Journal of Human Evolution 49 (2005) 468e481 Taxonomic status of purported primate frontal bones from the Eocene Pondaung Formation of Myanmar K. Christopher Beard a,*, Jean-Jacques Jaeger b, Yaowalak Chaimanee c, James B. Rossie a,d, Aung Naing Soe e, Soe Thura Tun f, Laurent Marivaux b, Bernard Marandat b a Section of Vertebrate Paleontology, Carnegie Museum of Natural History, 4400 Forbes Avenue, Pittsburgh, PA 15213, USA b Laboratoire de Paléontologie, Institut des Sciences de l’Evolution, C.C. 064, Universite´ Montpellier II, Place Euge`ne Bataillon, F-34095 Montpellier cedex 05, France c Department of Mineral Resources, Paleontological Section, Geological Survey Division, Rama VI Road, Bangkok 10400, Thailand d Section of Mammals, Carnegie Museum of Natural History, 5800 Baum Boulevard, Pittsburgh, PA 15206, USA e Department of Geology, Pa-an University, Pa-an, Myanmar f Department of Geology, University of Yangon, Yangon 11041, Myanmar Received 29 June 2004; accepted 30 May 2005 Abstract Two isolated cranial fragments from the late middle Eocene Pondaung Formation of central Myanmar have previously been interpreted as frontal bones of the amphipithecid primate Amphipithecus mogaungensis. Aside from a few maxillary fragments, these specimens provide the only potential source of information currently available regarding the cranial anatomy of Amphipithecidae. Were this taxonomic attribution correct, these specimens would indicate that amphipithecids retained numerous primitive skull features, including the absence of a postorbital septum, the retention of a voluminous olfactory chamber, and strong separation between the forebrain and the orbital fossa. However, several anatomical details observable on these specimens are incompatible with their attribution to any primate and strongly suggest that they cannot be ascribed to Mammalia. Particularly problematic in this regard are the extreme thickness of the dermal bone, the odd structure of the alleged ‘‘frontal trigon,’’ and the mediolateral orientation and uniquely robust construction of the descending process of the frontal bone (which partially segregates the orbital and temporal fossae). Because these isolated elements can no longer be attributed to Amphipithecus, the anatomical, * Corresponding author. Tel.: C1 412 622 5782; fax: C1 412 622 8837. E-mail address: beardc@carnegiemnh.org (K.C. Beard). 0047-2484/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.jhevol.2005.05.008 K.C. Beard et al. / Journal of Human Evolution 49 (2005) 468e481 469 phylogenetic, and behavioral inferences regarding amphipithecid paleobiology that have been drawn from these specimens can no longer be sustained. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: Amphipithecidae; Eocene; Pondaung Formation; Myanmar; Anthropoid origins Introduction Few fossil primates have engendered greater longstanding phylogenetic controversy than Pondaungia cotteri and Amphipithecus mogaungensis from the late middle Eocene Pondaung Formation of Myanmar (Pilgrim, 1927; Colbert, 1937). Both were initially described as primitive anthropoids, a view that has remained prevalent ever since (e.g., Simons, 1971; Ba Maw et al., 1979; Ciochon et al., 1985; Jaeger et al., 1998, 2004; Chaimanee et al., 2000; Ducrocq, 2001; Beard, 2002; Marivaux et al., 2003; Takai and Shigehara, 2004). Alternative ideas regarding the phylogenetic affinities of Pondaungia and Amphipithecus have ranged from von Koenigswald’s (1965) early assertion that Pondaungia is a condylarth to repeated claims that Pondaungia and/or Amphipithecus are uniquely specialized adapiform primates bearing only convergent similarities with early anthropoids (Szalay, 1970, 1972; Ciochon and Holroyd, 1994; Ciochon et al., 2001; Gunnell et al., 2002; Ciochon and Gunnell, 2002a,b, 2004). Recent phylogenetic analyses of early anthropoid relationships underscore this systematic uncertainty (Kay et al., 2004b). At least in part, this lack of consensus reflects the nature of the fossil material that has been attributed to Pondaungia and Amphipithecus. Isolated cranial and postcranial elements from the Pondaung Formation have been allocated to both taxa (Ciochon et al., 2001; Takai et al., 2003; Marivaux et al., 2003). With the exception of the talus described by Marivaux et al. (2003), these isolated finds emit a phylogenetic signal seemingly at odds with the evidence yielded by amphipithecid jaws and teeth. Possible explanations for this apparent character conflict include homoplasy, mosaic evolution, and the incorrect allocation of isolated and fragmentary fossils to taxa that were originally established solely on the basis of dental remains. Here, we show that the latter factor explains the difficulty in reconciling the anthropoid-like dentitions of Pondaungia and Amphipithecus with the very primitive craniofacial anatomy implied by two isolated cranial fragments that have recently been attributed to Amphipithecus. Previous work Both cranial fragments were collected by scientists working under the auspices of the ‘‘Myanmare Japan Pondaung Expedition Team’’ (Than Tun, 2000; Takai et al., 2000, 2003) (Figs. 1e2). The first specimen, National Museum of Myanmar Primate 19 (abbreviated hereafter as NMMP 19), was recovered in November 1999 at the Paukkaung kyitchaung 2 locality in the vicinity of Bahin village. In the field, NMMP 19 was found near a right maxillary fragment of Amphipithecus mogaungensis (NMMP 18), and the two specimens have usually been regarded as representing the same individual. Despite its fragmentary condition and ambiguous allocation to Amphipithecus, NMMP 19 has attained a surprising degree of prominence during the past five years. The specimen was illustrated, but neither described nor discussed in any detail, by Than Tun (2000) and Takai et al. (2000). Takai et al. (2000: Fig. 9) figured NMMP 19 alongside the NMMP 18 maxilla and referred to both elements as ‘‘the newest specimen of A. mogaungensis,’’ thereby implying that they pertained to the same individual. Slightly more than a year later, a second specimen (NMMP 27) bearing morphology comparable to NMMP 19 was found at the Paya Ama kyitchaung 1 locality in the vicinity of Sinzwe village. Although neither NMMP 19 nor NMMP 27 had yet been studied or described by members of the scientific team that recovered them, the specimens rapidly attracted the attention of other workers seeking to address the phylogenetic 470 K.C. Beard et al. / Journal of Human Evolution 49 (2005) 468e481 position of amphipithecid primates. Gunnell et al. (2002) published additional photographs of NMMP 19, announced the existence of NMMP 27, and made some brief anatomical observations pertaining to both specimens. In contrast to Than Tun (2000) and Takai et al. (2000), who identified NMMP 19 as a parietal fragment, Gunnell et al. (2002: 362) recognized both NMMP 19 and NMMP 27 as frontal bones that ‘‘potentially may represent Amphipithecus.’’ Despite their hesitation regarding the systematic allocation of NMMP 19 and NMMP 27, Gunnell et al. (2002) felt confident enough to score two important charactersdpresence or absence of postorbital closure and presence or absence of a fused metopic suturedin the cells pertaining to Amphipithecus in the character-taxon matrix they constructed for phylogenetic analysis (see Gunnell et al., 2002: Tables 5e6). In contrast to the derived character states that occur in undoubted anthropoids, Gunnell et al. (2002) determined that both postorbital closure and a fused metopic suture were absent in NMMP 19 and NMMP 27. They further indicated that these specimens exhibit a high degree of postorbital constriction of the braincase (similar to that found in Adapis). Based partly on the presence of these notably primitive cranial features in Amphipithecus, the phylogenetic analysis undertaken by Gunnell et al. (2002) yielded a clade containing large-bodied amphipithecids and the North American notharctid adapiform Hesperolemur. Having thus corroborated their earlier suspicions regarding the notharctid affinities of amphipithecids (Ciochon and Holroyd, 1994; Ciochon et al., 2001), these authors proceeded to hypothesize that the derived dental features shared by amphipithecids and undoubted anthropoids were merely convergent ecological adaptations for feeding on ‘‘hard objects and tough-skinned fruits’’ (Gunnell et al., 2002: 369; also see Ciochon and Gunnell, 2002a,b, 2004). In a study focusing on the maxillary morphology of Pondaungia cotteri, Shigehara et al. (2002) explicitly declared that NMMP 19 (which they referred to as NMMP-KU 0229) is a frontal bone pertaining to the same individual as the Amphipithecus maxillary fragment, NMMP 18 (which they referred to as NMMP-KU 0228). In agreement with Gunnell et al. (2002), they further noted: ‘‘On the frontal bone, the postorbital process was shaped like a bar and does not contribute a postorbital flange. These features...indicate that this taxon [Amphipithecus] did not exhibit postorbital closure’’ (Shigehara et al., 2002: 160). Without necessarily endorsing adapiform affinities for amphipithecids, they also emphasized that ‘‘this pattern [i.e., the absence of postorbital closure in Amphipithecus] presents problems for anyone who wishes to conclude that amphipithecids are stem anthropoids’’ (Shigehara et al., 2002: 161). Takai et al. (2003) reiterated the direct field association between NMMP 19 and NMMP 18. Based on new measurements, they concluded that NMMP 19 shows a lesser degree of postorbital constriction than Gunnell et al. (2002) had surmised. Takai et al. (2003) also observed polymorphism regarding fusion of the metopic suture in NMMP 19 (in which the suture appears to be fused) and NMMP 27 (in which the suture appears to be unfused), which they believed might be attributable to ontogenetic growth. On the basis of an endocranial cast of NMMP 19, Takai et al. (2003) reconstructed Amphipithecus as having relatively voluminous olfactory bulbs that project rostrally beyond the level of the frontal lobes of the brain. Morphologically, the olfactory bulbs of Amphipithecus were said to resemble the ‘‘bilobed’’ condition found in omomyids and microchoerids, as opposed to the more divergent condition characteristic of Eocene adapiforms. Kay et al. (2004a) accepted earlier determinations that Amphipithecus lacks postorbital closure, a condition that they regarded as being functionally linked with other features enhancing visual acuity in anthropoids, such as the loss of a tapetum lucidum and the presence of a retinal fovea. Accordingly, Kay et al. (2004a: 7) concluded that ‘‘the absence of postorbital closure in Amphipithecus suggests this animal did not possess the acute vision present in modern anthropoids.’’ Kay et al. (2004a) also interpreted the relatively large olfactory bulbs of Amphipithecus (described by Takai et al., 2003) as indicating that amphipithecids retained a strepsirrhine-like reliance on olfaction. In their recent analysis of anthropoid relationships, Kay et al. (2004b) emphasized the K.C. Beard et al. / Journal of Human Evolution 49 (2005) 468e481 phylogenetic instability of amphipithecids. Working under the assumption that Amphipithecus lacks postorbital closure, Kay et al. (2004b) found that amphipithecids group with adapiforms under most, but not all, of the character partitioning, ordering, and weighting schemes they investigated. Nevertheless, certain character assumption sets utilized by Kay et al. (2004b) yield phylogenies that recognize amphipithecids as anthropoids, despite their supposition that amphipithecids lack postorbital closure. Likewise, the phylogenetic analysis conducted by Seiffert et al. (2004) recovered amphipithecids as basal anthropoids, despite its assumption that amphipithecids lack postorbital closure. Critical reappraisal of NMMP 19 and NMMP 27 Given that skull features such as postorbital closure are widely regarded to be crucial for reconstructing the relationships of basal anthropoids, and given that NMMP 19 and NMMP 27 are currently the only specimens that potentially illuminate this aspect of amphipithecid anatomy, the morphology and taxonomic status of these pivotal specimens require further scrutiny. We observed several anatomical features in NMMP 19 and NMMP 27 that are inconsistent with their identification as frontal bones of primates. These traits are highlighted in the following sections. For convenience, we describe the specimens as if they were frontal bones (of some unknown taxon). In fact, we are not confident that even this anatomical (rather than taxonomic) attribution is correct. Frontal trigon In some living and extinct primates, the temporal crests converge posteriorly to produce a compound sagittal crest. In such forms, the area circumscribed by these crests and the superior orbital margins is known as the frontal trigon (Tobias, 1967). Among Eocene primates, certain adapiforms such as Adapis and Notharctus exhibit particularly pronounced frontal trigons (e.g., 471 Gazin, 1958; Gingerich and Martin, 1981). The putative frontal bones from the Pondaung Formation have been described as having frontal trigons as well, but several anatomical details militate against this assessment. As Takai et al. (2003) noted, the ‘‘frontal trigon’’ of both Pondaung specimens is convex (Figs. 1e3), which contrasts with the concave topography of the frontal trigon that typically occurs in primates (including such diverse fossils as Adapis, Notharctus, Aegyptopithecus, and Afropithecus). With the exception of Chiropotes, a convex frontal trigon is only found in those primates in which a paranasal sinus inflates the underlying bone (e.g., Archaeolemur, Smilodectes). However, according to radiographs, the remarkably thick squamous portions of NMMP 19 and NMMP 27 contain no pneumatized spaces (Takai et al., 2003). In Chiropotes, the convexity of the frontal trigon is due to the curvature of the large anterior lobes of the brain immediately beneath it. In contrast, the endocranial surface of the Pondaung fossils is hardly even concave in the rostrocaudal dimension. The slight curvature that does mark the endocranial surface of these specimens fails to influence the external topography of the skull due to the extraordinary thickness of the intervening bone. The frontal bones of some Eocene omomyids (e.g., Shoshonius, Tetonius, Hemiacodon) exhibit a mildly convex surface topography, but these forms lack a sagittal crest and their temporal crests do not converge until they approach the caudal end of the parietals. These cranial features strain the definition of a frontal trigon while yielding a total morphological pattern that diverges markedly from that shown by the Pondaung specimens. The temporal crests converge farther rostrally in the Eocene microchoerid Necrolemur, forming a modest sagittal crest. However, the frontal bones of Necrolemur (and omomyids) differ conspicuously from the Pondaung specimens in having the temporal crests and orbital rims raised above the level of the trigon. In addition to the important morphological differences noted already, two conditions observable in NMMP 19 and NMMP 27 conflict with the very definition of a frontal trigon. As described above, a frontal trigon is produced by the posterior 472 K.C. Beard et al. / Journal of Human Evolution 49 (2005) 468e481 Fig. 1. Cranial fragments NMMP 19 (A, B) and NMMP 27 (C, D) from the Pondaung Formation, previously interpreted as frontal bones of Amphipithecus mogaungensis. Specimens are depicted in dorsal (A, C) and ventral or endocranial (B, D) views. Scale bar equals 1 cm. convergence of the temporal crests to form a midline sagittal crest. This condition is by no means common in primates, and it usually arises as a result of hypertrophy of the temporalis musculature (Benefit and McCrossin, 1995). In both Pondaung specimens, the ‘‘sagittal crest’’ projects rostrally beyond the point where the temporal crests converge near the midline. In NMMP 27 the ‘‘sagittal crest’’ extends at least 18 mm rostrally beyond the location where the temporal crests converge, spanning the entire length of the frontal bone as it is preserved. In this specimen, the rostral part of the ‘‘sagittal crest’’ forms a ridge that is split by the metopic suture. In contrast, in NMMP 19 the ‘‘sagittal crest’’ divides into multiple ridges that fan out asymmetrically to give the frontal trigon a striated appearance. The frontal trigon of primates is never adorned with ridges, even when a metopic suture is present. To our knowledge, the only placental mammal in which a midline crest K.C. Beard et al. / Journal of Human Evolution 49 (2005) 468e481 473 Fig. 2. Explanatory drawing depicting the same specimens shown in the same orientations as in Fig. 1. Anatomical abbreviations are as follows: dpf, descending process of the frontal bone; fms, fused metopic suture; ft, frontal trigon; oc, olfactory chamber; orb, orbit; pb, broken root of postorbital bar; sc, sagittal crest; sss, superior sagittal sinus; tc, temporal crest; ums, unfused metopic suture. Scale bar equals 1 cm. occurs on the rostral part of the frontal bone is the yellow armadillo, in which it serves as the site for attachment of the orbitoauricularis muscle (Wible and Gaudin, 2004). In addition to the problematic rostral prolongation of the ‘‘sagittal crest’’ in the Pondaung specimens, the bilateral structures that have been interpreted as temporal crests also conflict with 474 K.C. Beard et al. / Journal of Human Evolution 49 (2005) 468e481 Fig. 3. Cranial fragment NMMP 19 in right lateral (A) and anterior (B) views. Anatomical abbreviations are the same as in Fig. 2, with the following additions: fbm, free bony margin of descending process of frontal bone; sif, superior intraorbital foramen. Scale bar equals 1 cm. anatomical patterns that are otherwise ubiquitous among primates. In all fossil and extant primates, whenever the temporalis musculature becomes sufficiently enlarged to yield confluence of the temporal crests (and particularly when this convergence occurs a short distance from the supraorbital costae, as is the case in Adapis, Notharctus, and the Pondaung fossils), the sagittal and temporal crests are distinct, raised features. In contrast, the ‘‘temporal crests’’ on the Pondaung fossils are low and rounded, and the ‘‘sagittal crest’’ is little more than a keel (Fig. 2A,C). The ‘‘temporal crests’’ are almost indiscernible on NMMP 19 near the midline, which is precisely where they should be most pronounced, and at no point are they raised topographically above the surface of the supposed frontal trigon. Even more troubling is the fact that, on both Pondaung specimens, the ‘‘temporal crests’’ do not appear to be bilaterally symmetrical. On NMMP 19, the left ‘‘temporal crest’’ reaches the midline rostral to the point where its right counterpart joins the ‘‘sagittal crest.’’ The opposite pattern holds in NMMP 27. In neither case does postmortem deformation account for this asymmetry. In summary, the ‘‘sagittal crest’’ in NMMP 19 and NMMP 27 does not appear to result from K.C. Beard et al. / Journal of Human Evolution 49 (2005) 468e481 the confluence of the temporal lines, and it is therefore not a true sagittal crest. The form of the temporal and sagittal ‘‘crests,’’ as well as the external topography of the adjacent bone, are inconsistent with significant hypertrophy of the temporalis musculature, which therefore cannot be responsible for the rapid angle of convergence of the ‘‘temporal crests.’’ Although various aspects of the structures in question are vaguely reminiscent of the primate frontal trigon, these similarities are superficial and certainly nonhomologous. Descending process of the frontal The descending process of the frontal bone, located at the junction of the orbital, temporal, and anterior cranial fossae, consists of a massive pillar of solid bone that is roughly rectangular in crosssection in both Pondaung specimens (Figs. 2e4). As noted by Gunnell et al. (2002) and Takai et al. (2003), the descending process of the frontal in NMMP 19 preserves a natural bony margin posteroinferior to the root of the supposed postorbital bar on the right side (Figs. 3e4). No equivalent bony margin exists in anthropoids and tarsiids because this is where the postorbital septum separates the orbital fossa from the adjacent temporal fossa in these taxa (Fig. 5A). Accordingly, several authors have emphasized that the free bony margin on the descending process of the frontal in NMMP 19 indicates that a postorbital septum did not occur in this specimen (Gunnell et al., 2002; Shigehara et al., 2002; Takai et al., 2003; Kay et al., 2004a,b; Ciochon and Gunnell, 2004; Takai and Shigehara, 2004). However, other aspects of the descending process of the frontal bone depart radically from the condition seen in other primates, as well as in mammals generally. In NMMP 19 the descending process of the frontal bone forms a mediolaterally extensive, coronally oriented bony wedge that would have partly segregated the orbit from the temporal fossa (Figs. 3e4). In living and fossil primates that lack a postorbital septum, the descending process of the frontal consists of relatively thin dermal bone that contributes only to the sidewall of the braincase 475 (cf. Daubentonia as depicted in Fig. 4; Lemur in Fig. 5B). The descending process of the frontal bone in Eocene adapiforms such as Adapis does not differ significantly from those of modern primates without a postorbital septum. In contrast, in NMMP 19 the descending process of the frontal flares laterally to form a coronally oriented partial posterior wall to the orbit, even in the region where the free bony margin occurs. Inferior to this, the bone is poorly preserved, but its broken surface indicates that the posterior orbital wall would have extended even farther laterally here. Although the free bony margin on the descending process of the frontal demonstrates that a conventional postorbital septum was absent in NMMP 19, the mediolateral orientation of the extensive bony wedge behind the orbit more closely approximates the condition in anthropoids and tarsiids than it does the primitive arrangement in which no bony barrier segregates the orbital and temporal fossae. Indeed, the presence of this partial postorbital wall (if that is what this structure is) distinguishes NMMP 19 from all vertebrates aside from tarsiids and anthropoids (Cartmill, 1980; Ross, 1994). In addition to the free bony margin noted earlier, an important distinction between the descending process of the frontal in NMMP 19 and the postorbital septum of anthropoids and tarsiids is the extraordinary robusticity of this structure in NMMP 19, as opposed to the paperthin construction of the postorbital septum in anthropoids and tarsiids (Figs. 4, 5A). At its inferior (broken) margin, the descending process of the frontal in NMMP 19 is 4.35 mm thick (measured rostrocaudally). Cartmill (1980: 245) noted that, in living haplorhines, ‘‘the postorbital septum is formed by bony processes that grow backward from the posteromedial edge of the [postorbital] bar (or forward from the braincase) along the cone-shaped sheet of free periorbita that demarcates the orbital contents from the anterior temporalis.’’ This ontogenetic pathway is difficult to reconcile with the extreme robusticity of the descending process of the frontal in NMMP 19. This structure in NMMP 19 further differs from the postorbital septum of living and fossil anthropoids in being composed of the frontal, rather than 476 K.C. Beard et al. / Journal of Human Evolution 49 (2005) 468e481 Fig. 4. Cranial fragment NMMP 19 (A) compared with skulls of the anthropoid Saimiri sciureus (BeC) and the strepsirrhine Daubentonia madagascariensis (DeE), all in oblique right lateral view. Note the unique structure of the descending process of the frontal bone in NMMP 19, which differs from that of primates in being much more robust and in flaring laterally rather than simply contributing to the sidewall of the braincase. The volumetric outline in (A) is meant to clarify the extraordinary robusticity of this element in NMMP 19; there is no reason to believe that the structure would have extended as far laterally as the outline indicates. The zygomatic arch and parts of the postorbital bar/septum have been removed in (C) and (E) to facilitate comparisons with the fragmentary Pondaung specimen. Anatomical abbreviations are the same as in Figs. 2e3. Specimens are not depicted at the same scale. the zygomatic bone (Simons and Rasmussen, 1989). In agreement with previous workers, we regard these differences as sufficient to rule out any hypothesis of homology between the descending process of the frontal bone in NMMP 19 and the postorbital septum of anthropoids and/or tarsiids. At the same time, we emphasize that nothing approaching the condition in NMMP 19 occurs in K.C. Beard et al. / Journal of Human Evolution 49 (2005) 468e481 477 Fig. 5. High-resolution computed tomography scans (lower image; oriented so that rostral is toward the top) through the orbital regions (orb) of the anthropoid Callicebus moloch (A) and the strepsirrhine Eulemur fulvus (B). The white transverse lines shown in lateral view (upper image) indicate the approximate plane of each scan. Note that, in primates, the descending process of the frontal bone (dpf) consists of relatively thin dermal bone that contributes to the sidewall of the braincase, which contrasts with the condition in the Pondaung fossils (see Fig. 4). In both taxa, the frontal sinus (fs) is labeled to avoid potential confusion with the nearby olfactory chamber. Scale bars apply to upper images only. other living or fossil primates (or in other mammals, for that matter). Postorbital constriction In order to obtain a relative measure of postorbital constriction, Gunnell et al. (2002) compared the postorbital breadth of NMMP 19 (16.97 mm) to the length of M1 in NMMP 18 (5.84 mm), yielding a ratio of 2.91. This value is smaller than the same ratio in Adapis parisiensis, a species that is notable for exhibiting one of the most severely constricted postorbital regions of any primate. However, we agree with Takai et al. (2003) that the points chosen by Gunnell et al. (2002) to estimate postorbital breadth in NMMP 19 were inappropriate. Using the corrected measurement (21.38 mm) from Takai et al. (2003), the ratio of postorbital breadth to M1 length becomes 3.66, which falls between the equivalent values for Hesperolemur and Adapis (Gunnell et al., 2002). However, if NMMP 19 does not pertain to Amphipithecus, as we believe, then its degree of postorbital constriction can only be estimated with respect to other dimensions found on the same specimen. One comparative metric that is useful in this regard is interorbital breadth, as it is measured between the preserved natural bone surfaces at the superomedial corners of both orbits (minimum interorbital breadth cannot be measured in this specimen). In NMMP 19, interorbital breadth is narrower than postorbital breadth, as is typical of primates with relatively little postorbital constriction (e.g., Smilodectes, Galago, Eulemur, Loris). The opposite condition describes primates having severe postorbital constriction, such as Leptadapis and Adapis (Lanèque, 1993). On this basis, NMMP 19 appears to exhibit relatively minor postorbital constriction, but the endocranial morphology of the specimen complicates this assessment. Regardless of whether the frontal squama of NMMP 19 was steeply inclined (cf. Takai et al., 2003: Fig. 2a) or relatively horizontal in life, the anterior cranial fossa would not have projected over the orbits to any appreciable extent. The forebrain of mammals always overlaps the orbits 478 K.C. Beard et al. / Journal of Human Evolution 49 (2005) 468e481 to some degree simply because the optic foramen lies caudal to the anterior cranial fossa, no matter how small the latter may be (de Beer, 1937). If the frontal squama of NMMP 19 was oriented in life so that it sloped down rostrally to meet the nasals, a small part of the orbital roof posterior to the broken descending process of the frontal may have extended beneath the forebrain in this specimen. However, the position of the broken roots of the postorbital bars (in a coronal plane that bisects the olfactory bulbs) points to an extreme separation of the forebrain and orbits. Such an arrangement is approximated only in primates with acute postorbital constriction. Yet, as discussed previously, the morphology of the orbitotemporal fossa and the ratio of interorbital to postorbital breadth are inconsistent with such extreme postorbital constriction in NMMP 19. It therefore appears that NMMP 19 displays greater anteroposterior separation between the orbits and the forebrain than occurs in any known primate, while lacking the strong postorbital constriction characteristic of all primates in which the forebrain and orbits are so disposed. The space that accounts for this discrepancy is largely occupied by the uniquely robust descending process of the frontal, which impinges on the region that would otherwise be occupied by the temporalis muscles. As with the weak temporal and sagittal ‘‘crests,’’ this extraordinary morphology indicates that the temporalis musculature was not particularly hypertrophied. We are left with the unresolved dilemma of why the orbits and forebrain of NMMP 19 should be so widely separated if hypertrophy of the temporalis musculature and postorbital constriction are not responsible for this unusual condition. Once again, NMMP 19 presents a combination of features that not only differs from that found in all other primates, but which also makes little structural or functional sense. Olfactory chambers The olfactory chambers preserved in NMMP 19 differ in several respects from those that we have been able to observe in primates. The olfactory chambers of living strepsirrhines can easily be viewed through the foramen magnum. Our comparisons with anthropoids are based on computed tomography scans of living platyrrhines (Rossie, 2003). The presence of superior intraorbital foramina (SIF), which are venous foramina in the dorsomedial orbital walls that open into the endocranial surface of the frontal and join with the superior sagittal venous sinus, is a condition shared by NMMP 19 and most strepsirrhines, tarsiers, and some anthropoids (Takai et al., 2003). In lemuroids, the venous foramina open into the posterior end of the olfactory fossae, somewhat as in NMMP 19, but the fossae themselves often end abruptly at a point immediately anterior to this, where the cribriform plate meets the roof of the chamber. As a result, in lemuroids, the olfactory fossae on the frontal bone are very small, unlike the condition in NMMP 19. In some specimens, the olfactory fossae extend farther anteriorly, but are much narrower than in NMMP 19, the remaining interorbital area being filled by the large frontal sinus (Fig. 5B). In lorisoids, the olfactory fossae and the mid-sagittal ridge that divides them are smaller than in NMMP 19, even in large species. When these features are discernible, there is also a coronally oriented ridge that partially separates the olfactory chambers from the frontal lobes of the brain. The venous foramina open into this ridge, behind the olfactory bulbs, but no further impressions of these veins or the superior sagittal venous sinus are discernible. In NMMP 19, the veins open into the tectum of the olfactory chambers above the olfactory bulbs before joining a deep trough for the superior sagittal sinus that cuts through the area where the post-olfactory ridge would be. Such a deep impression for the superior sagittal sinus was not observed in any of the primates examined here. Takai et al. (2003) noted this difference, but apparently were unconcerned by the extraordinarily strong development of this structure in NMMP 19. As expected, anthropoids show minimal development of the olfactory fossae because the olfactory bulbs are more or less covered by the frontal lobes. Even in Aotus, which has relatively large olfactory bulbs that protrude anterior to the 479 K.C. Beard et al. / Journal of Human Evolution 49 (2005) 468e481 frontal lobes (Stephan et al., 1981), the fossae are shallow. However, the impression for the superior sagittal sinus in Aotus is rather distinct, albeit shallow and rounded, and the post-olfactory ridge is not evident (Rossie, unpublished data). In total, the olfactory chambers of NMMP 19 show a combination of features that are not found together in any one primate, but only the sharpness of the olfactory fossae and the depth of the groove for the superior sagittal sinus are unlike any of the primates examined. The full taxonomic distribution of the orbital venous foramina discussed above is unclear, in part because of terminological problems. The superior intraorbital foramina are sometimes called ‘‘supraorbital foramina’’ in accounts of nonhuman primates (e.g., Berry, 1974), but these structures should not be confused with the supraorbital foramen of human anatomy, which contains the supraorbital nerve, artery, and vein. The true homologue of the SIF in humans is the foramen caecum, which enters the olfactory fossa through the anterior end of the crista galli (Simons, 1959). As a result of this potential confusion, one cannot discern the presence of a true SIF from any account that does not describe the contents or course of the foramen/canal (cf. Novacek, 1986). Fortunately, Le Gros Clark (1926) did distinguish between the two foramina, and reported the presence of the SIF in Ptilocercus, demonstrating that the structure is not limited to primates. Other oddities Beyond the topics elaborated above, several other unusual features of NMMP 19 and NMMP 27 merit brief mention here. In NMMP 19, the anterior borders of the parietals are V-shaped and incise the frontal bone, thereby creating a Wshaped coronal suture. The right medial orbital wall in NMMP 19 slopes inferolaterally, thereby reducing the already small volume available to house the orbital contents (Fig. 3B). Finally, the interorbital portion of the frontal squama in NMMP 27 extends a considerable distance rostrally without showing any change in curvature, and without showing any sign of the nasal suture (Fig. 2C). Conclusions The combination of features discussed above conflicts with the attribution of NMMP 19 and NMMP 27 to any primate (Table 1). In an attempt to ascertain what these enigmatic fossil specimens might be, we have compared them with skulls of various fishes, turtles, sphenodontians, lizards, crocodilians, and mammals; so far without success. Despite our current inability to identify these specimens taxonomically and anatomically, we are confident that they do not pertain to the Amphipithecidae and we doubt that they are frontal bones at all. Thus, we recommend that they not be Table 1 Significant differences between NMMP 19 and NMMP 27 and the frontal bones of primates Feature NMMP 19/NMMP 27 Primates Convex Concave, unless inflated by paranasal sinuses or strong curvature of frontal lobes Sagittal crest Extends rostrally beyond confluence of temporal crests/lines Never extends rostrally beyond confluence of temporal crests/lines Temporal crests/lines Bilaterally asymmetrical and indistinct Bilaterally symmetrical and raised above floor of trigon Descending process of frontal Thick, robust bone; flares laterally to form partial postorbital wall Thin dermal bone; contributes to sidewall of braincase Anteroposterior separation between orbits and forebrain Strong, apparently associated with minimal postorbital constriction Much less, but always associated with strong postorbital constriction when present (e.g., Adapis) Deep Shallow or absent External topography of the frontal trigon Superior sagittal sinus 480 K.C. Beard et al. / Journal of Human Evolution 49 (2005) 468e481 incorporated into future discussions of amphipithecid anatomy, taxonomy, and phylogeny. remarks on previous versions of the manuscript, and the financial support of CNRS and National Science Foundation grants BCS 0100825 and 0309800. Discussion Throughout the history of research on fossil primates, fragmentary and isolated specimens have been misallocated to various extinct primate taxa. In some cases, these mistakes resulted in nothing more than inaccurate anatomical reconstructions. In others, the resulting chimeras have yielded phylogenetic, biogeographic, and paleobiological inferences that are now discredited (e.g., Gregory, 1927; Cartmill et al., 1981; White et al., 1983; White and Gebo, 2004). While we understand that scientific mistakes are inevitable to some extent, we are surprised at how rapidly and uncritically the paleoprimatological community has accepted NMMP 19 and NMMP 27 as amphipithecid frontal bones. Based wholly or partly on these specimens, various researchers have drawn what must now be regarded as highly questionable conclusions about the anatomy (e.g., presence or absence of a postorbital septum; degree of postorbital constriction; size and conformation of olfactory lobes of the brain), sensory capabilities (e.g., degree of visual acuity; degree of reliance on olfaction), and phylogenetic position (adapiform versus basal anthropoid) of amphipithecid primates (Gunnell et al., 2002; Shigehara et al., 2002; Ciochon and Gunnell, 2002a,b, 2004; Takai et al., 2003; Kay et al., 2004a,b; Takai and Shigehara, 2004). We hope that future work will lead to the recovery of undoubted amphipithecid cranial remains, and that such specimens will finally illuminate the anatomy, behavior, and phylogenetic position of these intriguing southeast Asian primates. Acknowledgements We thank Than Tun and his colleagues for facilitating our fieldwork in Myanmar, Aye Ko Aung for his able assistance in the field, Mark Klingler for graphics and artwork, Bill Kimbel and three anonymous reviewers for their constructive References Ba Maw, Ciochon, R.L., Savage, D.E., 1979. Late Eocene of Burma yields earliest anthropoid primate, Pondaungia cotteri. Nature 282, 65e67. Beard, K.C., 2002. Basal anthropoids. In: Hartwig, W.C. (Ed.), The Primate Fossil Record. Cambridge University Press, Cambridge, pp. 133e149. Benefit, B.R., McCrossin, M.L., 1995. Miocene hominoids and hominid origins. Annu. Rev. Anthropol. 24, 237e256. Berry, A.C., 1974. Non-metrical variation in the prosimian skull. In: Martin, R.D., Walker, A.C., Doyle, G.A. (Eds.), Prosimian Biology. Duckworth, London, pp. 641e653. Cartmill, M., 1980. Morphology, function, and evolution of the anthropoid postorbital septum. In: Ciochon, R.L., Chiarelli, A.B. (Eds.), Evolutionary Biology of the New World Monkeys and Continental Drift. Plenum Press, New York, pp. 243e274. Cartmill, M., MacPhee, R.D.E., Simons, E.L., 1981. Anatomy of the temporal bone in early anthropoids, with remarks on the problem of anthropoid origins. Am. J. Phys. Anthropol. 56, 3e21. Chaimanee, Y., Tin Thein, Ducrocq, S., Aung Naing Soe, Benammi, M., Than Tun, Thit Lwin, San Wai, Jaeger, J.-J., 2000. A lower jaw of Pondaungia cotteri from the late middle Eocene Pondaung Formation (Myanmar) confirms its anthropoid status. Proc. Natl. Acad. Sci. 97, 4102e4105. Ciochon, R.L., Gunnell, G.F., 2002a. Eocene primates from Myanmar: historical perspectives on the origin of Anthropoidea. Evol. Anthropol. 11, 156e168. Ciochon, R.L., Gunnell, G.F., 2002b. Chronology of primate discoveries in Myanmar: influences on the anthropoid origins debate. Yearb. Phys. Anthropol. 45, 2e35. Ciochon, R.L., Gunnell, G.F., 2004. Eocene large-bodied primates of Myanmar and Thailand: morphological considerations and phylogenetic affinities. In: Ross, C.F., Kay, R.F. (Eds.), Anthropoid Origins: New Visions. Kluwer/Plenum, New York, pp. 249e282. Ciochon, R.L., Holroyd, P.A., 1994. The Asian origin of Anthropoidea revisited. In: Fleagle, J.G., Kay, R.F. (Eds.), Anthropoid Origins. Plenum Press, New York, pp. 143e162. Ciochon, R.L., Savage, D.E., Thaw Tint, Ba Maw, 1985. Anthropoid origins in Asia? New discovery of Amphipithecus from the Eocene of Burma. Science 229, 756e759. Ciochon, R.L., Gingerich, P.D., Gunnell, G.F., Simons, E.L., 2001. Primate postcrania from the late middle Eocene of Myanmar. Proc. Natl. Acad. Sci. 98, 7672e7677. Colbert, E.H., 1937. A new primate from the upper Eocene Pondaung Formation of Burma. Am. Mus. Novit. 951, 1e18. K.C. Beard et al. / Journal of Human Evolution 49 (2005) 468e481 de Beer, G., 1937. The Development of the Vertebrate Skull. Oxford University Press, Oxford. Ducrocq, S., 2001. Palaeogene anthropoid primates from Africa and Asia: new phylogenetical evidences. C. R. Acad. Sci. Paris (Sciences de la Terre et des planètes) 332, 351e356. Gazin, C.L., 1958. A review of the middle and upper Eocene primates of North America. Smithson. Misc. Collns. 136, 1e112. Gingerich, P.D., Martin, R.D., 1981. Cranial morphology and adaptations in Eocene Adapidae, II: the Cambridge skull of Adapis parisiensis. Am. J. Phys. Anthropol. 56, 235e257. Gregory, W.K., 1927. Hesperopithecus apparently not an ape nor a man. Science 66, 579e581. Gunnell, G.F., Ciochon, R.L., Gingerich, P.D., Holroyd, P.A., 2002. New assessment of Pondaungia and Amphipithecus (Primates) from the late middle Eocene of Myanmar, with a comment on ‘Amphipithecidae.’ Contrib. Mus. Paleontol. Univ. Michigan 30, 337e372. Jaeger, J.-J., Aung Naing Soe, Aye Ko Aung, Benammi, M., Chaimanee, Y., Ducrocq, R.-M., Than Tun, Tin Thein, Ducrocq, S., 1998. New Myanmar middle Eocene anthropoids. An Asian origin for catarrhines? C. R. Acad. Sci., Paris (Sciences de la vie) 321, 953e959. Jaeger, J.-J., Chaimanee, Y., Tafforeau, P., Ducrocq, S., Aung Naing Soe, Marivaux, L., Sudre, J., Soe Thura Tun, Wanna Htoon, Marandat, B., 2004. Systematics and paleobiology of the anthropoid primate Pondaungia from the late middle Eocene of Myanmar. C. R. Palevol. 3, 243e255. Kay, R.F., Schmitt, D., Vinyard, C.J., Perry, J.M.G., Shigehara, N., Takai, M., Egi, N., 2004a. The paleobiology of Amphipithecidae, south Asian late Eocene primates. J. Hum. Evol. 46, 3e25. Kay, R.F., Williams, B.A., Ross, C.F., Takai, M., Shigehara, N., 2004b. Anthropoid origins: a phylogenetic analysis. In: Ross, C.F., Kay, R.F. (Eds.), Anthropoid Origins: New Visions. Kluwer/Plenum, New York, pp. 91e135. Lanèque, L., 1993. Variation of orbital features in adapine skulls. J. Hum. Evol. 25, 287e317. Le Gros Clark, W.E., 1926. On the anatomy of the pen-tailed tree shrew (Ptilocercus lowii). Proc. Zool. Soc. Lond. 1926, 1179e1309. Marivaux, L., Chaimanee, Y., Ducrocq, S., Marandat, B., Sudre, J., Aung Naing Soe, Soe Thura Tun, Wanna Htoon, Jaeger, J.-J., 2003. The anthropoid status of a primate from the late middle Eocene Pondaung Formation (central Myanmar): tarsal evidence. Proc. Natl. Acad. Sci. 100, 13173e13178. Novacek, M.J., 1986. The skull of leptictid insectivorans and the higher-level classification of eutherian mammals. Bull. Am. Mus. Nat. Hist. 183, 1e112. Pilgrim, G.E., 1927. A Sivapithecus palate and other primate fossils from India. Memoirs of the Geological Survey of India, Palaeontologia Indica, n.s. 14, 1e26. Ross, C., 1994. The craniofacial evidence for anthropoid and tarsier relationships. In: Fleagle, J.G., Kay, R.F. (Eds.), Anthropoid Origins. Plenum Press, New York, pp. 469e547. 481 Rossie, J.B., 2003. Ontogeny, homology, and phylogenetic significance of anthropoid paranasal sinuses. Ph.D. Dissertation, Yale University. Seiffert, E., Simons, E.L., Simons, C.V.M., 2004. Phylogenetic, biogeographic, and adaptive implications of new fossil evidence bearing on crown anthropoid origins and early stem catarrhine evolution. In: Ross, C.F., Kay, R.F. (Eds.), Anthropoid Origins: New Visions. Kluwer/Plenum, New York, pp. 157e181. Shigehara, N., Takai, M., Kay, R.F., Aye Ko Aung, Aung Naing Soe, Soe Thura Tun, Tsubamoto, T., Tin Thein, 2002. The upper dentition and face of Pondaungia cotteri from central Myanmar. J. Hum. Evol. 43, 143e166. Simons, E.L., 1959. An anthropoid frontal bone from the Fayum Oligocene of Egypt: the oldest skull fragment of a higher primate. Am. Mus. Novit. 1976, 1e16. Simons, E.L., 1971. Relationships of Amphipithecus and Oligopithecus. Nature 232, 489e491. Simons, E.L., Rasmussen, D.T., 1989. Cranial anatomy of Aegyptopithecus and Tarsius and the question of the tarsieranthropoidean clade. Am. J. Phys. Anthropol. 79, 1e23. Stephan, H., Frahm, H., Baron, G., 1981. New and revised data on volumes of brain structures in insectivores and primates. Folia Primatol. 35, 1e29. Szalay, F.S., 1970. Late Eocene Amphipithecus and the origins of catarrhine primates. Nature 227, 355e357. Szalay, F.S., 1972. Amphipithecus revisited. Nature 236, 179e180. Takai, M., Shigehara, N., 2004. The Pondaung primates, enigmatic ‘‘possible anthropoids’’ from the latest middle Eocene, central Myanmar. In: Ross, C.F., Kay, R.F. (Eds.), Anthropoid Origins: New Visions. Kluwer/Plenum, New York, pp. 283e321. Takai, M., Shigehara, N., Tsubamoto, T., Egi, N., Aye Ko Aung, Tin Thein, Aung Naing Soe, Soe Thura Tun, 2000. The latest middle Eocene primate fauna in Pondaung area, Myanmar. Asian Paleoprimatology 1, 7e28. Takai, M., Shigehara, N., Egi, N., Tsubamoto, T., 2003. Endocranial cast and morphology of the olfactory bulb of Amphipithecus mogaungensis (latest middle Eocene of Myanmar). Primates 44, 137e144. Than Tun, 2000. The achievment [sic] of MyanmareJapan Fossil Expedition Team. Asian Paleoprimatology 1, 1e6. Tobias, P.V., 1967. Olduvai Gorge, Volume 2. The Cranium of Australopithecus (Zinjanthropus) boisei. Cambridge University Press, Cambridge. von Koenigswald, G.H.R., 1965. Critical observations upon the so-called higher primates from the upper Eocene of Burma. Proc. Kon. Ned. Akad. Wet. 68, 165e167. White, J., Gebo, D.L., 2004. Unique proximal tibial morphology in strepsirrhine primates. Am. J. Primatol. 64, 293e308. White, T., Suwa, G., Richards, G., Watters, J.P., Barnes, L.G., 1983. ‘‘Hominoid clavicle’’ from Sahabi is actually a fragment of cetacean rib. Am. J. Phys. Anthropol. 61, 239e244. Wible, J.R., Gaudin, T.J., 2004. On the cranial osteology of the yellow armadillo Euphractus sexcinctus (Dasypodidae, Xenarthra, Placentalia). Ann. Carnegie Mus. 73, 117e196.