Early primate evolution in Afro-Arabia

Setting the stage: “Island Africa” and Afrotheria

Combined geophysical and paleobiogeographic evidence indicates that the African and Arabian plates (Afro-Arabia) together formed a single landmass that had no permanent connections to other continents from ∼120 to ∼20 Ma; that is, through the later part of the Early Cretaceous, the entire Late Cretaceous and Paleogene, and part of the early Miocene.5 Despite Madagascar's proximity to the southeast, Afro-Arabia has not been directly connected to that island since the Jurassic.6 During the Late Cretaceous, Madagascar had close geographic and biogeographic links to the Indian subcontinent,7 but together the two bodies had already become isolated from other Gondwanan continents.6 India would later move north to collide with Asia, with overland vertebrate dispersal corridors likely first appearing in the early Eocene.8 Afro-Arabia's last connection before isolation was with South America, at ∼120 Ma, and its slow northward drift finally terminated when it collided with southwest Asia, permanently closing off the Tethys Sea, which previously had separated Afro-Arabia from Eurasia.9 The timing and nature of this collision is complex and currently a matter of debate. The process probably initiated in the later Paleogene, but the first major pulse of terrestrial vertebrate faunal exchange, which is assumed to reflect an ephemeral land connection between Arabia and southwest Asia, is not known to have occurred until ∼20 Ma.10 Faunal exchange continued to be strongly filtered well into the Miocene, for instance by a major marine transgression in the early middle Miocene.9

Very little is known about the terrestrial vertebrate fauna that inhabited Afro-Arabia during the Late Cretaceous and early Paleocene, between ∼95 and ∼60 Ma. The few fossils that have been discovered from Late Cretaceous sites show that Afro-Arabia was home to an archosaur fauna that included various crocodyliforms and both predatory theropod and herbivorous sauropod dinosaurs,7 but with the exception of a single caudal vertebra dating to about 95 Ma,11 no mammals are known. Several mammals have been found in later Cretaceous beds on other Gondwanan land masses (Madagascar, South America, and India) but almost all of these species are unequivocally derived from archaic, non-therian clades.12 The only clear records of Cretaceous-aged placental mammals in Gondwana are from India, and these taxa, of which Deccanolestes is best known, fall outside of the placental crown clade based on available evidence.13 Early crown therians and the stem lineage of Placentalia apparently diversified largely in the Cretaceous of Laurasia (Fig. 1).14

Given that so little paleontological work has been undertaken in the later Cretaceous of Afro-Arabia, the current lack of mammalian fossils from this time and place is more likely to be an artifact of inadequate paleontological sampling than a true absence. In fact, the now well-established molecular phylogeny of placental mammals,15, 16 when interpreted within the context of available paleontological evidence, predicts that the morphologically diverse supraordinal clade Afrotheria (which contains elephants, sea cows, hyraxes, aardvarks, sengis or elephant-shrews, golden moles, and tenrecs) originated in Afro-Arabia during the later part of the Cretaceous (Fig. 1).17, 18 This radical reorganization of mammalian phylogeny has been well-established for just over a decade, and is only now gaining broad acceptance.3 The existence of an ancient, endemic Afro-Arabian clade of placentals is a critically important piece of this biogeographic puzzle, which was not available to earlier students of mammalian evolution, but now provides an elegant explanation of the odd Fayum faunas that were first encountered by Andrews1 and Schlosser2: the diverse afrotherian fauna was the result of a Late Cretaceous origin in Afro-Arabia and tens of millions of years of evolution in isolation, while the presence of all other non-afrotherian placentals had to be the result of more recent overwater dispersals, most likely from Europe or Asia.

THE OLDEST AFRO-ARABIAN PRIMATES AND OTHER EARLY CENOZOIC IMMIGRANTS

Primates are not members of Afrotheria, but rather are most closely related to colugos or “flying lemurs” (order Dermoptera)16, 19, 20 and, more distantly, tree-shrews (order Scandentia), within a clade called Euarchonta (Fig. 1).15, 18 Euarchonta and the larger clade Euarchontoglires, which includes the more distantly related rodents (order Rodentia) and rabbits and pikas (order Lagomorpha), is almost certainly of Laurasian origin, based on each order's first appearance in the fossil record (Fig. 1).17 The oldest possible stem primate is earliest Paleocene Purgatorius from North America,21 but definitive members of the primate crown clade (that is, the clade containing all extant primates and their last common ancestor) do not appear in the fossil record until near the Paleocene-Eocene boundary, ∼56 Ma. By that time, cladistic stem strepsirrhines (basal adapiforms) and stem or crown haplorhines (basal omomyiforms) are already detectable on the northern continents,22-27 indicating a more ancient origin for crown primates in the Paleocene, if not earlier. Recent work refining molecular divergence dates for primate clades suggests that the last common ancestor of crown primates probably lived near the boundary between the Cretaceous and the Paleogene, about 66 Ma or slightly earlier.28, 29 If this is correct, the first 10 Ma of crown primate evolution are currently missing from the fossil record. Afro-Arabia would have been moving away from the peak of its geographic isolation during this interval (Fig. 1), but still would have been completely isolated, both by its geographic position and a relatively high (but falling) global sea level.30 At the close of the Cretaceous and through the early part of the Paleogene, Afro-Arabia would have been closer to southwest Europe (the Iberian Peninsula and nearby islands) than to Asia (Fig. 1).5

Despite Primates' probable Laurasian origin, by the close of the Eocene (34 Ma) a remarkable array of primate groups had made appearances on the isolated Afro-Arabian landmass4, 31: azibiids, djebelemurines, basal crown strepsirrhines and definitive lorisiforms, plesiopithecids, caenopithecine adapiforms, afrotarsiids, parapithecoids, proteopithecids, catarrhines, and two particularly enigmatic primates called Altiatlasius32 and Nosmips.33 The biogeographic origins of this diversity can only be understood by placing these primate lineages in a broader phylogenetic context.

The oldest euarchontan from Afro-Arabia is latest Paleocene Altiatlasius from Morocco.32 Given our current understanding of early primate evolution, Altiatlasius is enigmatic, due to its antiquity, geographic placement, and fragmentary nature (being known from only ten isolated teeth and a mandibular fragment holding a second lower molar). Phylogenetic analyses have placed Altiatlasius as either a basal stem anthropoid,22, 24, 25, 28, 34-36 a tarsiiform,36 a basal haplorhine of uncertain placement relative to anthropoids and tarsiiforms,24, 34 or even as a plesiadapiform.37 It has been argued that Altiatlasius suite of upper and lower molar characters is more similar to that of later primitive anthropoids than what is seen in the basal omomyiforms and adapiforms that are known from the early Eocene on northern continents,38-40 but its few anthropoid-like dental features, such as relatively well-developed lingual cingula on the anterior upper molars, have evolved convergently in non-anthropoids several times over the course of primate evolution. Unfortunately, in light of this, much more information is needed to establish Altiatlasius' affinities with any confidence.

On northern continents, the earliest Eocene was marked by the first appearances of crown primates and their rapid intercontinental dispersal at high latitudes, evidently facilitated by an intense episode of global warming that has been called the Paleocene-Eocene Thermal Maximum (PETM).41 Terrestrial mammalian response to the PETM has now been documented in Asia, Europe, and particularly North America, but practically nothing is known about this time period in Afro-Arabia because few mammal sites of earliest Eocene age are known. Primates have not been recovered at any of the early Eocene sites in Africa (for example, El Kohol in Algeria and Ouled Abdoun in Morocco). These localities also provide no unambiguous evidence of immigration of nonprimate mammals from northern land masses (marsupials and adapisoriculids being possible exceptions5), seemingly attesting to Afro-Arabia's continued isolation from Europe and Asia by the Tethys Sea. Definitive fossil evidence for filtered faunal exchange does not occur in the Afro-Arabian record for another five million years, near the boundary between the early and middle Eocene stages (∼49 Ma) (Figs. 2, 3, and 4). Important sites of this approximate age include Chambi in Tunisia and Glib Zegdou in Algeria, both of which have yielded remains of primates,35, 42-44 as well as Afro-Arabia's oldest rodents, the Zegdoumyidae.45 The latter are undoubted immigrants from Eurasia that ultimately gave rise to the Anomaluroidea, Africa's extant scaly-tailed flying squirrels (Anomaluridae) and springhares (Pedetidae).45, 46

The early-to-middle Eocene primates from northwest Africa have perplexed students of early primate evolution for decades, but new discoveries have finally helped to clarify their identities. The first primate recovered from Glib Zegdou was Azibius trerki, which Sudre described in 1975 on the basis of a partial lower jaw with three teeth.47 Before the discovery of more complete remains of Azibius reported in 2009, this odd species had been interpreted as a plesiadapiform, an adapiform, or even a non-primate mammal.35 Two isolated teeth, an M² and an M₃, of a smaller primate from Glib Zegdou, Algeripithecus minutus, were reported in 1992 by Godinot and Mahboubi.43 Morphologically similar but larger upper and lower molars were later reported by the same authors48 and formed the hypodigm for another new genus and species, Tabelia hammadae, which, like Algeripithecus, was interpreted as a primitive anthropoid. The anthropoid identity of Algeripithecus and/or Tabelia would come to be widely accepted by a number of other authorities24, 39, 40, 49-53 and, in the case of Algeripithecus, supported by phylogenetic analysis.26 The supposed presence of primitive anthropoids at such an ancient horizon in Afro-Arabia had a major influence on the anthropoid origins debate for about 15 years.

New material of Algeripithecus and Azibius described by Tabuce and coworkers35 has surprisingly revealed that Algeripithecus is, however, not an anthropoid at all, but rather is closely related to Azibius, and that together these “azibiids” are likely allied with crown (“toothcombed”) strepsirrhines (lemurs, lorises, and galagos). With upper teeth of Azibius now available for comparison, it is also clear that “Tabelia” is in fact Azibius, and that the hypodigm of another genus from Glib Zegdou, “Dralestes,” is composed of specimens that are attributable to both Algeripithecus and Azibius.35 In one fell swoop, these discoveries removed the only compelling records of Anthropoidea and Plesiadapiformes from this period in Afro-Arabia, and revealed that primate taxonomic diversity at Glib Zegdou was actually much lower than previously thought, with all taxa being members of one very strange primate family. With the benefit of hindsight, it is no wonder that isolated teeth of Algeripithecus, Azibius, “Dralestes,” and “Tabelia” engendered so much confusion – despite their great antiquity, aziibids are highly specialized, and indeed are some of the more bizarre primates known from the Eocene of any continent. They combine multicusped and superficially anthropoid-like upper molars with elongate, strepsirrhine-like lower molars, and have long, serrated upper and lower premolar teeth that have large accessory cusps of uncertain homology. Tabuce and colleagues'35 identification of azibiids as stem strepsirrhines has since gained additional support from the recovery of azibiid astragali that exhibit probable strepsirrhine synapomorphies.42

DJEBELEMURINES AND THE ORIGIN OF CROWN STREPSIRRHINES

The primate fauna from the roughly contemporaneous Chambi locality in Tunisia includes three primates: the tiny species Djebelemur martinezi, known from a mandible with P₃-M₃ and an isolated lower canine54; a small unnamed azibiid, known from isolated upper molars, that is similar to and possibly congeneric with Algeripithecus55; and a much larger adapiform represented by two lower molars.56Djebelemur was described as a member of the adapiform subfamily Cercamoniinae,54 a potentially paraphyletic group known largely from the Eocene of Europe, but soon thereafter was interpreted as a possible basal anthropoid38 or as a cercamoniine that “may lie very near the prosimian-anthropoidean transition.”53 The phylogenetic analysis of Seiffert, Simons, and Attia57 later placed Djebelemur not as an anthropoid or cercamoniine, but as an advanced stem strepsirrhine, more closely related to “toothcombed” strepsirrhines than any adapiforms from northern continents. This relationship is supported by a series of derived features of the lower postcanine teeth that Djebelemur shares with other advanced strepsirrhines known from younger sites in Afro-Arabia, but Djebelemur presumably falls outside of the strepsirrhine crown group because it lacked a toothcomb. Godinot4, 55 has since accepted and discussed this scenario in greater detail, and formally proposed that Djebelemur and its younger Afro-Arabian relatives be referred to as “djebelemurines.”

Key to the recognition of Djebelemur as an advanced stem strepsirrhine was the recovery of the first definitive Paleogene toothcombed strepsirrhines, Karanisia clarki and Saharagalago misrensis, at a younger (37 Ma, earliest late Eocene) locality in the Fayum Depression of Egypt in 2001.57 These discoveries marked a major breakthrough for the study of strepsirrhine evolution because fossil lorisids and galagids had long been known from early Miocene (< 20 Ma) sites in East Africa,58 suggesting a much more ancient diversification of lorisiforms in Afro-Arabia. Mysteriously, however, older and more primitive forms had never been found. The absence of crown strepsirrhines in the Paleogene defied explanation, particularly given that the late Eocene/early Oligocene Fayum sites had been worked intensively for decades and had otherwise produced a diverse primate fauna that included both large and small species. In the absence of a better explanation, this odd pattern led some to suggest that crown strepsirrhines might have originated in Asia and only later dispersed to Afro-Arabia.59Karanisia and Saharagalago were interpreted, respectively, as a crown lorisid and a stem galagid–identifications that required lorises and galagos to have already diverged by the middle Eocene; the lemuriform-lorisiform split must have been much older.57 More recent phylogenetic analyses incorporating a larger number of taxa and characters have called into question the initial identification of Karanisia as a crown lorisid and generally place the genus in a more basal position as either a stem lorisiform or a stem lemuriform.

Saharagalago is still known only from one upper molar and one lower molar, but the suite of upper molar features that it shares with Miocene-to-Recent galagos (notably a deep notch in the distal margin of the tooth and a large, distolingually protruding lobe for the hypocone) is unique among primates. The galagid relationship that was proposed for Saharagalago has since gained support from the discovery of more complete remains of a younger genus, Wadilemur, which is known from a ∼34 Ma Fayum site called Locality 41 (L-41) (Fig. 5E). Like Djebelemur, Wadilemur was originally identified as a cercamoniine adapiform based on the morphology of its P₃-M₃,60 but was later found to have a large and anteriorly protruding P₂ crown as in many toothcombed primates. In addition, Wadilemur is now known to have had galagid-like morphology of the upper molars and proximal femur.27 Like Saharagalago, Wadilemur appears to be a stem galagid,4, 27 but is situated closer to the galagid crown group than that taxon in phylogenetic analyses, which is consistent with its younger age.

Three other tiny primate species, one from the latest Eocene of Egypt (“Anchomomys” milleri) and two from the early Oligocene of Oman (Omanodon minor and Shizarodon dhofarensis) were, like Djebelemur and Wadilemur, initially identified as probable cercamoniine adapiforms, specifically aligned with a European group called anchomomyins, based on very limited material.60, 61 In light of what has recently been learned about early strepsirrhine evolution in Afro-Arabia, it now seems that these species are more likely to be djebelemurines than anchomomyin cercamoniines. The lower molars of djebelemurines differ from those of anchomomyins in very subtle ways, but there are obvious differences in the antemolar teeth. As in toothcombed strepsirrhines, “A.” milleri has three rather than four premolars, the second premolar is single-rooted, and the anterior premolar crowns project anteriorly. The lower canine of “A.” milleri is particularly interesting in also being relatively procumbent, mesiodistally elongate, and blade-like. It appears to represent an intermediate morphological condition between the more upright lower canine of typical adapiforms and the greatly procumbent and elongate canines of crown strepsirrhines. A clade containing “A.” milleri and Djebelemur (that is, a monophyletic Djebelemurinae) was first recovered by Seiffert, Simons, and Attia's phylogenetic analysis in 200357 and has since been found in some,27 but not all,25, 36 subsequent analyses of an expanded dataset.

It remains to be determined whether a clade including djebelemurines and crown strepsirrhines is in fact nested within a European radiation of anchomomyin adapiforms. Several recent phylogenetic analyses have recovered such a result.25, 28, 35, 36 This possibility is interesting in light of the recent discovery of Namaia bogenfelsi, a primate from the middle Eocene of Namibia.62 This species is remarkable both for its antiquity and its location, being the first Eocene primate from sub-Saharan Africa. Unfortunately, it is represented solely by a maxilla with the crowns of M2, 3 (the P₄ referred to N. bogenfelsi is from another locality that is probably considerably younger, and furthermore is too small to belong to the same species as the holotype maxilla). Namaia was described as an anthropoid, but, in my opinion, is likely a stem or crown strepsirrhine (again, for lack of a better term, a possible “djebelemurine”). Unlike anthropoids known from the later Eocene of North Africa, Namaia has an M2 that is transversely narrow and completely lacks a lingual cingulum around the protocone, but has a distinct hypocone. Among Eocene primates, this upper molar pattern is most similar to that of some European anchomomyins.

The only other Paleogene Afro-Arabian species of relevance to the problem of crown strepsirrhine origins is the bizarre plesiopithecid Plesiopithecus teras, known solely from the terminal Eocene Fayum Locality 41. While the dentition of djebelemurines apparently remained fairly conservative through the Eocene, the anterior dentition of Plesiopithecus is very specialized, bearing a single enlarged and procumbent tooth anterior to the premolars on each side of the lower jaw. This enlarged tooth resembles the canine teeth of crown strepsirrhines and is presumably homologous,63 though it does differ in some key details – for instance in having a root that is in line with the long axis of the crown, rather than being mesially inflected. The only enlarged tooth of the upper dentition is the canine,64 which would have occluded distal to the enlarged lower tooth (that is, there is no evidence for enlarged upper incisors that would have occluded tip-to-tip with the enlarged lower tooth).

Simons and coworkers^63:194 suggested that Plesiopithecus used these enlarged anterior teeth to gnaw on resistant food items, and that the species “filled the niche in the food chain of the African Paleogene occupied elsewhere by plesiadapiforms and arboreal squirrels.” One possibility is that Plesiopithecus is derived from a lineage that initially had a toothcomb like those of generalized crown strepsirrhines, but subsequently lost the intervening incisors and enlarged and reoriented the canine roots as part of the transformation of these teeth into a gnawing or gouging apparatus. Plesiopithecus was initially identified as an anthropoid65 before the recovery of its cranium,64 which revealed that it had enlarged orbits and no postorbital closure. Its lower molars are superficially similar to those of later Paleogene anthropoids in being quite bunodont, buccolingually broad, and low-crowned; the lower premolars are also superficially anthropoid-like in being obliquely implanted (Fig. 5F). The upper molars of Plesiopithecus are very generalized and have a complete lingual cingulum, as in early anthropoids, but also early strepsirrhines such as Karanisia. These anthropoid-like dental features occasionally pull Plesiopithecus over to the anthropoid side of the primate tree in phylogenetic analyses,25, 33, 36 though that position is here considered to be unlikely.

AN UNEXPECTED RADIATION OF ANTHROPOID-LIKE ADAPIDS

As noted earlier, the early-middle Eocene Chambi locality has also yielded two lower molars of a relatively large primate species. At the time of its description as “an enigmatic mammal” by Court,56 the only candidate adapiform from Afro-Arabia was Djebelemur, explaining his reluctance to attribute these specimens to Primates. He noted similarities to both sivaladapid adapiforms and non-primate hyopsodontid condylarths. Late Eocene beds of the Fayum have since confirmed that there was a radiation of true adapids that diversified alongside djebelemurines and basal crown strepsirrhines after an independent trans-Tethyan crossing (possibly from Europe, Fig. 4). The Chambi teeth indicate that this dispersal had occurred by the beginning of the middle Eocene.

The existence of this adapid clade was definitively confirmed in 1995 with the description of Aframonius dieides, a ∼1.5 kg species from the terminal Eocene Fayum Locality 41.66Aframonius has elongate lower molars with well-developed shearing crests (Fig. 5D), greatly reduced upper and lower second premolars, large and sexually dimorphic canines, and fairly generalized upper molars that bear large hypocones.67 As in some other adapiforms, the mandibular symphysis fused in older individuals. Aframonius was initially interpreted as another Afro-Arabian cercamoniine that was relevant to the problem of anthropoid origins, but subsequent discoveries have shown that Aframonius lacks postorbital closure and has fairly enlarged orbits (work in preparation). Godinot68 argued that Aframonius should be removed from Cercamoniinae and placed in a new subfamily of Adapidae, the Caenopithecinae, along with the middle Eocene genera Caenopithecus (from Europe), Mahgarita (from North America), and Adapoides (from Asia). The hypothesis of a monophyletic Caenopithecinae has since been supported by large-scale phylogenetic analyses.25, 27, 33

The later discovery of a highly specialized caenopithecine, Afradapis longicristatus (Fig. 5C), at the ∼37 Ma Locality BQ-2, revealed that the group must have undergone a major diversification during the middle Eocene in Afro-Arabia. At ∼3 kg, Afradapis was the largest primate known from the Afro-Arabian Eocene, and indeed one of the largest primates known from the Eocene of any continent. Afro-Arabian anthropoids did not come to occupy this body mass range until well into the early Oligocene, about 6 Ma later. Afradapis had dorsoventrally deep and robust mandibular corpora, a fused mandibular symphysis, and numerous elongate shearing crests on its upper and lower molars that are arranged in a pattern like that of folivorous howler monkeys (Alouatta). The premolar dentition of Afradapis is particularly remarkable; uniquely among “prosimians,” the genus lost its upper and lower second premolar and enlarged the lower third premolar into a massive bladelike tooth that sheared against the upper canine. This arrangement is otherwise seen only in catarrhine anthropoids that appear much later in the Fayum succession, but it was never taken to such an extreme among Paleogene catarrhines. Afradapis' molar morphology leaves little doubt that the species was highly folivorous,33 while its loris-like astragalus suggests a highly mobile ankle joint and a probable preference for cautious climbing rather than leaping.36 Perhaps Afradapis occupied a niche similar to those of extant Alouatta species, while living alongside advanced stem and basal crown anthropoids whose feeding and locomotor ecology more closely resembled that of various living cebid species. It is fascinating that Afradapis might provide a phylogenetically independent example of this adaptive strategy because a similar niche has been inferred69 for propliopithecid catarrhines such as Aegyptopithecus, which occur in deposits that are several million years younger, in the same restricted part of northeast Africa.

THE ORIGIN AND DIVERSIFICATION OF AFRO-ARABIAN ANTHROPOIDS

Anthropoid diversity in the Fayum succession was long assumed to be the result of an ancient anthropoid presence in Afro-Arabia.40, 51, 53 For many years, this hypothesis appeared to be bolstered by the early occurrence of forms such as Algeripithecus, Altiatlasius, and “Tabelia.”38 With the affinities of late Paleocene Altiatlasius very much open to debate, the combined removal of early-middle Eocene Algeripithecus and “Tabelia” from the anthropoid clade35 and the discovery of several candidate basal anthropoids in the middle and late Eocene of Asia70, 71 has forced a reconsideration of this long-held view.72 There are no longer any compelling records of Anthropoidea in the early and middle Eocene records of Afro-Arabia; only a tiny lower molar trigonid and talonid from the ∼44-46 Ma (middle Eocene) Aznag site in Morocco, described as a possible tarsiiform,73 remains as a possible middle Eocene anthropoid. A P₄ from Namibia referred to middle Eocene Namaia62 displays some anthropoid-like features, but was found at a site that is of contested age, as it also contains rodents that are similar to those from the early Miocene of East Africa.45, 74

In light of this, the oldest undoubted Afro-Arabian anthropoids are those that are known from the ∼37 Ma BQ-2 Locality in Egypt26 and the roughly contemporaneous Bir el-Ater (Nementcha) site in Algeria.75 Recently, it has been incorrectly claimed that an anthropoid-bearing primate fauna from Dur at-Talah in Libya is 38-39 Ma in age,76 but it is probably several million years younger, perhaps ∼35-36 Ma, and certainly younger than BQ-2 based on mammalian biochronology.77 The Bir el-Ater site was the first to yield remains of Biretia, a genus that was long known from only a single lower molar.75 Work at BQ-2 has since considerably expanded our understanding of the genus, with multiple specimens representing most of the upper and lower postcanine dentition, as well as fragmentary mandibles and a partial maxilla that has revealed some interesting details of Biretia's orbital floor.26Biretia is now known from three species, all of which are tiny (< 300 g) bunodont anthropoids with molar cusp patterns that are similar to, but more primitive than, those of later parapithecids known from the latest Eocene/early Oligocene Jebel Qatrani Formation. The upper molars of Biretia have well-developed hypocones and variably developed conules intervening between the three primary cusps, while the lower molars have distinct hypoconulid cusps and reduced or absent paraconids on the more posterior teeth. Phylogenetic analyses that have included this new material of Biretia have placed the genus as a basal member of a parapithecoid clade,25, 26, 28, 33, 34, 36, 72 a result that is consistent with the diversification of parapithecoids that has been documented in younger beds in the same area. Biretia is unique among living and extinct anthropoids in having a very compressed orbital floor; the orbital lamina is fused to the palate and the lingual root of M² is not only exposed in the orbital floor, but is itself apically compressed. This pattern was interpreted as a spatial correlate of orbital hypertrophy, potentially implying that Biretia was nocturnal,26 but a more recently recovered maxillary specimen that preserves a portion of the orbital rim indicates that Biretia probably did not have particularly enlarged orbits.78

One other as yet undescribed anthropoid species occurs at BQ-2. It is much more common than Biretia, and now is represented by the entire upper and lower dentition, some cranial parts, and numerous postcranial bones. As reported in an earlier abstract,79 this species shares dental features with slightly older Bahinia [from Burma (=Myanmar)], and younger oligopithecids such as Catopithecus and Oligopithecus but, unlike oligopithecids, retains three upper and lower premolars and has relatively simple premolars. It is now clear that this new species is also similar to the poorly known younger genus Talahpithecus from Libya.76 Unlike Biretia, the undescribed BQ-2 form and Talahpithecus lack conules on the upper molars, have tiny hypocones, and entoconid and hypoconulid cusps are closely appressed on the lower molars.

Work at BQ-2 has also uncovered isolated teeth of a very odd primate, Nosmips aenigmaticus, whose higher-level affinities are currently unclear (Fig. 5A).33 The molars of Nosmips are similar to those of later anthropoids in being relatively short (mesiodistally) and having reduced paraconids and broad talonid basins; however, they differ in having poorly developed hypoconulid and entoconid cusps. The P₄ is also somewhat like those of derived anthropoids in having an enlarged metaconid cusp that is connected to the protoconid by a tall crest, but the tooth is relatively elongate, with a long, curved paracristid. This pattern is further exaggerated on P₃, which bears a particularly long, arcuate paracristid. The upper molars appear to be quite conservative, with no conules and no hypocone, while the P³ has elongate buccal crests that would shear against the elongate P₄ paracristid. Seiffert and coworkers33 ran several phylogenetic analyses, altering assumptions about ordering and scaling of multistate characters, and found that Nosmips was not consistently placed in Anthropoidea. Intriguingly, it would occasionally emerge as the sister taxon of Plesiopithecus. Nosmips' phylogenetic position simply cannot be determined without more informative remains.

The younger Talahpithecus-bearing primate fauna from the Dur at-Talah beds in Libya, at present made up entirely of isolated teeth, also includes species of Karanisia, Biretia, and, surprisingly, Afrotarsius,76 a form that is otherwise known from early Oligocene beds in the Fayum area. Afrotarsius has been the topic of ongoing debate since its initial description as a tarsier relative in 1985.80 Some authors have argued that it might be a basal anthropoid.52 Previously known only from a partial mandible with M_1-3 and the P₄ talonid, the new discoveries from Libya include the first upper molars of the genus, which reveal a remarkably primitive occlusal morphology. Though Jaeger and colleagues76 interpreted Afrotarsius as a stem anthropoid “eosimiiform,” it shares features with the primitive Miocene tarsier Tarsius sirindhornae81 (from Thailand) that are not seen even in the most primitive eosimiids.71 The new evidence suggests to me that Afrotarsius is indeed a tarsier relative.

THE ASIAN CONNECTION

The anthropoids from BQ-2 and Bir el-Ater are the oldest from the Eocene of Afro-Arabia, but they occur at the end of a long (∼10 Ma) gap in the Afro-Arabian fossil record that has been filled only by the poorly documented Namaia-bearing Black Crow site in Namibia and the Moroccan Aznag locality (Figs. 2 and 3). Anthropoids could have arrived in Afro-Arabia at any point during that interval. So where did they come from? This is not a question that can be answered by considering Afro-Arabia alone, because many specialists now consider Asian eosimiids to be basal stem anthropoids, while the much more derived amphipithecids, known primarily from Burma and Thailand, are generally situated in a more advanced position within Anthropoidea, potentially within the crown clade (Fig. 6).33, 82, 72 Amphipithecids pose major problems because they are not represented by complete cranial remains. Also, there is considerable uncertainty surrounding attribution of isolated primate postcranial bones from the primary amphipithecid-bearing sites in Burma; some of these bones resemble those of adapiforms while others more convincingly resemble those of undoubted anthropoids.83, 84 By the late middle Eocene, the teeth of large-bodied amphipithecids are certainly anthropoid-like, but are already very highly specialized, leading some to argue that amphipithecids are adapiforms that evolved their anthropoid-like dental features convergently.83 Furthermore, the calibration-free molecular dating analyses of Steiper and Seiffert28 suggest that the anthropoid crown clade originated close to, or even after, some of the most highly specialized amphipithecids had left their remains in the ∼37 Ma Pondaung Formation of Burma – again arguing against their placement as crown anthropoids.

Several recent phylogenetic analyses have included Paleogene anthropoids, but these studies differ in their sampling of characters and taxa, and, not surprisingly, in their results. One way of summarizing all of these different results is to use a “supertree” method (matrix representation with parsimony or MRP), which converts the source trees from each study into their own matrices, in which members of each clade are assigned a derived state; all of the resulting matrices from each source tree can then be analyzed together to find the most parsimonious result from all of the studies combined. An MRP analysis of early anthropoids is presented in Figure 6. The resulting supertree places eosimiids and Altiatlasius as the most basal stem anthropoids, while amphipithecids intervene between crown Anthropoidea and a proteopithecid-parapithecoid clade. When biogeography is mapped onto this supertree, four trans-Tethyan dispersals from Asia to Africa are reconstructed, one that explains the early presence of Altiatlasius, one that explains the later presence of Afrotarsius (placed as a tarsiid in the supertree), and two independent dispersals to explain the presence of 1) a proteopithecid-parapithecoid clade and 2) a crown anthropoid clade in Afro-Arabia. This reconstruction hinges on the position of the early Oligocene primate Bugtipithecus, from Pakistan, a genus known from only a few isolated teeth that was described as an amphipithecid,24 but is placed outside that clade in the supertree. If Bugtipithecus is moved to be a basal amphipithecid, the biogeographic reconstruction for the entire anthropoid stem lineage changes to being ambiguous (that is, either Afro-Arabian or Asian).

At present, I consider it likely that amphipithecids are stem anthropoids that originated and diversified entirely in Asia, while the proteopithecids, parapithecoids, and crown anthropoids that are known from the Paleogene of Afro-Arabia are derived from a single middle Eocene dispersal from Asia, with a subsequent dispersal of platyrrhines to South America from Afro-Arabia. This pattern is consistent with evidence for strongly filtered trans-Tethyan exchange between Afro-Arabia and Asia in the middle and late Eocene, leading to competition between newcomers from Asia (anthropoids and hystricognathous rodents) and previously established clades derived from colonizing lineages that arrived in the early-middle Eocene, possibly from Europe (caenopithecines, djebelemurines, and zegdoumyid rodents) (Fig. 4). Anthropoids would ultimately come to dominate the later Paleogene primate faunas of northern Africa in the same way that hystricognaths dominated the North African rodent faunas (Fig. 7). A trans-Tethyan dispersal in the opposite direction has also been documented, with the anomaluroid rodent Pondaungimys, unambiguously of Afro-Arabian origin,45, 46 appearing in southeast Asia in the ∼37 Ma fauna that also includes amphipithecids and Bahinia. Hyaenodontid creodonts probably also participated in this exchange, though the direction and timing of dispersal(s) are currently difficult to evaluate because their relationships are poorly understood. Anthracotheriid artiodactyls, which would also enjoy great success in the later Paleogene of Afro-Arabia, dispersed from Asia to Afro-Arabia, apparently in the later Eocene, after the deposition of BQ-2.85 A close biogeographic connection between Afro-Arabia and Asia in the Eocene is also supported by the shared presence of afrotarsiids.72

By the terminal Eocene, three anthropoid clades–oligopithecids, parapithecoids, and proteopithecids–were present in Afro-Arabia (Figs. 2, 3, 6, and 8) and living alongside the last strepsirrhines known from the Paleogene of northern Africa (Aframonius, “Anchomomys” milleri, Plesiopithecus, and Wadilemur).53 All four of these strepsirrhine lineages went extinct, at least locally, near the Eocene-Oligocene boundary, and these disappearances have been linked to the early Oligocene global cooling event.86 In contrast, judging from their abundance at the ∼34 Ma Fayum Locality 41, all of these anthropoid clades were quite successful, presaging the later success and further diversification of catarrhines and parapithecids in the early Oligocene. By the terminal Eocene, basal catarrhines, best represented by the oligopithecid Catopithecus browni, had already evolved a number of distinctive dental apomorphies (notably the loss of P²/₂ and the presence of sexual dimorphism in P³/₃ size) and postcranial apomorphies that are shared with propliopithecids and crown catarrhines.87-89 Like all other Paleogene anthropoids known from crania, Catopithecus and Proteopithecus evidently had relatively small brains when compared with those of extant anthropoids, but otherwise exhibited anthropoid synapomorphies such as full postorbital closure, small and convergent orbital apertures, fused metopic sutures, and intraorbital lacrimal foramina.89, 90 Unlike crown anthropoids, Catopithecus and Proteopithecus clearly lacked full fusion of the mandibular symphysis, but the suite of apomorphies that Catopithecus shares with later catarrhines suggests that fusion evolved independently three times – i.e., in platyrrhines, catarrhines, and parapithecoids. The postcranial remains of Catopithecus show that it was probably a more cautious climber than sympatric Proteopithecus, which exhibits postcranial features that are consistent with more rapid locomotion and pronograde leaping. These reconstructions have now been supported by analysis of these species' semicircular canal sizes.91 The molar teeth of Catopithecus have relatively well-developed shearing crests, suggesting that the species was either a frugivore that supplemented its diet with insects, or was partially folivorous,92 while Proteopithecus was more clearly a generalized frugivore. Like parapithecids and propliopithecids, both Catopithecus and Proteopithecus had sexually dimorphic canines that presumably reflect a polygynous mating system.93 These genera occur alongside Arsinoea, still known only from very limited remains, and Serapia, which is very similar to Proteopithecus and probably a close relative.31

Propliopithecids such as Aegyptopithecus, Moeripithecus, and Propliopithecus appear in northern Africa in the early Oligocene, and have teeth like those of generalized crown catarrhines from the early Miocene of East Africa. Aegyptopithecus has long been known from well-preserved cranial and postcranial remains,31 the latter of which closely resemble those of extant howler monkeys. A small juvenile female Aegytopithecus cranium has recently helped to confirm the presence of extreme sexual dimorphism in molar size, body mass, and craniofacial morphology within the species94 – a pattern that is continued in early Miocene catarrhines. The recently discovered facial and basicranial remains of Saadanius hijazensis, from the late Oligocene of Saudi Arabia, strongly suggest that the facial morphology of propliopithecids was characteristic of advanced members of the catarrhine stem lineage.95

The parapithecid clade is represented at L-41 by the tiny Abuqatrania, whose bulbous molar cusps are very similar to those of ∼1 Ma younger Qatrania and the highly derived genera Apidium and Parapithecus, which are well-known from sites that are ∼3-4 Ma younger. Parapithecids were apparently largely frugivorous,92, 96 and abundant postcranial remains have been interpreted as indicating that these forms were agile leapers, though semicircular canal size suggests a “slow” mode of locomotion.91 At the time of deposition of the youngest primate-bearing sites in the Fayum succession (the ∼29-30 Ma Quarries I and M), Apidium and Parapithecus were very common in northern Africa, but until recently members of this clade were not known from younger beds. Parapithecids have now been found farther south, possibly in the latest Oligocene of Tanzania97 and definitively from the late early or late Oligocene Lokone site in Kenya, which has yielded a mandible and isolated upper teeth of a form called Lokonepithecus.98 It appears that parapithecids did indeed continue to diversify, at least in sub-Saharan Africa,86 but have never been found in the Miocene. Parapithecids were once thought to be closely related to cercopithecoid catarrhines due to dental and postcranial similarities that the two groups shared, but are now generally placed as stem anthropoids (Fig. 6). Perhaps competition with cercopithecoids played a role in their final extinction near the Oligocene-Miocene boundary.⁸⁶

EVOLUTION OF AFRO-ARABIAN PRIMATE COMMUNITIES IN THE PALEOGENE

It is now clear that primate communities in Afro-Arabia underwent several major phases of restructuring during the Paleogene. During the late Paleocene and Eocene, individual colonists from northern continents would make it into Afro-Arabia via chance dispersals across the Tethys Sea, go on to adapt to local environments and diversify, and, millions of years later, be faced with competition from another immigrant lineage. Later, environmental change associated with global cooling appears to have played an important role, as it probably constricted the latitudinal range of thermophilic crown strepsirrhine clades (e.g., galagids), and, in some cases, might have driven stem strepsirrhine lineages into extinction (adapids, plesiopithecids, and djebelemurines).86 Anthropoids persisted into the Oligocene in northern Africa in what might have been relatively marginal, unpredictable environments.

One interesting way of envisioning adaptive trajectories within these communities is to analyze a large morphological character matrix with a phenetic approach, such as a multivariate analysis of pairwise distances between taxa. In this case, the distances between species provide an approximation of the overall similarity or difference of species based on the character states that they have been assigned, without taking into account the polarity of traits or the historical (phylogenetic) basis for how they came to exhibit such a character complex. Species that share many character states are separated by short pairwise distances, while those that share few character states are separated by large pairwise distances. Analysis of one such matrix of morphological features from the dentition, cranium, and postcranium34 (Fig. 9) reveals that known living and extinct primates occupy a fairly restricted character space, and that one zone of that space was repeatedly occupied by distantly related lineages that converged on an “anthropoid-like” pattern, either on northern continents or in Afro-Arabia. On northern continents, this zone was occupied by several adapiforms that have, at one time or another, been identified as candidate anthropoids (for instance, adapines, caenopithecines, Darwinius, and sivaladapids). In Afro-Arabia, this zone was first occupied by azibiids – one of which, Algeripithecus, was initially identified as an anthropoid43 – and later by the caenopithecine clade that would give rise to Afradapis and Aframonius, the latter of which was also initially identified as a possible anthropoid relative.66Plesiopithecus occupies this zone, and it, too, was first identified as an anthropoid.65 Finally, the enigmatic Nosmips, a taxon that shows both anthropoid-like and strepsirrhine-like features, is also placed in this general zone. Though the oldest eosimiids fall close to tarsiiforms on this plot, the younger (and, in my opinion, more convincingly anthropoid-like) Bahinia later came to occupy the same “anthropoid-like” character space. Could it be that a form like Bahinia dispersed into Afro-Arabia in the middle Eocene, faced competition with adaptively similar, but distantly related, “endemic” forms such as caenopithecines and plesiopithecids, and that this competition played a critical role in determining the adaptive trajectory of later anthropoid evolution? The oldest Fayum anthropoid, Biretia, falls closest to this zone, and later – via independent phylogenetic pathways – the members of sequentially younger anthropoid communities documented in the Fayum area diverge along a seemingly predictable trajectory, taking this generalized pattern to an extreme by convergently evolving craniodental features that were long thought to define Anthropoidea as a whole. The Afro-Arabian record is, unfortunately, simply not yet complete enough to allow for large-scale, temporally constrained tests of hypotheses related to competitive exclusion or character displacement, but these are certainly intriguing possibilities, and are ideas that could not have even been proposed without the wealth of surprising fossil evidence that has accumulated over the course of the last two decades.