard et al. 2009) adds interesting information. In size and several aspects of its morphology, it appears intermediate between Myanmarpithecus and Pondaungia, confirming the place of the former. It also has an enormous canine showing very strong apical wear. P/3 and P/4 are less anteroposteriorly compressed than in Pondaungia. P/4 has a relatively large metaconid, P/3 has none. Both premolars have an anteriorly elongated preprotocristid with a paraconid, especially high on P/3, resulting in a very unusual profile for this tooth. An unworn upper molar, interpreted as M1/ or M2/, has a rounded outline as in Myanmarpithecus, crenulated enamel, no conules, and only a small cuspule on the posterior cingulum that might be an incipient hypocone. Like other amphipithecids, Ganlea was probably a hard-object feeder (Shigehara et al. 2002; Kay et al. 2004a; Beard et al. 2009). The apical wear of its large canine is similar to what is seen in living pitheciines, which are specialized on hard fruits. The same kind of apical wear on canine and premolars is present on Myanmarpihecus, which is smaller than living pitheciines.

The large Siamopithecus from the Krabi mine in Thailand (Late Eocene) is probably the most extreme in its bunodonty (Chaimanee et al. 1997; Ducrocq 1999). Its upper molars are particularly extended transversely. Their hypocone bears unusual links, being anteriorly directly linked to the protocone, posterolingually directly to the posterior cingulum, and, in the case of M2/, also directly linked to the metacone by a lingual crest. The absence of the crista obliqua linking protocone and metacone, replaced on M2/ by a crest linking hypocone and metacone, as in Pondaungia, stands in stark contrast with the morphology of most anthropoideans. The massive, anteroposteriorly long and transversely short P3/ and P4/ are also very distinctive – much more reminiscent of a giant omomyiform than of an anthropoid. To what degree are these premolars autapomorphic, obscuring phylogenetic signals? Maxillary and facial fragments associated with a lower jaw of Krabi have been used to produce a reconstruction of facial elements through synchrotron microtomography, followed by adjustment of digital images (Zollikofer et al. 2009). Despite the fragmentary nature of the remains, the positioning of two parts of the orbital rim suggests that Siamopithecus had highly convergent and frontated orbits, as is typical of living anthropoideans. Convergence does not seem so high from the figures, whereas frontation does; it looks higher than in the African simian Parapithecus. Estimations of these parameters as well as a morphometric study based on landmarks place Siamopithecus phenetically within anthropoideans. However, this fossil is still fragmentary and does not show the key characters of postorbital septum or middle ear morphology that would definitively prove an anthropoidean status.

A small species from the Oligocene of Pakistan, Bugtipithecus inexpectans, is represented by six isolated teeth having unusual characters (Marivaux et al. 2005). It is less bunodont than the genera discussed above; it shows a well-formed small cingular hypocone with a prehypocrista linking it to the postprotocrista, a cuspidate metaconule, and labial and lingual cingula on the upper molars. This small survivor may well illustrate primitive character states for amphipithecids. These would point toward a possible connection with eosimiids (posterior waisting, parastyle, salient hypoparacrista on M1/, morphology of P/4), but not with primitive African simians.

The controversies about amphipithecid affinities have been fueled by the discovery of a small set of postcranial specimens, probably of the same individual, in the Pondaung Formation. These bones (“NMMP 20”) have the right size to be attributed to species of Pondaungia, but have a series of characters that are difficult to reconcile with anthropoidean affinities (Ciochon et al. 2001). An almost complete humerus shows a reduced deltopectoral crest, a shallow bicipital groove, a rounded capitulum with an expanded capitular tail, a groove separating the capitulum and the trochlea and a lateral trochlear rim, an expanded lateral epicondyle, and the absence of a dorsoepitrochlear pit (Fig. 27). On the one hand, these characters and others constitute important differences from anthropoidean humeri, while on the other they present similarities with notharctine adapiforms. However, there are also differences from known notharctines, e.g., a distomedially sloping trochlea, a proximally salient humeral head, and a few other traits. There are also a proximal ulna, a calcaneum lacking its proximal tuber, and other fragments. The calcaneum is quite different from that of simians, showing similarities with notharctines or adapines, as well as a few differences (Fig. 27). The calcaneocuboid facet has similarities with that of Adapis in its marked dorsoventral orientation (Ciochon and Gunnell 2004). Functionally, many characters reflect mobile joints and stability, as in slow-mowing arboreal quadrupeds (Kay et al. 2004a). In terms of affinities, most authors agree that these characters are at odds with anthropoidean affinities, leading to two different propositions: (i) they confirm a notharctid origin for amphipithecines (Ciochon et al. 2001; Ciochon and Gunnell 2004); (ii) they must not pertain to Pondaungia, considered an anthropoidean, but to a yet unsampled large sivaladapid adapiform, because small adapiforms were found in the Pondaung Formation and large sivaladapids did exist in Asia (Beard et al. 2007). A partial pelvis, found at the same locus and probably pertaining to the same individual, shows strepsirhine rather than anthropoidean affinities, according to Marivaux et al. (2008).

Fig. 27

Some elements pertaining to the controversial NMMP 20 set of postcranials possibly pertaining to Pondaungia savagei. Proximal left humerus in posterior (a) and proximal (c) views; distal left humerus in anterior view (b); and distal calcaneum in dorsal view (d). All at the same scale (Courtesy of R. L. Ciochon)

Two isolated astragali from the Pondaung Formation have been successively described, both of which have the right size to be attributed to a large-bodied amphipithecid (Marivaux et al. 2003, 2010). They show no adapiform/strepsirhine character but similarities with omomyids and simians, and have been interpreted in favor of simian affinities for amphipithecids (Marivaux et al. 2003, 2010; Dagosto et al. 2010). These bones are clearly lacking the derived strepsirhine/adapiform astragalar characters. However, their similarities with anthropoideans are not definitive proof of simian affinity because, as will be explained below, many of these characters are essentially primitive in primates, allowing only a “haplorhine” (sensu non-strepsirhine) identification (Dagosto et al. 2010).

The postcranial evidence for large-bodied primates from the Pondaung Formation is embarrassing. When the bones are possibly anthropoidean-like, they are attributed to amphipithecids. If they are clearly non-anthropoidean in their characters, proponents of the anthropoidean status of amphipithecids ascribe them to a yet unsampled large sivaladapid. However, it is strange that the growing number of fossil finds in the Pondaung Formation has not to date included discovery of such a large sivaladapid, whereas the number of large amphipithecid specimens has increased considerably. The alternative possibility is that the NMMP 20 bones and the astragali pertain to amphipithecids, and that this group is neither adapiform nor anthropoidean, but instead constitutes some kind of large “haplorhine” primate.

On the whole, craniodental remains of amphipithecids show tantalizing similarities with anthropoideans in the height and robustness of the mandibles, vertical symphysis, and many aspects of the bunodont dentition. But at the same time there are many differences with African simians which prevent researchers from positing a close affinity with propliopithecids or parapithecids (e.g., the latter have big conules and a large hypocone that is never linked to the metacone on the upper molars, short upper premolars with continuous lingual cingulum, lower molars with big hypoconulids, short M/3 lobe, etc.). Some characters of amphipithecids, such as the unfused symphysis in large bunodont species and the waisted lower molars, even more strongly suggest convergent evolution with African simians in other bunodont features. If they were anthropoideans, they would be very distantly related to African families, contrary to the results of numerous recent cladistic analyses. The weakness of parsimony analyses including large dental data sets, applied to very distantly related groups and a fossil record full of gaps, is stunningly apparent in the analysis of Kay et al. (2004b), in which Amphipithecidae are jumping from a place within crown anthropoideans to a place within fossil adapiforms, as a result of a slight change in analytic procedure. The program is evidently trying to ally them with the group presenting the highest number of shared derived characters. However, in an analysis including such primitive taxa as Teilhardina and Donrussellia, almost all characters of amphipithecids are derived relative to these primitive genera. In the absence of truly diagnostic characters, a high number of similar derived characters of the dentition and the postcranials can show up in two very different groups. What such analyses produce is more a signal of global phenetic similarity than a signal of phylogenetic affinity. The two extremely different results reveal that there is no clear phylogenetic signal in this analysis. Clearly, the data at hand do not allow elucidation of the problem. A similar analysis performed by Beard et al. (2009) nested amphipithecids within African simians; however, we know that this phylogeny is incorrect because Amphipithecus and Pondaungia lack complete postorbital closure. This demonstrates that a considerable amount of convergence in dental characters occurred, as indicated by the dental characters themselves. In spite of the appealing anthropoidean-like reconstruction of Siamopithecus facial parts, Amphipithecus and Pondaungia lack one of the diagnostic anthropoidean characters; furthermore, their suborbital region is lower than in anthropoideans of similar size (Shigehara et al. 2002), confirming convergence in cranial characters among haplorhine groups. Amphipithecids appear as highly bunodont diurnal primates, convergent with some simians in their craniodental specialization for hard-object feeding.

A sister group relationship between amphipithecids and eosimiids has been suggested by Jaeger et al. (1999). The probability of such a connection has since been increased by the discovery of Myanmarpithecus, which partly bridges the gap. It has been strengthened further by the discovery of Bugtipithecus and Phileosimias. The phylogenetic analysis of Marivaux et al. (2005) showed the two groups to be close, although their analysis continued to ally amphipithecids with African anthropoideans. Such a connection suggests the possibility that amphipithecids are derived from eosimiids and should be placed along with them in Tarsiiformes. As seen above, such a placement fits with the facial and postcranial characters. The two families would considerably broaden the adaptive spectrum of tarsiiforms, and in doing so reset the balance between the two major clades of haplorhines, simiiforms and tarsiiforms.

African Paleogene Anthropoidea or Simiiformes

A whole radiation of anthropoideans, or simians, is documented from the Late Eocene and Oligocene of Egypt. Extensive exposures in the Fayum desert have been yielding fossil primates since the beginning of the twentieth century. Renewed efforts of field crews led by Elwyn Simons and the Egyptian Geological Survey since the 1960s have resulted in the establishment of a succession of faunas, situated in a precise stratigraphic sequence (e.g., Simons 1995). Comparisons with similar faunas from Oman, which are linked with marine deposits, and magnetostratigraphy have led to a new dating of the section (Seiffert 2006). The most productive beds are those of the Jebel Qatrani Formation, which is mostly Oligocene : only one third of the “lower sequence” is Late Eocene in the new temporal framework. The older fauna called BQ-2, found in the Birket Qarun Formation and dated around 37 Ma, has yielded the oldest known undoubted anthropoids (Seiffert et al. 2005b). All families have at least one genus known by its skull. These fossils can thus be included in character-based anthropoideans; they possess their hallmarks, that is, a simian type of postorbital closure and the typical ear region (detailed below for Parapithecus).

Parapithecidae

The family Parapithecidae contains a group of four tightly united genera, Apidium, Parapithecus, Qatrania, and Abuqatrania, which can be considered as Parapithecinae, and the more primitive genus Biretia. The oldest and smallest species, Biretia fayumensis from BQ-2, shows relatively bunodont upper molars with a large cingular hypocone, a distinct paraconule on M1/, and well developed metaconules on M1–2/, P3/ and P4/ which have small hypocones and a P2/ transversely elongated, with a protocone and three roots (Seiffert et al. 2005b). These characters are shared with some later parapithecines. The lower molars show a moderate bunodonty and a centrally placed hypoconulid. They also retain a complete protocristid and a premetacristid, which will be lost in the later, more extreme bunodont species. P/4 has a cuspidate metaconid linked to the anterolingual cingulid, a hypoconid, and lingual talonid basin; its outline is somewhat anteroposteriorly elongated, whereas it becomes shorter and broader in several later species. It is a small species, with molars between 2 and 3 mm in length and a weight estimate of roughly 200 g. The slightly larger B. megalopsis, from the same locality, differs in a few dental details. Its maxilla reveals surprising features: the suborbital region is very shallow, resulting in the exposure of the roots of M1/ and M2/ in the orbital floor, which is formed at their level by the same bony lamina as the palate. The authors infer large orbits and a nocturnal adaptation for Biretia, whereas all later African anthropoideans are diurnal (Seiffert et al. 2005b).

Species of Qatrania and Abuqatrania are also small (length of P/4–M/3 less than 1 cm). Their lower molars are extremely bunodont and somewhat high-crowned (Fig. 28). Their crests are reduced, and their cusps are low and rounded, expanded into the talonid basins. There is a large centrally placed hypoconulid. M/1 bears a small paraconid. On M/2 and M/3, protoconid and metaconid are separated by an anteroposterior groove; the protocristid is lost. The P/4 is short and simple, with a small metaconid and hypoconid. The mandibular rami of these species have been reported as very shallow (Simons and Kay 1988), and this is beautifully illustrated by the mandible of Abuqatrania basiodontos, which has both a very shallow ramus and a high and anteroposteriorly short coronoid process (Simons et al. 2001). One upper molar of Qatrania has a very large hypocone. Species of these genera differ by small dental details. The absence of a small posterolingual shelf (fovea) on the M/2 of A. basiodontos is emphasized as different from species of Qatrania. However, this is a primitive character state, found in the oldest species (from L-41). Placing that species in a different genus may obscure its likely ancestral status relative to the later species, Q. wingi and Q. fleaglei, from Oligocene levels.

Fig. 28

Dentition of parapithecids. Right lower mandible of Qatrania wingi (a); maxilla (b) and mandibles (c and e) of Parapithecus, probably P. grangeri; and left mandible of a juvenile Apidium phiomense with worn milk teeth DP/3 and DP/4, and definitive M/1 and M/2 (d). The high degree of bunodonty is striking, and the maxilla also shows the supplementary cusp present on the three premolars and the great size of the hypocone, larger than the protocone on M1/ (b). All in occlusal views except (e), which is a labial view with M/1-M/3. Photographs of white-coated epoxy casts, all at the same scale

Apidium and Parapithecus are clearly larger species found in the Oligocene beds. They are much better known and allow the description of skull and postcranials of parapithecids. There are two species of Parapithecus: the type species P. fraasi, known through its type mandible described by Schlosser (1911), from an unknown level, and P. grangeri (sometimes “Simonsius” grangeri), known by more material from Oligocene quarries and now including a cranium (Simons 2001). The lower molars and premolars are larger and accentuate the bunodont characters of Qatrania, showing, e.g., a longer and deeper groove between protoconid and metaconid on the lower molars, a larger metaconid isolated from the protoconid by a groove, an anterolingual cingulid and an abbreviated talonid on P/4. P/3 is similar to P/4 on P. fraasi, and P/2 is simpler, unicuspid, and bordered by a continuous lingual cingulid. The canine is moderate in size and shows an anterolingually ascending cingulid. The two species differ in size, in some dental details, and in the retention in the type mandible of P. fraasi of one incisor on each side, which is of light color and suspected to be a retained deciduous incisor, whereas P. grangeri has no more lower incisors at all – a morphology exceptional among simiiforms. The upper teeth are also extremely bunodont. On M1/ and M2/, the rounded cusps are inflated and the conules large, reducing the trigon basin (Fig. 28). The hypocone is almost as large as the protocone on M1/, smaller on M2/ (Kay and Williams 1994). P3/ and P4/ have a paraconule and P2/ appears transversely shorter on the cranium (Simons 2001). The alveoli reveal that there was a moderately large upper canine and two small incisors. Two anterior lower canines appressed against each other with intersticial wear, and wearing flat apically against the upper incisors, constitute a unique dental device – one that is strange in a species reported to be dentally dimorphic. The highly bunodont molars of parapithecids are interpreted as broadly adapted to frugivory, with nuances; e.g., a higher shearing capacity in Parapithecus may reflect a slightly more folivorous diet (Kay and Simons 1980).

The cranium of P. grangeri is fascinating. It has the complete postorbital closure typical of simians and at the same time an elongated profile recalling more primitive primates (Simons 2001, 2004). Its orbits are posterolaterally inclined in dorsal view, giving a convergence angle of about 105°, and posterodorsally inclined in lateral view, showing a frontation lower than in other simians, similar to some prosimians (Fig. 29). Among its noticeable characters are a large zygomatico-facial (or malar) foramen, as found in some platyrrhines; a glenoid fossa shaped as a transverse trough; a transversely extended postglenoid process; a lateral pterygoid wing extending posteriorly to overlap the lateral bullar wall; a relatively anteriorly located foramen magnum; and a relatively small braincase revealing an encephalization much lower than in living anthropoideans, comparable to that of Eocene prosimians (Simons 2001). This latter assessment depends on body size estimates; those based on dental regressions had initially given body weights ranging from 1.6 to 3 kg. More reliable estimates based on postcranials showed that 1.6 kg is the more likely value (Simons 2004). Even with this low estimate, a better calculation using a CT-scan reconstruction of the brain cavity confirms an encephalization comparable to that of living strepsirhines (Bush et al. 2004). The complete bullae preserved on the cranium appear relatively distant from each other. They confirm two characters which had been established earlier through the study of isolated parapithecid petrosals: the anular ectotympanic co-ossified with the petrosal around the auditory meatus, and the posteromedial location of the large carotid foramen (Cartmill et al. 1981). The latter study had revealed that there are septa between the promontorium and the ectotympanic. One such septum separates the tympanic cavity from an anterior accessory cavity; the carotid canal travels through this septum, moving for a short distance along the anterior part of the promontorium as a large promontory canal, without a stapedial branch (the perbullar pathway of Cartmill et al. 1981). There is a relatively high degree of petrosal pneumatization, especially of the anterior accessory cavity; a large postglenoid foramen indicates a primitive petrosquamous venous drainage. On the whole, the ectotympanic, pneumatization, and characters of the carotid canal show that parapithecids possessed the distinctive and highly specialized otic complex of living anthropoideans. This morphology bears no resemblance to that of adapiforms. It is in large part shared with Tarsius but not with known omomyiforms (Cartmill et al. 1981), and it is a key element at the base of the restricted character-based notion of haplorhines, as endorsed above.

Fig. 29

Schematic drawings of the cranium of Parapithecus grangeri in lateral (a) and dorsal (b) views. It shows the overall elongation of the cranium in lateral view. Straight lines indicate the approximate value of the convergence angle (b) and the frontation angle (a) of the orbits

Apidium is a closely related genus, with three species (A. moustafai, A. phiomense, A. bowni). It differs from Parapithecus by some characters of the dentition. For example, in A. phiomense, lower molars increase in size from M/1 to M/3, which is uncommon in anthropoideans; the extremity of their cristid obliqua may be inflated, isolated from the hypoconid by a groove, and forming a centroconid. On the upper molars of A. moustafai, the large hypocone is lingually shifted and linked to a thick anterolingual cingulum, which gives the impression that it is larger than the protocone on M1/ and M2/. The anterolingual cingulum can even be cuspidated, resulting in a thick pericone (in A. phiomense; Kay and Williams 1994). P2/ is transversely extensive and bears a well-formed paraconule. The upper canines of Apidium present an anterior groove which extends into the root (a convergence with cercopithecoids). The highly bunodont teeth of A. phiomense would suggest frugivory; however, the species possesses a relatively much thicker enamel than other Fayum primates, and its enamel is of the primitive radial type (not decussating). A microwear study did not find evidence of hard-object processing (Teaford et al. 1996). Was the diet of this species highly abrasive, and in that case, what was it?

A large sample of postcranials attributed to A. phiomense has been studied in detail, which makes it the postcranially best-known Paleogene simian (Fleagle and Simons 1995). The scapula of Apidium has a glenoid fossa similar in shape to that of Saimiri. It was described by Anapol (1983), who conducted an analysis of scapular angles and the surface of the glenoid fossa, and found the bone to be most similar to those of quadrupedal primates. No humerus shows a complete proximal head. There is a prominent deltopectoral crest, which extends roughly one third of its length. The proximal part is deep anteroposteriorly and narrow mediolaterally. On the distal shaft, the brachialis flange is relatively narrow. There is a large entepicondylar foramen. The articular part shows a relatively small, mediolaterally elongated capitulum, the medial part of which is continuous with the broad trochlea (there is no pronounced groove or lip between them). The trochlea is conical and its medial lip extends further distally than the capitulum. The deep coronoid fossa, the proximal extent of the capitular surface, and the proximal extent of the medial part of the trochlea all suggest that extreme flexion was common in this species. The medial epicondyle is prominent, directed posteriorly at an angle of about 20°, as in arboreal quadrupeds. On the posterior side the trochlear surface is limited by prominent lips, and there is a dorsoepitrochlear pit, a common feature in omomyids and platyrrhines. The ulna has a relatively long olecranon process. The sigmoid cavity, described in detail by Conroy (1976), shows greatest similarity with the platyrrhines Cebus and Saimiri. The radius presents a head that is oval in shape. Its shaft is very broad and robust in comparison with living primates. Its distal part becomes relatively broad and flattened for a large pronator quadratus muscle, as in arboreal quadrupeds. Its distal articulation shows a sharp styloid process.

Several partial hip bones were recovered (Fleagle and Simons 1979). The ilium shows a broad gluteal plane, rectangular in its preserved part, which is characteristic of anthropoideans, and an expanded iliac plane that narrows proximally, as in Saguinus. The ischium is long and proximally broad, and the ischial tuberosity extends slightly above the level of the acetabulum. The dorsal rim of the acetabulum is thicker than the ventral rim. Most characters indicate a quadrupedal primate, although it seems that relative iliac length was lower than in other primates. The femoral head has a rounded articular surface which is restricted anteromedially and expanded posteromedially, as in Cantius. Several characters of the femur are typical of leaping primates: a thick and short neck, perpendicular to the shaft; a relatively distal fovea capitis; a prominent intertrochanteric line; a proximal surface of the greater trochanter that is relatively flat and broad in anterior view, anteriorly overhanging the shaft. There is no third trochanter. The lesser trochanter is very large, more rectangular than in living primates, and it joins the greater trochanter to wall off the posterior femoral fossa. The femoral shaft is relatively robust. The distal articulation is higher anteroposteriorly than in all other anthropoideans, and the patellar groove is deep and narrow, bordered by a prominent lateral lip – all characters reflecting a leaping adaptation. The tibia also presents a series of characters associated with leaping: an extensive, convex lateral condyle and a smaller, concave medial condyle; a proximal shaft that is very narrow mediolaterally; a cnemial crest extending 20–22 % along the shaft, indicating the proximal insertion of the hamstring muscles; tibia and fibula distally appressed for roughly 40 % of their length; a deep medial malleolus and prominent beak on the anterior side, which match the midline groove of the astragalar trochlea, indicating that movements at the ankle joint were restricted to flexion-extension in the parasagittal plane (Fleagle and Simons 1995). The astragalus has a broad dorsal articular surface with pronounced lips and steep sides. The medial trochlear articulation is very shallow, and the fibular facet is steep and relatively shallow. The articular head is ovoid in shape, and its main axis is rotated dorsolaterally. The calcaneum is relatively robust, with a thick tuber and no plantar process. The posterior calcaneal facet is relatively small and very convex, with a steep distal part; the similar size and shape of the matching facet on the astragalus show that there was very little rotation between them at this level. The sustentacular facet is small and circular. There is a prominent peroneal tubercle opposite the sustentaculum. The anterior articular facet faces medially and has little connection with the sustentacular facet, which is unusual and resembles many cercopithecids. The cuboid facet, slightly concave and lunate-shaped, surrounds a deep pit for the calcaneo-cuboid ligament. Below this pit, the tuberosity for the short plantar ligament is more strongly developed than in most platyrrhines. The cuboid is wedge-shaped, broader proximally than distally, and relatively narrow and deep. Its distal face has a single T-shaped facet for one metatarsal, the fourth (MT IV), and the lateral face shows an anterior broad facet for the fifth (MT V). Such a lateral articulation of the fifth metatarsal is unusual among primates; today it is found only in Tarsius (Fleagle and Simons 1995). On the navicular, the distal face has three distinct facets for the three cuneiforms, aligned in an L-shaped arrangement as in omomyids, Tarsius, and anthropoids (Dagosto 1988).

The bones ascribed to A. phiomense show an unusually high variability in size. Few long bones are intact, but the best preserved, completed by others, allow estimations of their median length (e.g., humerus 5.5 cm, femur 8.9 cm). An estimation of the intermembral index gives a score of 62, less than in any other simian and similar to leaping prosimians (Fleagle and Simons 1995). On the whole, Apidium shares most characters with small leaping platyrrhines like Saimiri or Saguinus; however, it was more specialized for pronograde leaping (not VCL), and it also shows a series of other traits – some shared with cercopithecids and others unique among living primates.

The systematic status of Parapithecidae has changed. They have been considered sometimes primitive catarrhines, sometimes platyrrhines, sometimes related to cercopithecoids. Following the detailed analysis of Fleagle and Kay (1987) they have often been considered a primitive sister of all other anthropoideans. However, more recent analyses including Proteopithecidae conclude to an unresolved polytomy between catarrhines, platyrrhines, proteopithecids, and parapithecids (Seiffert et al. 2005b, 2010a). The phylogenetic relationships between these early families and platyrrhines are not yet resolved.

Proteopithecidae and Arsinoea

The species Proteopithecus sylviae has been found in the locality L-41 of the Fayum. Incomplete dental remains first led researchers to rank it among the oligopithecines; however, the discovery of more complete dental specimens, crania, and some postcranials resulted in revised placement, together with Serapia eocaena, in a family Proteopithecidae (Simons 1997a, b; Simons and Seiffert 1999). Members of this family have three premolars above and below. One cranium shows that Proteopithecus has spatulate central upper incisors, a relatively large upper canine somewhat transversely compressed, bearing an anterior vertical groove and a lingual cingulum, and upper cheek teeth of large size in relation to palatal breadth. A P2/ smaller than P3/ has only a small lingual lobe with a small protocone and a continuous cingulum. P3/ and P4/ are transversely extended. They both have a strong posterior cingulum which broadens lingually in a shelf reminiscent of the hypocone of the molars. M1/ and M2/ are transversely elongated, have no conule (or only a small paraconule on M1/), a protocone reported to be smaller than the labial cusps, and a relatively large crestiform hypocone. M3/ is reduced. In the lower dentition, P/2 is larger than P/3; both are conical. P/4 is much larger than P/3 and has a large metaconid and a strong lingually recurved preprotocristid, which in occlusal view gives the impression that P/4 has a trigonid similar to that of M/1. It has a short talonid (Fig. 30). The lower molars are generalized, with a trigonid higher than the talonid, which is broad. M/1 has a small paraconid; M/2 and M/3 have a paralophid without cusp. A hypoconulid is frequent on M/1–2, twinned with the entoconid. M/3 is small and shows high variability in its talonid morphology.

Fig. 30

Dentitions of proteopithecids and oligopithecids. Right mandible of Proteopithecus sylviae showing P/3, P/4, and the lower molars (a). Partial palate with upper teeth (b) and mandible with lower teeth (c) of Catopithecus browni. The maxilla in (b) shows from left to right the elongated I1/, the shorter I2/ displaced from its broken root, the large and high canine, subtriangular in occlusal view, and the two upper premolars and three molars. On the lower jaw are also preserved the large canine, two premolars, and three molars. Photographs of white-coated epoxy casts, all at the same scale

The species P. sylviae is relatively small; one cranium is 4.4 cm in length. It presents all the simian characters of orbits and auditory region which have been given above for Parapithecus, plus a fused metopic suture and lacrimal bone within the orbit (also simian features). Among the distinctive characters of the genus are small premaxillae, large posterior palatine foramina, lack of a posterior palatine torus, and a jugal foramen smaller than in Parapithecus. The rostrum is proportionately shorter than in Catopithecus, and the anterior margin of the orbit is more forward located, being above a line between P2/ and P3/. The temporal lines converge far posteriorly and join to form a slightly elevated sagittal crest. The latter joins two less distinct nuchal crests and a salient vertical occipital crest. The posterior cresting, more developed than in similar-sized platyrrhines, is probably correlated with the relatively smaller brain size. The glenoid fossa is anteroposteriorly and mediolaterally broad and flat; there is a well-developed postglenoid process and just posteromedially a distinct postglenoid foramen. Both this region and the facing articular process of the mandible are reported as similar to those in Catopithecus (Simons 1997a).

Postcranials of Proteopithecus include two humeri, a partial femur and hip bone, one femur and two tibiae found in association with a mandible, and an astragalus and a calcaneum (Gebo et al. 1994; Simons and Seiffert 1999; Seiffert et al. 2000; Seiffert and Simons 2001; Gladman et al. 2013). The humerus is very similar to those of parapithecids. In profile view, its shaft has a distinct sigmoidal curvature. Differences with parapithecids include a less extended supinator crest and less developed brachial flange, narrow and deep bicipital groove (as in Qatrania fleaglei), relatively narrow distal articular surface, and broad capitulum confluent with a narrow trochlear surface; on its posterior side the latter shows a well-developed lateral lip within the olecranon fossa. The partial hip bone is similar in many respects to the hip bone of Apidium, but has an even larger crest separating the gluteal and iliac planes, and both gluteal and iliac planes are broader in their caudal parts than in Apidium. Quantitative ratios used to describe the acetabulum match best with those of extant catarrhines, rather than with platyrrhines or prosimians (Gebo et al. 1994). The femoral neck angle is 108°, and head and neck length are less than in Apidium. There is a crista paratrochanterica on the posterior side, between head and greater trochanter. There is a third trochanter at the distal end of the latter, which is more prominent than in any living anthropoidean. The lesser trochanter is large and flange-like. Its lateral crest meets the posterior crest of the greater trochanter, closing the trochanteric fossa as in Apidium. However, these crests are less prominent, and the lesser trochanter is less expanded and rectangular, resulting in a trochanteric fossa less extended than in Apidium. The trochanteric fossa is interpreted as a more primitive version of that of Apidium (Gebo et al. 1994). Distally, the patellar groove is broad and has a prominent lateral lip. The ratio of anteroposterior depth to mediolateral width is 78, lower than in Apidium (103). The tibiae are longer and more gracile than those of Apidium. Other characters of the proximal part are as in Apidium and reflect a leaping adaptation. On the distal part, the fibular articular facet is short (10 % of total length), as in some platyrrhines, contrasting with the long syndesmosis of Apidium. The humero-femoral index is 73, higher than in Apidium (62). The crural index is 106 (vs. 106–111 in Apidium), most similar to those of callitrichines (100–101). The astragalus is very similar to those of parapithecids and small-bodied platyrrhines, but with a lower body and a long axis of the head mediolaterally oriented in distal view (more dorsoplantar in Apidium). The calcaneum shows a distal placement of the peroneal tubercle, a distal plantar tubercle, and separated sustentacular and distal astragalar facets – characters which may be shared-derived with parapithecids (Gladman et al. 2013). It also differs from parapithecids in being more slender and in having a greater distal elongation (43 %). Characters of the knee, tibia, and ankle clearly indicate a quadrupedal arboreal locomotion, including frequent rapid running and pronograde leaping.

Proteopithecidae appear to be the most generalized of the African Eocene-Oligocene anthropoideans. They have a large number of similarities with small living platyrrhines, though many of these are considered primitive for simians. It has been suggested repeatedly that proteopithecids are the only serious candidates for a stock from which to derive platyrrhines (Simons 1997a; Simons and Seiffert 1999), because parapithecids and propliopithecids have derived specializations which preclude their playing such a role. However, living platyrrhines also have their own autapomorphies, and until now it has not been possible to identify clear shared derived characters that would ally platyrrhines with proteopithecids. On the contrary, recently found postcranials suggest that proteopithecids may well be more closely related to parapithecids.

The systematic placement of Arsinoea kallimos, known only by its lower dentition, has fluctuated. It is a small primate: the complete tooth row of the type mandible, from I/1 to M/3, is less than 2 cm long. Among its distinctive characters are a low-crowned canine, a relatively large, anteroposteriorly elongated I/2, and three simple premolars slightly increasing in size from P/2 to P/4 (Simons 1992). M/2 is reported as low-crowned. There is a large paraconid well separated from the metaconid on M/1, a smaller paraconid close to the metaconid summit on M/2, and no paraconid on M/3. This type of posterior paraconid reduction is unlike what is found in other simians, recalling very different primate groups. Arsinoea has been described as “family uncertain,” or found to form a trichotomy with other higher taxa and proposed as member of a new family Arsinoeidae (Simons et al. 2001). More recently, phylogenetic analyses have proposed a place of primitive sister group to other parapithecids (Seiffert et al. 2005a). A closer comparison with Biretia would be interesting, and upper teeth would help specify its place.

Catarrhini : Oligopithecidae and Propliopithecidae

For many years, Oligopithecus savagei was known only by its type mandible, found in Quarry E of the Fayum. The anterior part of this mandible slightly increases in height anteriorly, and bears a moderate-sized canine with an anteriorly ascending lingual cingulid. There are only two premolars and a P/4 that is relatively short and molarized, with a large metaconid as anterior as the protoconid, and an anteriorly elongated preprotocristid turning lingually into a brief and thick paralophid (almost a paraconid). The P/3 is larger, simpler, and much higher. Its main cusp has a long anterior preprotocristid with wear indicating its honing function for the upper canine. The lower molars are transversely broad, crested with a low relief, with a cristid obliqua reaching the posterior base of the protoconid. M/1 has a small paraconid, and M/2 clearly shows a lingually placed hypoconulid, twinned with the entoconid. The presence of only two premolars, morphologically derived and close to those of Aegyptopithecus, makes it a tantalizing catarrhine. However, its molars are quite different from those of the latter, being crested with twinned entoconid and hypoconulid, instead of bunodont with a centrally placed hypoconulid. On the other hand, these molars resemble those of the adapiform Hoanghonius, leading to the idea of a status possibly transitional between adapiforms and simians (Gingerich 1977b). This notion was further strengthened by the discovery of an upper molar of O. savagei with high crests recalling cercamoniines, interpreted in terms of phylogenetic affinity (Rasmussen and Simons 1988). However, subsequent discoveries, and especially analysis of more complete material of the closely related Catopithecus, definitively proved the anthropoidean status of O. savagei. A second species of Oligopithecus, O. rogeri, was found in the Oligocene locality of Taqah, Oman (Gheerbrant et al. 1995). It completes the knowledge of the genus and shows specific differences. Interesting is the P4/ with a large protocone and well-developed crests forming the lingual part of a trigon (with one labial cusp), and a complete lingual cingulum. The upper molars have an anteroposteriorly more elongated labial part, marked medial waisting, and a thicker and more continuous lingual cingulum, with posterolingual extension on M2/. The trigonid of the lower molars seems less anteroposteriorly compressed.

Catopithecus browni is found in the Latest Eocene locality L-41. Many specimens are known, allowing the description of the dentition, cranium, and parts of the limb skeleton (Fig. 30). The mandibles are anteriorly unfused, a unique case in anthropoideans which suppresses one of their synapomorphies, and also a unique case for catarrhines (Simons and Rasmussen 1996). The lateral incisor is spatulate. The canine is large, projecting, and sexually dimorphic in size. It has a marked cingulid descending anteriorly from the tip and curving posteroventrally on the lingual face. P/3 is also dimorphic. It has a high main cusp and a small talonid. Its preprotocristid serves as honing blade for the upper canine; it is longer in males, which have a broader P/3. P/4 is more complex, with small paraconid and metaconid. The molars are transversely broad. Their crests are not high and their cusps are moderately bunodont. M/1 has a small median paraconid, whereas M/2 and M/3 have a paralophid which seems to enclose the narrow trigonid basin. The talonid basin is relatively wide, and entoconid and hypoconulid are tightly twinned on all molars. M/3 has roughly the same length but is much narrower than M/1 and M/2; its third lobe is only slightly salient posteriorly. The upper incisors are described and figured by Simons and Rasmussen (1996). They are spatulate and resemble those of small patyrrhines. I1/ is larger and elongated in the mesiodistal plane, and relatively flat on its lingual face, which has a straight margin (Fig. 30). I2/ is smaller and less elongated, with a rounded lingual margin. The upper canine is a simple tooth, transversely compressed, bearing an anterolingual cingulum and an anterior vertical groove, and is dimorphic. P3/ and P4/ are relatively simple, with a large paracone, small styles, the lingual lobe narrower than the labial part, and a continuous thin lingual cingulum at least on P4/. The protocone is smaller on P3/ than on P4/. M1/ and M2/ are relatively simple. Their broad trigon basin is limited lingually by a raised and curved crest, which is formed by a preprotocrista running to the parastyle and a postprotocrista running toward the metacone. A continuous lingual cingulum bears a crestiform hypocone, larger on M2/ than on M1/, and a small pericone on M2/ alone. The labial cingulum is faint and shows a slight mesostyle thickening. M3/ is small, and is reduced in its posterior part.

Six crania of Catopithecus are known that are crushed in different ways, allowing many anatomical traits to be recognized (Simons and Rasmussen 1996). In many respects they resemble Parapithecus and small extant platyrrhines like Callithrix. Thus only some peculiarities are described below. In the facial region, the nasals are broad and long, and the premaxillae have a very broad ascending process. There is a broad interorbital distance, and a broad intercanine distance on the palate. The orbital fissure is made of two parts, an anterior one, which is lenticular, and an inferior one of moderate size. The zygomaticofacial foramen is smaller than in parapithecids and platyrrhines, but larger than in most catarrhines. The relative orbit size indicates diurnality, and convergence is estimated at 120–130°. The anterior margin of the orbit is above the contact between P3/ and P4/, more posterior than in most small platyrrhines. On the ventral side, the palatine bone shows robust pyramidal processes running posterolaterally from the palate, as in Aegyptopithecus and some catarrhines; these processes are commonly more gracile in other anthropoideans. On the dorsal side of the cranium, there is no metopic (interfrontal) suture. Strongly developed orbital crests link the posterolateral part of the supraorbital ridges to the temporal lines, which are arcuate and join far posteriorly, bordering a long frontal trigon. The sagittal crest and prominent occipital protuberance suggest a relatively flat and vertical nuchal plane, contrasting with the ballooned shape of small extant platyrrhines. Rough estimates of brain size indicate a low encephalization, even lower than in living prosimians. The relatively large olfactory bulbs project in front of the frontal lobe, as they do in prosimians, whereas they project more downward in extant anthropoideans and Aegyptopithecus (Simons and Rasmussen 1996).

Several limb bones have been ascribed to Catopithecus. A complete but proximally crushed humerus and two partial distal humeri show that the humerus must have been relatively long; the deltopectoral crest is reduced in comparison with Proteopithecus, and the brachial flange is moderately developed. The entepicondylar foramen is smaller and more medially placed than in Apidium; it is lost on one of the three specimens; its lateral wall is confluent anteriorly with the medial edge of the trochlea. A large medial epicondyle projects more medially than in other Fayum taxa. On the articular part, the capitulum is very round and sharply demarcated from a long and prominent capitular tail. Compared with Apidium, the capitulum and zona conoidea are relatively shorter, whereas the trochlea is broader. The angle of translation of the ulna is small, as in other Fayum primates, whereas it is larger in platyrrhines (Gebo et al. 1994; Rose 1988). A morphometric study of the distal articular surface shows that this surface is phenetically closer to those of propliopithecines than to those of other Fayum anthropoideans (Seiffert et al. 2000). The femur has a relatively short head and neck length, a head more rounded than in Proteopithecus, and a notch between the greater trochanter and the head that is shorter and deeper than in Apidium. A third trochanter extends farther distally than in Proteopithecus and appears variable, ranging from prominent to poorly developed (Seiffert and Simons 2001). The large lesser trochanter does not contact the long trochanteric crest, which runs distally. The astragalus of Catopithecus is very different from those of parapithecids and proteopithecids. It has an increased curvature of the distomedial margin of the trochlea, a relatively deep and medially projecting cotylar fossa, and an elevated lateral trochlear rim and increased trochlear wedging (anterior broadening). It also exhibits a laterally projecting fibular facet, a mediolaterally broad head, a low medial body, and a relatively wide distal half of the ectal facet. These limb bones suggest that Catopithecus moved more deliberately and climbed more frequently than the smaller parapithecids and proteopithecids. The astragalus reflects an emphasis on inverted and abducted foot postures, as well as powerful hallucial grasps. As with Aegyptopithecus, which it resembles, this astralagus may indicate regular hindlimb suspensory behavior.

The recent description of Talahpithecus parvus, from the late Middle or Late Eocene of Dur At-Talah in Libya, proves the existence of very small and highly crested oligopithecines (Jaeger et al. 2010). As mentioned above, the dental characters of O. savagei first led researchers to hesitate between catarrhine and adapiform affinities. After its anthropoidean status was clearly established, the task remained to assess whether the highly crested oligopithecines had been a primitive sister group of the bunodont (parapithecids + propliopithecids), with convergence in the anterior part of the dentition, or whether they were more closely related to the propliopithecids. Postcranials found subsequently have added a considerable amount of evidence in favor of the second view. The differences in molar morphology, which had previously impressed many scholars, can now be explained by a process of upper molar crest increase associated with a reduction of hypocone and lingual part, as seen in sivaladapids (Seiffert et al. 2004). Thus the loss of P/2 and the similarity in premolar morphology, instead of being convergences, probably reflect the close affinity of oligopithecids and propliopithecids. The presence of solely two premolars remains a hallmark of the catarrhines (and sometimes the oligopithecines are considered a subfamily of the propliopithecids, a possible choice).

The family Propliopithecidae, known only in the Oligocene until now, includes the genera Aegyptopithecus, Propliopithecus, and Moeripithecus. The best-known species by far is A. zeuxis, known by many skulls and several postcranial elements from quarries I and M of the Fayum. A cranium of a young male, found in 1966, has been described and depicted many times (e.g., Simons 1967, 1972; Szalay and Delson 1979). It has a relatively long muzzle (Fig. 31), orbits with a slight dorsal orientation in lateral view, tooth rows with a slight posterior divergence, etc. However, this cranium had been reconstructed from a fossil that had been shattered in its facial region. The later discovery of several nondistorted partial male crania brought to light some distortion in this first reconstruction, as well as very high morphological variability (Simons 1987). Comparison of four male crania revealed enormous differences between them: development of the sagittal crest from very low and starting somewhat posteriorly, to very high and shifted anteriorly just behind the orbits (Fig. 31); orbit orientation, in profile view, from slightly turned dorsally to vertical; zygomatic root and suborbital height, from moderate to stronger and much higher. These differences accompany differences in age as visible from tooth eruption and wear. They are themselves attributed to aging, with enlargement of the temporal ridges due to growth of the temporalis muscle and even deepening of the face. Aging also produces differences in ventral view: zygomatic arches mainly above M2/ in young individuals, and above M3/ in older ones; posterior concavities of the palate between M2/ and M3/ in a young male, and well posterior to M3/ in an old one (all indicating an anterior shift of the cheek teeth with age). These differences are further accentuated when males are compared with females, a small individual of which was subsequently described (Simons et al. 2007). The female specimen is minimally damaged and is the only cranium showing the exact placement of the braincase relative to the face in Aegyptopithecus (Fig. 32). An angle of 150–160° is measured between the plane of the palate and that of the basioccipital. Several quantitative analyses of tooth dimensions have been done to see if two species could be distinguished in the assemblage from the two quarries; however, until now they all conclude to one species with high cranial variability (Kay et al. 1981; Simons et al. 2007). In fact, A. zeuxis reveals for the first time in the Paleogene a sexual dimorphism affecting not only canines and cranial superstructures, but also the size of the postcanine dentition – something found to be the case in only the most dimorphic extant species.

Fig. 31

Schematic drawings of several crania of Aegyptopithecus zeuxis. (a) is the male cranium reconstructed with some possible deformation of the muzzle, (b) is the undistorted cranium of a female, and (c) is the dorsal view of the same; (d–f) are two other faces of males, one in profile (d) and anterior (e) views, the second only in anterior view (f). The extreme sexual dimorphism present in this species can be seen through the size difference between the female (b and c) and the males and also through the absence of sagittal and nuchal crests in the female. All these crests are present in (a), and a strong variability in the development of the sagittal crest can be seen among the males

Fig. 32

The undistorted cranium of a female Aegyptopithecus zeuxis from the Fayum, Egypt, in lateral (a) and dorsal (b) views at slightly different scales. This is the only specimen showing the exact relationships between face and braincase in this species (Courtesy of, and copyright by, P. D. Gingerich)

Among other noticeable features of Aegyptopithecus are its moderate rostrum length, anteriorly concave nasofrontal region in profile view, and broad interorbital region with a convexity in the medial orbital wall, resembling African great apes; the orbital convergence is 130–135°. The angle of divergence between the tooth rows is 12–13°. The characters of the otic region are those of anthropoideans as seen in parapithecids. However, the lateral pterygoid wing does not contact the auditory bulla. On three specimens, the dorsal part of the ectotympanic extends out in a process, suggesting the incipient development of a tubular ectotympanic (Simons et al. 2007). The use of micro-CT-scan has produced images of the endocast, as well as much better estimates of endocranial volumes than the previous ones: 14.6 cm³ for the female, instead of the earlier 27 cm³ for a comparable female cranium (Simons 1993); a new estimate of the young male endocranial volume, 20.5–21.8 cm³, is again lower than the earlier estimation of 27 cm³ by Radinsky (1977). The authors conclude that encephalization in Aegyptopithecus was very low, “at best strepsirhine-like, and perhaps even non-primate-like” (Simons et al. 2007), although the range of body size estimates is broad. Further scrutiny of body weight should allow more precise estimates of the brain-to-body mass relationship.

The dentition of Aegyptopithecus is well known. The two upper incisors are spatulate and have a complete lingual cingulum. I1/ is much larger than I2/ and has a lingual cingular bulging. The robust canines, much larger in males than in females, have an anterior vertical groove and a lingual basal cingulum. The cheek teeth are quite bunodont. The two premolars are subequal in size, bicuspid, and have an almost complete lingual cingulum. The anterolabial part of P3/ can be salient in males. The molars have moderately low and rounded cusps. M2/ is much larger and much more transversely elongated than M1/. The trigon is well formed, with preprotocrista going toward the anterolabial corner and postprotocrista going toward the metacone summit, forming a complete crista obliqua. On M1/ and M2/ the large hypocone is almost as high as the protocone, and linked to a complete lingual cingulum. The M3/ are highly variable. The two mandibles of A. zeuxis are high, robust, solidly fused anteriorly, and anteriorly increase in height in males. The incisors are relatively small and narrow, and moderately proclive. They have an elongated crown and a mesial cingulid (seen on DPC 1112). I/2 is slightly larger than I/1. The lower canine is high and pointed, more gracile than the uppers. Its lingual cingulid runs from the posterior base of the crown, curving and ascending toward its tip. The two lower premolars are different. P/4 is short and molarized, with a metaconid almost as high and as anterior as the protoconid; it has an arcuate paralophid, and a small talonid basin limited by a low hypoconid and a subhorizontal postcristid. P/3 is higher and more pointed. Its height in lingual view is underlined by a deep ventral expansion of the posterior cingulid, followed by a continuous anterodorsal ascending course of the cingulid. This difference is accentuated on the P/3 of males, which are larger, higher, and more pointed, and which have a longer preprotocristid for honing with the upper canine. The lower molars differ from those of oligopithecids by some features: absence of a paraconid on M/1, increased bunodonty, transversely broader M/2, M/2 larger than M/1, well-developed posterior cingulid, and a more centrally located hypoconulid, leaving space for a small valley or fovea between entoconid and postcingulid. M/3 is larger than M/2; it is essentially made longer through its much larger, and quite variable, hypoconulid.

Among the Fayum primate limb bones studied by Conroy (1976), one ulna was referred to Aegyptopithecus. It was found similar to that of Alouatta, the extant howler monkey. Several other bones were subsequently described; a summary of those attributed both to that genus and to Propliopithecus was given by Gebo (1993). The humerus of Aegyptopithecus is robust and shows strong muscle crests, especially the deltopectoral, and the brachial flange is more laterally extended than in any other anthropoidean. Among the characters which indicate arboreal quadrupedalism are the prominent and laterally placed tubercles bordering the posteriorly facing head, a slight flattening on the top of the head, and the very distal crest for the teres major muscle. There is a wide medial epicondyle (strong flexors), a dorsoepitrochlear fossa, and an entepicondylar foramen. On the distal articular surface, the capitulum is round and has a small capitular tail, the trochlea is relatively wide, the zona conoidea wide and shallow. The olecranon fossa, which is shallow, is one more primitive/quadrupedal feature, whereas the articular surface is of the nontranslatory type as in other catarrhines (Rose 1988; Gebo 1993). Most characters of the ulnae also indicate arboreal quadrupedalism: robust bone, proximal convexity, a relatively long olecranon and low coronoid processes, and a broad and shallow sigmoid notch with oblique orientation of the flexion-extension axis. The prominent pronator crest indicates climbing capabilities. The femur of Aegyptopithecus is a very robust bone, which retains a third trochanter (or gluteal tuberosity) – a character independently lost in the parapithecids and later catarrhines (Ankel-Simons et al. 1998). The tibia of Propliopithecus chirobates is short and robust, exhibits asymmetrical condyles, and shows a distal articular surface similar to those of Early Miocene proconsulids (Fleagle and Simons 1982). The astragalus of Aegyptopithecus shows a relatively long and moderately high body, and a short neck angled medially. The trochlea is asymmetrical, with a higher lateral rim, as well as a medial rim anteromedially curved above a deep medial malleolar cup (cotylar fossa), which produces abduction associated with dorsiflexion of the foot. The fibular facet projects quite far laterally, as in Catopithecus. The calcaneum is very broad mediolaterally in both Aegyptopithecus and Propliopithecus. Its plantar surface is slightly concave anteroposteriorly, and its proximal part slightly bent medially. Its distal part is moderate in length (38 % of total length). It exhibits an elongated posterior facet and an anterior extension of the sustentacular facet, both indicating extensive rotation and sliding of the astragalus on the calcaneum. Distally, a deep pivot indicates high rotational capacities. The first metatarsal is relatively long, curved, and possesses a relatively short peroneal tubercle, as in other anthropoideans. It presents a facet for a prehallux, a primitive character; two other metatarsals also present primitive characters on their narrow proximal part (Gebo and Simons 1987). A long, slender, and moderately curved proximal phalanx is typical of climbers. On the whole, most characters suggest arboreal quadrupedalism and climbing similar to Alouatta; increased foot mobility and strong hallucial grasping possibly reflect frequent suspensory feeding postures, as in the latter.

Other species and genera of propliopithecids include three species of Propliopithecus, Moeripithecus markgrafi, and an unnamed propliopithecid from Taqah, Oman. The three species of Propliopithecus are smaller than A. zeuxis, and they differ from it by a series of dental characters: lower crowned lower incisors, lower molars with a well-developed labial cingulid, more peripheral cusps and straight-sided crowns, and a transversely broader M1/. P. haeckeli, coming from an unknown level and described long ago (Fig. 33), is considered more primitive than the larger P. chirobates, known by samples from Quarries I and M of the Fayum (Schlosser 1911; Kay et al. 1981). P. ankeli from Quarry V is even larger and has transversely very broad lower molars. The type and only known specimen of Moeripithecus markgrafi is also an early find without precise stratigraphic provenance. It is a partial mandible with M/1 and M/2, showing a unique combination of characters. The molars have a strong basal crown inflation and a transversely short talonid basin; they are more crested than in other propliopithecids, and their twinned entoconid and hypoconulid recall oligopithecids. Treated for many years as a species of Propliopithecus, M. markgrafi is now recognized as a valid genus with probable transitional significance. In a similar vein, a better documented Oligocene species from Taqah, Oman, was first described as pertaining to M. markgrafi, but in recent years specialists have tended to consider it a distinct taxon – one also presenting characters that are primitive for propliopithecids (Thomas et al. 1991; Seiffert et al. 2010a). Whereas the lower dentition is close to those of other propliopithecids (a male has an enormous P/3), the upper molars are more transversely elongated and show much smaller hypocones, which are clearly primitive characters. The Taqah species, large and Oligocene, underlines the diversity of propliopithecids, which likely deserve the distinction of a fourth genus.

Fig. 33

Right lower dentition in occlusal (a) and lingual (b) views and reconstruction of the lateral view of the mandible of Propliopithecus haeckeli at two different scales (From Kälin 1961)

Systematic interpretations of the propliopithecids, mainly Aegyptopithecus, have changed dramatically through time, due to the discovery of more complete fossils and the increasing role of locomotor characters for primate phylogenetic reconstruction. The partial dentitions that were found first could reasonably be compared with those of living great apes. Discovery of the skull revealed the primitive, platyrrhine-like state of the ectotympanic. An entepicondylar foramen found on the humerus was also a primitive feature. These two characters prevented placement of Aegyptopithecus among crown catarrhines; however, a frequent move was to include Aegyptopithecus as a primitive catarrhine because it had only two premolars, and to modify the defining characters of the group accordingly. However, within the catarrhines, classifying Aegyptopithecus as a hominoid implied that the two characters at issue had evolved convergently in hominoids and cercopithecoids. For example, Simons (1987) insisted that the similarities shared by Aegyptopithecus and the great apes (deep face, temporal cresting, broad lacrimals and interorbital region) were shared-derived and implied a hominoid status for the propliopithecids. However, a more detailed study of the interorbital region and the distribution of its characters concluded that it would be more parsimonious if an African ape-like system of the ethmofrontal sinuses were the primitive condition for crown catarrhines (Rossie et al. 2002). Aegyptopithecus and probably Proconsul also were considered primitive sister groups of a clade (cercopithecoids + hominoids). The latter view has gained general support based on consideration of locomotor adaptation and associated characters: pronograde arboreal quadrupeds such as Aegyptopithecus preceded the more lightly built and tailless proconsulids, which themselves preceded the adaptive divergence of semi-terrestrial cercopithecoids and orthograde arboreal hominoids. Propliopithecids are primitive catarrhines, and likely a primitive sister group of crown catarrhines. They include the ancestors, or good approximates of the actual ancestors, of Miocene–Recent catarrhines. As such they are one of those indispensable paraphyletic taxa that are needed if evolutionary history is to be reconstructed.

The First Proconsuloid

At the end of the Oligocene, a large catarrhine of more modern aspect is found in Kenya. Kamoyapithecus hamiltoni is known by dental and gnathic remains only. It is a large animal; the length of its upper tooth row is roughly 1.5 times that of Aegyptopithecus zeuxis. The species is documented by a maxilla bearing P4/–M3/, the tip of an upper canine, an anterior mandibular fragment, and an I/2 (Leakey et al. 1995). Its cheek teeth are bunodont, low-crowned, with moderate labial and lingual flaring. P4/ is oval in outline. M2/ is only slightly larger than M1/ and has roughly the same size as M3/. The molars are transversely broad in comparison with Later Miocene proconsulids, but quite short and square in comparison with propliopithecids. A large hypocone is close to the protocone, present and slightly smaller on M3/ than on M1/ and M2/. The cingulum is not crenulated. The large robust upper canine may be a derived character shared with later forms, as is the position of the hypocone. The very well-developed superior mandibular torus is also found in large-size Proconsul species. As seen with the upper molar proportions, K. hamiltoni in many ways appears intermediate between earlier propliopithecids and Later Miocene–Recent catarrhines. It can be considered a primitive member of the informal proconsuloids.

Anthropoidean and Platyrrhine Origins, Afrotarsiidae, and Further Phylogenetic Questions

Platyrrhine Origins

Before discussing anthropoidean origins, it is necessary to envisage the origin of the Platyrrhini, the South American monkeys. They are documented in Bolivia in the “Salla Beds,” Salla-Luribay Basin, which were long believed to be Early Oligocene (Hoffstetter 1969). However, the Deseadan land mammal age was subsequently redated, and the Salla Beds are now considered Late Oligocene, or Oligo-Miocene (an age of 25–26 Ma is often mentioned). The genus Branisella with its only species B. boliviana is documented by dental material (Fig. 34). The upper cheek teeth are moderately bunodont; the lower ones are also clearly high-crowned. The roots indicate that P2/ was small. P3/ and P4/ have a well developed lingual part, with large protocone. On P4/, the protocone has a postprotocrista continuous toward the posterolabial part; there is a transversal groove between protocone and paracone, and a continuous and thick lingual and posterior cingulum. The upper molars have the three usual main cusps. It seems that the crista obliqua is sometimes interrupted by a transverse groove. There are important variations in the extension of the lingual cingulum and hypocone, leading to a triangular or more rectangular lingual outline. This led to the initial distinction of some specimens as another taxon “Szalatavus multicuspis,” but further specimens showed that these differences reflected high intraspecific variations (Rosenberger et al. 1991; Takai and Anaya 1996; Takai et al. 2000). M3/ is smaller than the other molars and can have a very reduced metacone (variable). The mandibles are fused and the dental arcades are close to a V. I/2 is larger and set more posteriorly than I/1. It has a complete lingual cingulid. The canine is oval in outline, with its longitudinal axis bent anterolaterally; it bears a continuous lingual cingulid. P/2 is not reduced. It has roughly the same size as P/3, is unicuspid, and bears a strong lingual cingulid, curving into a posterior one labially ascending toward the main summit. P/3 has a well-formed metaconid, lower and posterior to the protoconid, whereas P/4 has a metaconid roughly as high and as anterior as the protoconid. P/4 also has a small talonid. The three lower molars are transversely broad, with long labial slopes resulting in transversely short trigonid and talonid basins (Fig. 34). Their anterior paralophid is transverse and continuous; there is almost no labial cingulid, only a basal closing of the labial trough between protoconid and hypoconid. M/3 is especially short posteriorly; a hypoconulid may be identified on its rounded postcristid, but there is no third lobe.

Fig. 34

Dentition of Branisella boliviana. Maxilla with two upper premolars and three molars in occlusal view (a). A lower jaw with fused symphysis, preserving the anterior teeth, three premolars on the right side and the three molars on the left side, in occlusal view (b) and lingual view of its left anterior teeth P/3, P/2, and partial canine and I/2 (c). Photographs of white-coated epoxy casts, all at the same scale

Branisella is evidently the oldest known platyrrhine. It is sometimes considered as a good ancestral morphotype for them; or as more closely related to one living lineage, implying an earlier diversification (e.g., relatedness to callitrichines for Takai et al. 2000); or, more commonly, as an early primitive sister group of all other platyrrhines. All specimens found in the Salla Beds show strong dental wear, and it has been suggested that their high-crowned lower molars were an adaptation to an abrasive diet, probably related to the semi-arid environment revealed by the fossils and geology of these beds. Therefore, the likely existence of other taxa living in contemporaneous humid and forested areas has been postulated. Whatever its relationships with later platyrrhines, Branisella testifies that the group was present in South America by the Late Oligocene. Concerning its origins, all specialists agree that South American platyrrhines must have a common origin with African catarrhines – but when and where? Close affinity between the two groups suggested to Hoffstetter (1972) and others an African origin and subsequent dispersal to South America. However, a scenario of primates rafting over the Atlantic Ocean, which was already wide at the time, has often met with considerable skepticism (e.g., Conroy 1976). Alternatively, a common source on northern continents and subsequent dispersal to Africa and through North America to South America has been advocated – a proposal sometimes revived in conjunction with putative Asiatic anthropoideans (eosimiids and amphipithecids). However, as seen above and discussed further below, this Asiatic origin is not only far from established, it appears unlikely. With respect to the platyrrhines, it would imply a dispersal through North America – a continent with a rich and continuous fossil record, and yet one in which no suitable ancestor has ever been found. Furthermore, exchanges between the two Americas were interrupted between the Paleocene and the Plio-Pleistocene. This dispersal route for primates thus appears extremely unlikely. On the contrary, new finds in Africa have revealed that the proteopithecids were relatively close to the platyrrhines. Branisella and Proteopithecus both share the unusual combination of a reduced P2/ and unreduced P/2 (Takai et al. 2000). As seen above, Proteopithecus cannot yet be ranked among the platyrrhines. There is still a morphological gap between known African Eocene fossils and platyrrhines, although a common origin is the most consensual view. It is paralleled by the stronger case of an African origin for the South American caviomorphs. A chance dispersal from Africa to South America is thus postulated (Holroyd and Maas 1994). It probably implied the conjunction of several factors: intermediate land on the Mid-Atlantic ridge, a sea-level drop, the rafting of small animals able to survive seasonal food shortage, etc. If this is what occurred, African platyrrhines or stem platyrrhines should be found. It is a real possibility, given that all African anthropoideans named to date come from North Africa and Arabia, leaving the possibility that other groups lived in more southern regions (see below on intriguing fossils found in Namibia).

Anthropoidean Origins and Postcranial Characters

Anthropoidean, or simian, origins have been a field of lively debate during the last two decades. This field has become progressively polarized between three hypotheses: first, a hypothesis rooting the anthropoids in Eocene adapiforms; second, a more commonly adopted hypothesis linking them to the two Asiatic families Eosimiidae and Amphipithecidae; and third, a hypothesis of ancient origin in Afro-Arabia. On the whole, the Fayum anthropoid evidence of three distinct clades differentiated by the early Late Eocene and the implications of platyrrhine origins for a fourth clade jointly provide sufficient evidence to allow researchers to infer a long undocumented history of early simians in Africa, unless the Asiatic connection turns out to be true. As the Paleocene Altiatlasius (see below) is found to have possible simian characters, it reinforces the possibility of an ancient African differentiation for the group.

Can an adapiform ancestry be defended? This old hypothesis was revived by the discovery of similarities existing between teeth and dentaries of cercamoniine and hoanghoniine adapiforms, on the one hand, and those of oligopithecids, on the other (Gingerich 1975, 1977b). It was further advocated in studies of both adapiforms (Rasmussen 1990) and Fayum anthropoideans (Simons 1987; Simons and Rasmussen 1996). It was adopted in a study of Messel adapiforms (Franzen 1994) and further argued for in the recent study of Darwinius, supposed to be a haplorhine (Franzen et al. 2009). However, the discovery of oligopithecid crania showed that the dental similarities invoked earlier were in fact convergences. As mentioned above, key characters of the auditory region emphatically rule out the adapiform hypothesis (Cartmill et al. 1981). The key haplorhine cranial characters cannot be checked on Darwinius because its cranium is crushed; however, Darwinius is certainly a cercamoniine adapiform, and in this group the well-preserved skull of Pronycticebus shows that they are strepsirhines and certainly not haplorhines. Difficulties linked to the haplorhine concept (Tarsius + simians) have been mentioned above. If they were to lead to a view of convergence instead of homology for haplorhine characters, this would not be in favor of the adapiform view, but rather in favor of a third group yet to be identified.

The Asiatic connection could imply multiple dispersals between Asia and Africa. For example, the schema arrived at by Beard et al. (2009) allies eosimiids with African simians and suggests a closer phylogenetic link between amphipithecids, propliopithecids, and platyrrhines. This implies two dispersals, one for eosimiids, and a second one for amphipithecids in either direction, return to Asia in their schema or second dispersal to Africa with a less parsimonious tree. As seen above, amphipithecids cannot be close to propliopithecids, and a more consensual schema would imply one dispersal of an eosimiid giving rise to the African simian radiation (Williams et al. 2010). This is a real possibility (Fig. 36). However, it implies that the Shanghuang petrosal attributed to an eosimiid by MacPhee et al. (1995) be ascribed to an omomyid instead (Kay et al. 1997), which may appear unlikely. That hypothesis is partly sustained by postcranial characters, which have been interpreted as shared-derived between eosimiids and anthropoideans. This requires a critical look.

As indicated above, the Shanghuang fissure fillings have yielded not only eosimiids but a series of four morphological groups based on tarsal characters (Gebo et al. 2001). Study of the calcanea allows their placement along a morphocline between small primates with short and broad calcanea, adapted to quadrupedalism and climbing (“protoanthropoids: new taxon”), to markedly elongated calcanea, adapted to leaping (“prosimians: tarsiids”). Intermediates are the “protoanthropoids: eosimiids,” which are slightly broad, and the “prosimians: unnamed haplorhines,” which are narrow and very omomyid-like (Fig. 35). Astragali complete the tarsal characters of these four groups. The locomotor diversification between these groups took place among tiny Middle Eocene primates of the same region in China, in the presence of a larger omomyid (Macrotarsius) and a larger and more climbing-adapted adapid. Is it reasonable in such a context to hypothesize that some of these small primates would be in the process of reversal from a prosimian-like leaping (omomyid-like, not extreme) to quadrupedalism? This change would have had to be so marked that a medial tibial facet extended to the ventral border of the astragalus would retract to a more dorsal position – something that can hardly be explained in adaptive terms. Much more parsimonious is the view that, because a reduced medial facet is primitive in primates, quadrupedalism was the likely primitive primate locomotor mode (or, if a component of leaping was present, it was low or recent enough to have led to some long bone lengthening as in Archicebus, but not yet to medial astragalar and to tibial modification). This fits with the notion of short and broad calcanea as probably primitive in primates, as indicated by extra-group comparisons, reinforcing quadrupedalism and climbing as primitive in the group. Several degrees of leaping were successively reached by different groups, with probable parallelisms between them (such convergent evolution is demonstrated among living primates by the fact that VCL specialists have evolved convergently in at least four families of extant primates). Increasing specializations are expected during evolutionary radiations. Reversals can happen, but usually only under special conditions (further adaptive transitions). Under these assumptions, the anthropoidean-like tarsals of Shanghuang in large part retained postcranial characters that are primitive for primates. They do not prove anthropoidean affinities. This also applies to the astragali of amphipithecids from the Pondaung Formation (Marivaux et al. 2003, 2010; Dagosto et al. 2010). Eosimiid tarsals are probably not primitive for all their characters. For example, might their high astragalar neck angle be convergent with that of adapiforms? The peculiar cuboid facet, with its removed articular wedge, is clearly derived. However, analogous modifications of the cuboid facet have been described, for example in adapines and amphipithecids (Godinot 1992a; Ciochon and Gunnell 2004). A more detailed study of this facet in many groups is needed, including functional aspects, to decipher homologous and convergent changes. At present, the anthropoidean affinities of eosimiids based on tarsal characters do not appear convincing.

Fig. 35

Schematic drawings of the four types of calcanea identified by Gebo et al. (2001) in the Shanghuang fissure fillings, all in dorsal view. They constitute a morphocline from a broad and short morphology typical of generalized arboreal quadrupeds (a) to progressively narrower and more elongated calcanea until the extremely elongated one in (d), which reflects a high leaping specialization. (d) is ascribed to a tarsiid, (b) to an eosimiid; (c) appears omomyid-like or cheirogaleid-like in proportions. This morphocline can also be most simply interpreted from primitive in (a) to most derived in (d). In this case, the anthropoid-like proportions in (a) would not imply phylogenetic affinity with simians but only primitiveness among primates

The significance of the anthropoidean postcranium is an old problem. Anatomists have recognized for a long time that living simians are generally more primitive in their postcranial anatomy than living prosimians. Small simians do not show the degrees of specialization reached by some prosimians for leaping or for hand and foot powerful grasping. Because for dental and cranial anatomy there is evidence that early simian characters evolved from prosimian characters, many authors since Gregory (1920) have assumed a similar scenario for postcranial anatomy. This model, Model 1 of Dagosto (1990), implies the reversal of a whole suite of characters – of the knee joint, the first metatarsal joint, the tarsals, etc. Other scholars have considered it more likely that anthropoideans retained their primitive postcranial characters from the euprimate morphotype (Ford 1988; Godinot 1992b). Dagosto’s Model 1 explained these reversals through an emphasis on above-branch quadrupedalism, which could have accompanied a marked increase in size. Indeed, at the time when this model was proposed, known fossil anthropoideans were larger than most early fossil prosimians. Strangely, the same scenario of reversals has been maintained for the extremely small eosimiid fossils discovered more recently in China (Gebo et al. 2000, 2001). However, their Middle Eocene age and extremely small size render such a scenario questionable – and quite unparsimonious, as seen above for the tarsal characters (and as could be argued for first metatarsal and distal phalangeal characters). The series of tarsal groups found in the Shanghuang fissure fillings gives strong support to the view that primitive anthropoidean-like postcranial features are retained from a primitive primate ancestor. And in fact, the discovery of an anthropoidean-like calcaneum in the Earliest Eocene Archicebus provides an even more striking confirmation (Ni et al. 2013). A full understanding of the locomotor adaptation of the earliest primates will require further study and more fossils. At present, the postcranial evidence for eosimiids is roughly that of primitive primates, which does not prove their anthropoidean affinities.

The Asiatic connection rooting African simiiforms in the Asiatic eosimiids is essentially based on parsimony analyses of large datasets (e.g., Marivaux et al. 2005; Beard et al. 2009; Seiffert et al. 2005a, b, 2010b; Ni et al. 2013; Fig. 36). As mentioned above concerning amphipithecids, these analyses can be misled by convergences due to similar selective forces (e.g., bunodonty linked with similar frugivory and/or hard food requirements). They also suffer recurrent difficulties concerning the independence of characters, their weighting, missing data, etc. The number of characters and taxa will never replace the absence of intermediates or make up for uncertainties due to the lack of what would be the most diagnostic characters. An example in primate phylogeny was given by Marivaux et al. (2001), in which a global parsimony analysis nested Bugtilemur within living cheirogaleids when this fossil lacked a real tooth comb. It thus could not be a lemuriform. Linked to the discovery of a related and more complete genus, Bugtilemur was subsequently recognized as a probable adapiform (Marivaux et al. 2006). In this case, the many similarities with cheirogaleid jugal dentition overweighted the phylogenetic signal issued from the tooth comb. A high number of convergences in dental details can obscure the true phylogenetic relationships. Such will be the case for all the large data sets used to uncover primate phylogeny as long as several fossil groups lack key phylogenetic information, and at the same time are separated by large gaps in the fossil record, preventing the recovery of enough intermediates in other suites of characters. It is worth noting here that such analyses can also generate trees which imply as many as six dispersals of stem anthropoideans between continents (Ni et al. 2013), which appears historically unrealistic. A morphologically parsimonious tree can be historically unlikely or even impossible.

Fig. 36

The two main current hypotheses of anthropoid origins. Left is one of the consensual views emerging from the cladistic analyses of large data sets supported, for example, by Williams et al. (2010). In this schema, African simians are rooted in early Asiatic Eosimiidae, which are considered simians or stem-simians having no close relationship with tarsiids. Right, another view, considered more likely here, implies a longer history of early African simians with a probable African origin of the typical anthropoid characters. In this case, Eosimiidae and possibly Amphipithecidae could be related to tarsiids in an extended tarsiiform radiation (however, some of the taxa included here might also appear as derived omomyiforms as indicated by ?). Both hypotheses preserve a haplorhine clade, stem-based and extended to include omomyiforms in Williams et al. (2010), more restricted and character-based here. If Tarsiidae themselves appeared rooted in omomyiforms, anatomical haplorhinism would have lost any phylogenetic meaning and reflect only a convergence in cranial adaptation

Among the reasons to doubt an Asiatic origin of simians is lack of historical likelihood. Anthropoideans are a well-adapted, successful group in Africa and South America, including the Caribbean islands. If they had been present in Asia, why didn’t they leave some group of successful survivors behind there? Asiatic forests are the refuge of many primitive mammals, many tree shrews, and Tarsius found refuge in islands; they were the refuge of adapiforms until the Late Miocene, until the spread of African monkeys. Would they have been the place of the first radiation of anthropoideans without leaving any Asiatic descendants? Such a scenario seems very unlikely. If there had been a radiation of Eocene-Oligocene anthropoideans in Asia, successful enough to colonize Africa through one or multiple dispersals, they would have left descendants in Asia too, alongside the cohort of primitive Asiatic mammals. The extinction of all early Asiatic anthropoideans would be a very unlikely historical event – incongruous with the remarkable ability of all kinds of groups, including primates, to survive over long periods of time. That argument is not definitive, of course, because the past is also full of surprises. However, paleontologists should pursue a coherent account of history as much as, if not more so than, a parsimonious distribution of morphological characters considered in isolation from the geological and geographic context of the fossils to which they belong. The argument of historical likelihood, added to the lack for now of convincing evidence, should lead us to consider favorably the third scenario – that of an ancient differentiation in Africa (Fig. 36).

Afrotarsiidae

The enigmatic genus Afrotarsius is gaining great importance in connection with the scenarios laid out above. A. chatrathi is known through one mandible bearing M/1–M/3 and parts of the base of the crown of P/4 and P/3. Its molars with high pointed cusps and its trigonids with paraconid similar from M/1 to M/3 are close to those of Tarsius, and it was initially described as a tarsiid (Simons and Bown 1985). Unsurprisingly, some large parsimony analyses place it close to Tarsius (Seiffert et al. 2005a). It differs from Tarsius by several characters, including an M/1 larger than M/2–3 and M/3 without an elongated third lobe. Possible ties with early anthropoideans were also found (Fleagle and Kay 1987; Kay and Williams 1994), and similarities with eosimiids were noted (Ross et al. 1998; Godinot 2010). New material attributed to the new species A. libycus includes two upper molars and two P3/, an important addition (Jaeger et al. 2010). These teeth add to the similarities with eosimiids; however, the upper molars also show the postmetaconule-crista directly linked to the posterolabial corner of the tooth, a very primitive character which must have been retained from an ancestor more primitive than eosimiids, leading to use of the family Afrotarsiidae proposed earlier (Jaeger et al. 2010). The P3/ are very small and their labially deflected postparacrista suggests a marked insectivorous specialization.

Another putative afrotarsiid, Afrasia djidjidae, was recently described from the Pondaung Formation of Myanmar (Chaimanee et al. 2012). This species is known by four isolated teeth only, coming from three different localities. The type M2/ is very similar to that of Afrotarsius libycus, but there are some differences, and several characters of the other teeth are more reminiscent of eosimiids. The placement of A. djidjidae in afrotarsiids is tentative, and is associated with a cladogram showing several evidently incorrect parts (Teilhardina as a terminal branch of Omomyidae + Tarsiidae). The authors conclude that an afrotarsiid dispersal took place between Asia and Africa shortly before these Middle Eocene localities. Such a conclusion is hasty. Characters linked to an insectivorous specialization are difficult to disentangle from primitive characters. Much more material will be needed to understand the differences between eosimiids (Eosimias with long P/3 and P/4) and afrotarsiids (Afrotarsius with broad and short P/4). What will Afrasia be? The real afrotarsiids might equally well have issued from a much earlier dispersal, which would have given birth to a new adaptive radiation of stem simians or simians. This would increase and broaden the spectrum of adaptations in the two parallel radiations of African simiiforms + stem simiiforms, and Asiatic tarsiiforms. Earlier and more complete fossils are needed to clarify these hypotheses.

Further Important Phylogenetic Questions, and a Provisional Conclusion

A number of important issues are associated with the understanding of fragmentary fossils. Such finds are tantalizing, because the implications of their analysis might be far-reaching; but they are also frustrating, because the associated discussion is very technical and the results are bound to remain tentative, due to such specimens’ incompleteness or isolation from other, better understood fossils. First in the list is Altiatlasius koulchii, from the Late Paleocene Adrar Mgorn locality in Morocco. As the only Paleocene primate known to date, it is very important. Yet it is documented only by eight isolated teeth. The upper molars show extremely primitive characters (stylar shelf), but also some characters – such as a slight bunodonty and a continuous lingual cingulum on M2/ – which would make it derived in comparison with Donrussellia and Teilhardina, and possibly a sister group of the African simian radiation (Sigé et al. 1990; Godinot 1994). A recent phylogenetic analysis, based on a very large data set, positions Altiatlasius as the earliest member of the stem anthropoideans (Ni et al. 2013). Such far-reaching conclusion needs to be strengthened by more complete fossil evidence, especially since some early African strepsirhines appear to have been very bunodont (Tabuce et al. 2009). Yet this unique fossil is an important signal, because it might indeed document stem haplorhines or stem anthropoideans.

Two fossils from the Middle Eocene of Namibia are the first Eocene primates named from Sub-Saharan Africa. They are a maxilla with M2/ and M3/, which is the type of Namaia bogenfelsi (Pickford et al. 2008), and a smaller P/4, which is also referred to it. The upper maxilla belongs to a primitive primate that has overall similarity with European anchomomyin adapiforms, but also significant differences. Its lingual half is anteroposteriorly more reduced than in anchomomyins, and it bears a well-formed cuspidate metaconule, such as is never seen in anchomomyins and extremely rare in adapiforms. The P/4 is unlike that of any adapiform and might suggest simian affinities, although it has a smaller size than the maxilla and could pertain to a different taxon. More material is needed to determine the number and the affinities of the primates present in this locality.

Finally, to give a last example, the enigmatic Nosmips was described from the BQ-2 locality of the Fayum, Egypt (earliest Late Eocene). Its surprising characters led to the name N. aenigmaticus (Seiffert et al. 2010b). Its lower molars have lost their paraconids, but their trigonids nevertheless increase in length from M/3 to M/1. Inasmuch as their trigonids are much higher than their talonids, they are more reminiscent of the lower molars of prosimians than of simians; however, the anteroposterior cristid obliqua and the short M/3 talonid look simian-like (primitive?). One upper molar appears very simple, with an incomplete lingual cingulum and no hypocone. A posterior waisting gives it a primitive stamp. This species is specialized through its premolars, with both P3/ and P/3 anteroposteriorly long and high. The lower P/3 has a voluminous metaconid and an anteriorly extended and curved protocristid, which presumably had a honing function with a large upper canine. The P/4 is lower, more molarized, and morphologically intermediate with the lower molars. Parsimony analyses of 361 morphological characters give, depending on different assumptions, very different results: stem anthropoidean, stem lemuriform, or adapiform status (Seiffert et al. 2010b). This example once more demonstrates the unability of such large analyses to uncover a good phylogenetic signal in the presence of overlapping dental similarities with species belonging to different infraorders. Lower molar morphology, as well as long, highly molarized, and specialized P/3–4 are more suggestive of stem lemuriform affinities, which represent the most rational assumption. However, other affinities are also possible. In any case, this genus reveals one more ancient lineage in Africa, which, along with the genera cited above, highlights the fact that our knowledge of early African fossil primates remains very incomplete.

To conclude our discussion of primate phylogeny: the last decade has witnessed a number of surprising discoveries, which show that our knowledge of the Paleogene primate fossil record is still only very partial. The European and North American radiations are relatively well circumscribed, with only a few open questions for now: Where does Rooneyia come from? Where do the adapines come from? The African and Asiatic records are much less completely known and have been the subject of the greatest surprises: Afradapis, an adapiform convergent with catarrhines in its anterior dentition; the azibiids, tiny strepsirhines with upper molars convergent on bunodont simians; Bugtilemur, an adapiform convergent with living cheirogaleids; Nosmips, an enigmatic form for which there is hesitation between strepsirhine and simian! All of these findings call for caution concerning the interpretation of dental characters for fragmentary fossils pertaining to distantly related groups. Nevertheless, great advances have been made. Among them, the discovery of Eocene lemuriforms and stem lemuriforms in Africa is decisive for our understanding of the origin of this living infraorder. It supersedes all previous, speculative efforts to root these groups near different genera of adapiforms. The discovery of Teilhardina asiatica shows that intercontinental dispersals can be precisely followed also from Asia to other continents (this should be possible for the putative anthropoidean ancestors as well, then!). A broad and monophyletic Asiatic family of amphipithecids is also a step forward, even if their origin and phylogenetic place remain controversial. Archicebus bears directly on concepts of the earliest primate diversification. A complete fossil, such as the skeleton of Darwinius masillae, should produce further information on rarely found parts of the primate skeleton. Ongoing functional and phylogenetic study of hands and feet, tarsals, and phalanges should continue to contribute significantly to the understanding of primate phylogeny. This field is very active, and still has a lot to discover.

Conclusion: Evolution in Paleogene Primates

The fossiliferous part of the Paleogene, Eocene plus Oligocene, spans roughly 30 My. Over the course of this time interval, it is possible to observe major steps in primate evolution. These steps can best be studied in the regions where the fossil record is the richest, the North American and the European Eocene, even if the fossil groups found there became extinct without leaving descendants among the living groups. The well-known Eocene radiations started with the arrival of one genus each, e.g., Teilhardina, Cantius, or Donrussellia. From this genus a diversification took place, accompanied by increases in size in some of the lineages. When the record is dense enough, dental morphology allows an almost direct reading of lineages. More often the record is not as good as that, but the relevant evolutionary trends are so general that reading the polarity of dental traits is usually straightforward. Only the large number of convergences can obscure the reconstruction of phylogeny, as seen for example in omomyines. These dental trends raise interesting questions. Why does the paraconid become reduced? And, why does a hypocone develop at the same place in so many different lineages? Hypocones occur at roughly the same size, suggesting the existence of a developmental constraint, which it would be important to understand. Incidentally, this question brings up the difficult issue of character coding: should characters that we know not to be homologous be coded in the same way? For example, the hypocone in omomyines is not homologous with that of microchoerids, and the latter is not homologous with the hypocones which develop in several lineages of adapiforms, etc. This information is most often ignored in cladistic analyses. Reduction of some antemolar teeth is also a common trend, which is sometimes related to a size increase in the anterior incisor and “compression” of teeth between it and P/4, but at other times occurs without evident dental specialization and is then probably related to muzzle shortening.

Changes in molar morphology are related to changes in diet, and diet is highly correlated with size. Small species weighing below 500 g are usually predominantly insectivorous, whereas species above 500 g become more frugivorous or folivorous (Kay 1975). There are many mixed feeders among living and fossil primates, and most often their protein intake comes from insects for the small species, and from leaves for the large ones. The broad picture is that the Late Paleocene–Earliest Eocene primates were mainly insectivorous and more or less mixed feeders, with Cantius already quite frugivorous. By the end of the Early Eocene, some larger species had evolved more specialized dentitions. For example, Notharctus had acquired molars with increased shearing crests, forming the labial W on upper molars that we see in living folivores. By the Middle and Late Eocene, other genera, such as Caenopithecus and Afradapis, had developed similar morphologies, and lower molar shearing crests had strongly increased in the adapines. More bunodont dentitions evolved many times. Extreme degrees of bunodonty, adapted to frugivory and/or the processing of hard food, developed in Asiatic amphipithecids and in African parapithecids and propliopithecids. Some fossils developed dental specializations unknown in living primates. Necrolemur, Microchoerus, the tarkadectines, and other genera developed highly crenulated teeth, probably adapted to some kind of resistant food. The very small azibiids are astonishing: how could such tiny species have become so extremely bunodont? Paleogene primates broaden the spectrum of dental and dietary adaptations of the order.

Alongside the Paleogene primate diversifications, locomotor adaptations evolved as well. Quite a number of the Early Eocene prosimians are reconstructed as “cheirogaleid-like” –which indicates a mixture of quadrupedalism, climbing, and leaping. However, there is debate about the ancestral morphotype of locomotor adaptation, with or without much leaping (as mentioned above in the anthropoidean origin part). In some groups, it seems that locomotion changed essentially through a more or less marked increase in leaping propensities (weak in omomyids, more pronounced in the larger notharctines). It has recently been suggested that some tarsal lengthening might be a compensatory effect in grasping foot postures, and not an indication of leaping propensities (Moyà-Solà et al. 2012). This hypothesis needs to be closely examined. Only during the Middle and Late Eocene did one lineage, the Necrolemur–Microchoerus group, increase its leaping specialization to a degree similar to the living VCL, which move through the forest by long leaps and adopt vertically clinging postures at rest. However, we know that very different adaptations developed in some other groups. Quadrupedalism and climbing (slow climbing?) are present in amphipithecids and adapids. The Late Eocene–Oligocene anthropoideans also show quadrupedalism and climbing, with more leaping propensities in parapithecids. A variety of adaptations developed among European adapines, including different degrees of quadrupedalism and climbing. To date, there is no clear evidence of ateline-like forelimb suspension in Paleogene primates.

Important aspects of the evolution of sense organs and the brain can be traced as well in Paleogene primates. Most characteristic since their origin are the large eyes of primates. It has been known for a long time that nocturnal primates can be distinguished from diurnal ones by their possession of relatively larger eyes, as mentioned above for many fossil groups. A deeper understanding of visual evolution became possible once size measurements of the optic foramen were used to infer the degree of retinal summation and visual acuity in fossils (Kay and Kirk 2000). Visual characteristics are typically shared by large groups; however, fossils also reveal multiple changes. Most omomyiforms are reconstructed as nocturnal, yet Teilhardina asiatica was found to have small orbits, like diurnal species. Among cercamoniines, Europolemur was found to have small orbits, like diurnal forms (such as the adapines and Notharctus); however, Pronycticebus and Darwinius had orbits indicating nocturnal habits. The anthropoidean Biretia megalopsis exhibits nocturnal-sized orbits as well. All of these examples reveal that important behavioral shifts took place in the Eocene. There will be more to learn about this crucial aspect of primate behavior. Adapis is found to have a strangely high degree of retinal summation. Why? No adapiform or omomyiform studied by Kay and Kirk (2000) showed optic foramen dimensions as large as those in extant diurnal anthropoideans, whose visual acuity (inferred for diurnal ancestral haplorhines because Tarsius has a fovea) is extremely high. Finding intermediate values in some fossils would be crucial.

The use of CT-scan techniques allows access to quantitative parameters of the bony labyrinth. Two kinds of sensory capacities can be extracted. Cochlear labyrinth volume is correlated with hearing abilities (Kirk and Gosselin-Ildari 2009). Ongoing studies have already yielded results. For instance, assessments of high and low frequency limits of hearing show that Necrolemur antiquus had better high frequency hearing than three other fossils, similar to that of the living Galago senegalensis, whereas Adapis had capacities similar to those of the living Perodicticus (Ludeman et al. 2013). The size and morphology of the semi-circular canals also give interesting information. The first studies interpreted relative semi-circular canal size, estimated via the radius of curvature, to correlate directly with relative degrees of agility among species (Spoor et al. 2007; Silcox et al. 2009). This way, adapines were found to be less agile than Smilodectes and Notharctus, and Microchoerus faster and more agile than the latter, which was in line with the modes of locomotion of these taxa as reconstructed from postcranials. However, further studies have put these results in question, because it appears that it is the orthogonality of the semi-circular canals, more so than their radius of curvature, that is related to the speed of head movements (Malinzak et al. 2012).

The size of the infraorbital foramen is highly correlated with the size of the infraorbital nerve, which transmits signals from the mechanoreceptors of the orofacial region. It had been assumed that the infraorbital foramen, which is smaller in primates than in most non-primate mammals, was larger in strepsirhines than in haplorhines (Kay and Cartmill 1977). This size difference within mammals was interpreted to roughly correlate with vibrissae number, and was used in this way in several studies to support a haplorhine status for fossils showing a relatively small infraorbital foramen (e.g., Beard and Wang 2004), or conversely, to infer a strepsirhine and probably nocturnal habit for others (e.g., Tabuce et al. 2009). However, a much larger study of primates and other mammals has failed to confirm such differences in terms of vibrissae counts. Haplorhines and strepsirhines appear not to differ in infraorbital foramen area, nor in macrovibrissae counts, but to differ in microvibrissae counts – and in the opposite direction from what was expected: it is haplorhines that have more (Muchlinski 2010a). Information about the mechanoreceptors of the oral region can be gained from the relative size of the infraorbital foramen; however, for now the implications appear to be primarily ecological, with frugivores having larger foramina and more vibrissae than insectivorous and folivorous primates (Muchlinski 2010b). This approach already fits better with the relatively small infraorbital foramen found in Adapis (Gingerich and Martin 1981).

Important aspects of brain evolution in primates are obtained through the study of endocranial casts. The latter provide not only quantitative information – that is, absolute brain measurements – but also qualitative information, because most neocortical sulci delimit functional or somatotopic areas. If these sulci can be confidently homologized, behavioral and sensory specializations can be inferred. The oldest primate endocast is that of Early Eocene Tetonius homunculus, known since its preliminary description by Cope. Taking into account the modest deformation of the skull, Radinsky (1967) proposed a reconstruction with a width–length index of 1.07. He noted that the frontal lobes were smaller relative to the rest of the cerebrum than in any extant primate. Only a shallow Sylvian fissure can be recognized, but this is not an indication of primitiveness because similar-sized living prosimians have similarly smooth brains. The olfactory bulbs of Tetonius are relatively larger than in any living primate. As noted by Radinsky, these primitive characters notwithstanding, compared to contemporaneous ungulates Tetonius had a very advanced brain with enlarged temporal and occipital lobes and reduced olfactory bulbs. A natural endocast of the Middle Eocene Smilodectes gracilis was described in detail and beautifully illustrated by Gazin (1965). It is also very smooth, shows only a shallow sulcus lateralis not far from the midline of the brain, and, surprisingly, does not feature a recognizable Sylvian sulcus. Gazin noted the extended neopallium, which is relatively greater than in any other Middle Eocene mammal for which such information is available. He also noted a close similarity with the brain of extant Eulemur, insofar as the cerebellum is not overlapped by the cerebrum. The cerebellum of Smilodectes is short and narrow in comparison with its width, and the vermis and lateral lobes are prominently developed. Both Tetonius and Smilodectes, an Early and a Middle Eocene primate, show expanded temporal and occipital areas, suggesting well-developed acoustic and auditory capacities, respectively. Gazin (1965) rightly correlated the enlarged visual cortex area with the large, forward-facing orbits of Smilodectes. Recent analyses show that binocular vision is correlated with the expansion of visual brain structures, and consequently with expansion of overall brain size, in primates (Barton 2004). More generally, advances in visual capacities have been correlated with increases in brain size in primate evolution (Kirk 2006).

The endocast of the Late Eocene Adapis was the first to be described (Neumayer 1906) and has been mentioned many times by many authors. A better preserved endocast was described by Gingerich and Martin (1981). Both endocasts exhibit a true Sylvian sulcus, a universal trait in extant primates but lacking in Smilodectes. Among other possible differences between notharctine and adapine brains are the large and pedunculate olfactory bulbs in Adapis, relative to which those of Smilodectes are small (Radinsky 1970). This distinction aside, both appear primitive in comparison with living primates. Attempts to quantify encephalization through a quotient relating brain size to body size have led to some controversy, mainly due to difficulties in estimating body size, i.e., body weight, in fossil primates (Radinsky 1977; Jerison 1979). In any case, there is consensus that both Adapis and Smilodectes had smaller brains in relation to body size than do living strepsirhines. On the other hand, the Late Eocene Necrolemur and the younger Rooneyia are found to have encephalization levels similar to those of living prosimians. The endocast of the Early Oligocene Aegyptopithecus was mentioned above. It shows a central sulcus as in living anthropoideans, and appears advanced over strepsirhines in having relatively more visual cortex and smaller olfactory bulbs. However, in quantitative terms it is smaller than in any living anthropoidean, having an encephalization level similar to some prosimians or even below (Radinsky 1973; Simons 1993; Simons et al. 2007). The endocast of Parapithecus (Simonsius) grangeri has been extracted by CT-scanning techniques (Bush et al. 2004). It also appears smaller relative to body size than in living anthropoideans. Its olfactory bulbs are at the lower limit of those of strepsirhines.

A last source of information relating to behavior is provided by sexual dimorphism, which was long suspected in Notharctus (Gregory 1920; Gingerich 1979) and convincingly shown in one assemblage of N. venticolus (Krishtalka et al. 1990). Evidenced through canine size and shape, it was shown subsequently to affect skull shape in N. tenebrosus and Smilodectes gracilis (Alexander 1994; Alexander and Burger 2001). Sexual dimorphism is suspected in Cantius torresi based on the size difference between two canines, and size differences in mandibular depths in six specimens (Gingerich 1995). It thus seems to be a characteristic of notharctines. Its presence in adapines is unknown, due to the lack of homogeneous assemblages (the only available one, for the large Magnadapis from Euzet, does not present any dimorphism; Gingerich 1977b). A marked sexual dimorphism has been found, however, in the latest Eocene adapid Aframonius (Simons et al. 1995). An extremely high dimorphism, including body size dimorphism, is found in propliopithecids. The conclusion that a marked sexual dimorphism is also found in the smaller Proteopithecus and Catopithecus modified the view, based on observations in living primates, that sexual dimorphism is highly correlated with body size. It appears to be a characteristic of simiiforms, acquired early and probably linked to their diurnal habits (Simons et al. 1999). Sexual dimorphism is an indication of distinctive social structures , usually interpreted as indicating life in polygynous groups, high male-male competition when it is pronounced, and single male dominance when it is highly pronounced. However, this is a complex question; precise social structures cannot be inferred, especially for low levels of dimorphism. What is important is that social evolution along these lines also happened in some adapiforms in the Early Eocene, and was advanced in Late Eocene anthropoideans. Social life is an important predictor of increased brain size, which is believed to have played a major role in later simian brain evolution. In sum, not only was binocularity present in the earliest primates; higher visual acuity and social factors as well were already present in Paleogene primates, enabling later brain developments.

When the fossil record is good, each primate species can be studied for its adaptations, and faunistic aspects can also be analyzed. In the big picture, primate radiations are constrained by climatic and biogeographic factors, as laid out in the Introduction to this chapter. Within these radiations, interesting primate faunal successions are known, which are more or less understood. For example, the replacement of cercamoniines by adapines in Europe around MP13–MP14 (Figure 7) is due to a dispersal. But why did mid-sized or large cercamoniines disappear? Were the more folivorous adapids better adapted to changing environments? In the Late Eocene of Africa, in the Fayum, adapids were the large-sized primates, whereas anthropoideans and lemuriforms of the same localities were smaller. Adapids disappeared by the end of the Eocene, and anthropoideans became progressively larger during the Early Oligocene. Were these developments also related to environmental changes? A much better-studied example, made possible by a more extensive geological and paleontological record, is the Early–Middle Eocene transition in several of the Rocky Mountain basins of Wyoming and adjacent states, from anaptomorphine dominance in the Wasatchian to omomyine dominance in the Bridgerian. Furthermore, the diversification of the omomyines in the Bridgerian is striking. This raises two interesting issues. First, continued research in basin margin areas (southwestern Bighorn Basin, northeastern edge of the Green River Basin), has led to the realization that anaptomorphines remain in fact relatively diverse in these marginal areas (e.g., Bown 1979; Muldoon and Gunnell 2002). The shift from anaptomorphine to omomyine dominance in lowland habitats probably was an ecological replacement, with anaptomorphines being pushed toward upland refugia when omomyines had found favorable habitats and reached dominance in basin areas. The second issue relates to the rapid diversification of the omomyines: did basin margins provide heterogeneous habitats favorable to the rise of evolutionary innovations? Answering such a question requires detailed paleoenvironmental studies. What is remarkable in the Rocky Mountain Eocene record is that it provides, in the Bighorn and in other basins, some of the best examples of detailed lineages – phylogeny followed through time as close as one can get, and also, at a more regional scale, fine-grained aspects of replacements, refugia, and their possible role in evolution. In such locations, the study of fossil primates can contribute to the detailed elucidation of the mechanisms of evolution.