Phiomia major, Sanders & Kappelman & Rasmussen, 2004

Sanders, William J., Kappelman, John & Rasmussen, D. Tab, 2004, New large-bodied mammals from the late Oligocene site of Chilga, Ethiopia, Acta Palaeontologica Polonica 49 (3), pp. 365-392 : 377-384

Genus Phiomia Andrews and Beadnell, 1902 Phiomia major sp. nov.

Figs. 7–11 , Tables 2 and 3.

Holotype: CH 17−1, associated right I2 ( Fig. 8C ), partial left I2, right P2?, right partial M1?, right partial M3, and symphysis with right and left i2 ( Fig. 11A, B ).

Referred specimens: CH 3−53, right M1 ( Fig. 7B ), CH 4−2, left P4 ( Fig. 8A ); CH 4−3, right p3; CH 4−4, right p3 ( Fig. 9A ); CH 5−15, left p4; CH 9−1, left M2 ( Fig. 8B ); CH 9−2, right m3 ( Fig. 9C ); CH 9−23, right p4; CH 10−4, right molar fragment; CH 13−1a, b, associated right and left M3; CH 14−2, left? I2; CH 22−1, left M3 ( Fig. 7C ); CH 25−1, left m1 ( Fig. 9B ; antimere of CH 25−2); CH 25−2, right m1 (antimere of CH 25−1); CH 25−5, left P3 ( Fig. 7A ); right molar fragment; CH 27−6, right d4; CH 33−5, right partial molar; CH 33−V−7, right partial I2; CH 68−V−1, right partial p4, m1; CH 75−V−7, right partial m2; CH 75−V−9; right m1; CHS4−1, left m2 ( Fig. 10B ).

Etymology: From Latin major , greater, larger, in reference to dental and symphyseal dimensions exceeding those of other species of the genus. Type locality: Chilga 17, Upper Guang Section, Chilga region, northwest Ethiopia.Referred specimens are from other localities of the Upper Guang and Gahar Valley Sections.

Age and distribution: Late Oligocene, 28–27 Ma. Only known from the Chilga region.

Diagnosis.—Large species of Phiomia ; molar size range surpasses dimensions of Fayum palaeomastodonts and symphysis and incisors much longer than in Palaeomastodon and other species of Phiomia ( Table 3; Fig. 12 ); further distinguished from Palaeomastodon by absence of posttrite cristae (“zygodont crests”), presence of a central conelet in the posterior loph of P4, and by full trilophodonty of molars, including M3/m3. Differentiated from Miocene elephantoid taxa (for example, Hemimastodon , gomphotheres, mammutids) by smaller size of molars ( Fig. 12 ), lack of features such as zygodont crests, crescentoids, and pretrite anterior and posterior accessory central conules throughout the crown, and trilophodont m3 with a diminutive posterior cingulid. Distinguished from Moeritherium by larger size, development of incisors into tusks, and trilophodont intermediate molars.

Description.—Most proboscidean teeth from Chilga closely resemble those of primitive elephantiform species of Phiomia from Fayum. Individual specimens in the Chilga sample of Phiomia comprise nearly the entire dental series and show sufficient morphological uniformity and size progression along the tooth row to be accommodated in a single species ( Table 2). The adult dental formula is 1−0−3−3/1−0−2−3. P4 is bilophodont, has a central cusp in its posterior loph, and its posterior cingulum connects with the hypocone to form a distocrista ( Fig. 8A ; see Tassy 1994a). The fourth deciduous premolar is trilophodont and has posterior accessory central conules. Adult molars are fully trilophodont ( Figs. 7–10 ). Cheek teeth are low−crowned, bunodont and have simple half−loph(id)s composed of a dominant outer cusp and a single, smaller adaxial conelet. With few exceptions, accessory central conules are restricted to the pretrite side of the first transverse valley, and trefoil enamel wear figures are rudimentary or absent. Cingular development is weak. Enamel is thick and ornamented basally by horizontal striations and rugosities. There is no cementum. First molars are markedly smaller than second molars.

Relative to the cheek teeth, the mandibular symphysis and incisors in the sample are notably outsized compared with those of Fayum palaeomastodont species ( Table 3). Upper incisors are more robust and curved than lower incisors. The most complete I2, from CH 17−1, has a length of 450 mm. Toward its base, it is of flattened piriform shape, higher than wide. It is more rounded in cross−section toward the tip. When in its alveolus, the tusk would have curved down and outward. There is a prominent wear facet covering a distance of 160 mm along the ventromedial surface of the tusk towards the tip ( Fig. 8C ). Longitudinal striations mark the surface, and enamel is present as a band along the outer face of the tusk. Another I2, CH 14−2, also has a band of posttrite

α, greatest diameter in cross section.

enamel along its lateral surface. A medial sulcus runs the length of the tusk.

Specimen CH 17−1 also includes the symphyseal portion of the mandible, as well as the lower tusks. The length of the tusk projecting from its alveolus on the right side is 205 mm; complete i2 length is 460 mm. At its alveolar insertion, the i2 is dorsoventrally elongated and has a transversely flattened, piriform shape in cross−section. It is more rounded in cross−section toward the tip. A shallow sulcus runs nearly the entire length of the medial surface of the tooth. The lower tusk is longitudinally straight ( Fig. 11B ); it exhibits a distinct, flattened wear surface ventromedially at its distal end that is 90 mm long. Longitudinal striations are present circumferentially the length of i2. In anterior view, the i2 is obliquely angled within its alveolus ventromedially to dorsolaterally at 30° from vertical.

The symphysis ( Fig. 11A ) is 360 mm long in the midline from the posterior symphyseal margin to its distal tip. The incisor alveoli are closely appressed and are demarcated by a shallow midline sulcus on the ventral side of the symphyseal portion of the mandible. Downward angulation of the incisor alveoli on the symphysis measures only 7–8° from horizontal. The width of the alveolar area of the symphysis is remarkably constant along its entire length, measuring 89 mm proximally and 93 mm distally. The anterior chambers of the mandibular canals can be seen on each side of the symphysis. Dorsally, prominent longitudinal ridges on each side demarcate a broad, shallow midline channel running the length of the symphysis ( Fig. 11A ).

Comments.—Palaeomastodonts ( Palaeomastodon and Phiomia ) are best known from the Jebel Qatrani Formation of the Fayum, Egypt ( Andrews 1901, 1904, 1905, 1906, 1908; Andrews and Beadnell 1902; Matsumoto 1922, 1924; Osborn 1936; Lehmann 1950; Simons 1968; Gagnon 1997). The Jebel Qatrani Formation is late Eocene to early Oligocene in age, the youngest beds of which are ca. 32 Ma (Kappelman 1992; Kappelman et al. 1992; Gingerich 1993), and therefore its proboscideans are of greater antiquity than the new specimens from Chilga. Palaeomastodonts are also known from the late Eocene Qasr el Sagha Formation of the Fayum ( Coppens et al. 1978), the early Oligocene site of Taqah, Oman ( Thomas et al. 1989 a,b, 1999), and the less securely dated sites of Jebel Bon Gobrine, Tunisia ( Arambourg and Burollet 1962), Zella Oasis and Dor el Talha, Libya ( Arambourg and Magnier 1961; Savage 1969), and possibly Malembe, Angola ( Pickford 1986a). There are no radiometric dates or other definitive evidence showing that the proboscideans from these sites are as young geologically as those from Chilga, although based on stratigraphic position Arambourg and Burollet (1962) believed the fauna from Jebel Bon Gobrine to be late Oligocene in age.

Despite a substantial temporal gap, Fayum and Chilga palaeomastodonts share a common suite of dental features (see Andrews 1906; Matsumoto 1922, 1924; Lehmann 1950; Tobien 1978). Along with smaller molar dimensions ( Fig. 12 ), these features distinguish palaeomastodonts from gomphotheres and mammutids. Chilga incisors and most cheek teeth, however, are longer than those of Fayum palaeomastodonts ( Table 3, Fig. 12 ), documenting phyletic size increase over the course of the late Eocene–late Oligocene.

Palaeomastodont taxonomy has had a complicated and unstable history. Originally, Fayum palaeomastodonts were placed in a single genus, Palaeomastodon ( Andrews 1901) . A second genus, Phiomia , was subsequently recognized in the Fayum palaeomastodont collection ( Andrews and Beadnell 1902; Andrews 1904, 1905, 1906). Later, Fayum palaeomastodonts were further sorted into multiple species of Palaeomastodon and Phiomia ( Matsumoto 1922, 1924; Osborn 1936). Recent classificatory schemes have tended to accept both genera (e.g., Tobien 1978; el−Khashab 1979; Shoshani and Tassy 1996: appendix B), or to separate them at even higher taxonomic levels ( Kalandadze and Rautian 1992; McKenna and Bell, 1997). Periodically, however, Phiomia has been taxonomically subsumed into or made a subgenus of Palaeomastodon ( Andrews 1908; Lehmann 1950; Tobien 1971; Coppens et al. 1978), reflecting sorting difficulties posed by the enormous morphometric variability encompassed in the Fayum assemblage.

Several conflicting hypotheses have also been advanced about palaeomastodont systematics. In one view, Palaeomastodon was considered a predecessor of mammutids, while Phiomia was thought to be ancestral to other elephantoids, such as Gomphotherium ( Matsumoto 1924; Tobien 1971, 1978). Others have conjectured a special ancestor−descendant relationship between Phiomia and amebelodontines ( Osborn 1919, 1936; Borissiak 1929; Tobien 1973). There is currently little enthusiasm for these hypotheses, and more recent parsimony treatments of proboscidean phylogeny suggest that Phiomia is the sister taxon to all elephantoids, including mammutids; Palaeomastodon is corre−

Chilga, Ethiopia ( CH 9-2, CH 22-1, CHS4-1, Phiomia major sp. nov.; CH 14-V-12, CH 35-V-23, aff. Palaeomastodon sp. nov. A; CH 14-11, aff. Palaeomastodon sp. nov. B, CH 25-V-12a, b,

cf. Gomphotherium sp. nov.)

Fayum specimens of Phiomia

Fayum specimens of Palaeomastodon

Fayum palaeomastodont specimens, gen. et. sp. indet.

Hemimastodon crepusculi, Dera Bugti , Pakistan

Miocene African and Eurasian Gomphotherium

Eozygodon morotoensis , Moroto, Uganda and

Meswa Bridge, Kenya

Zygolophodon turicensis , numerous early-late Miocene sites, Europe, and Z. gobiensis, Tung Gur , China

Z. aegyptensis, Wadi Moghara , Egypt

Z. metachinjiensis , upper-middle Chinji Formation, Siwalik Series, Pakistan spondingly seen as the sister taxon to Phiomia +elephantoids ( Tassy 1994a, 1996b; Shoshani 1996).

Numerous craniodental features purportedly differentiate Phiomia from Palaeomastodon and are inventoried in Matsumoto (1922, 1924), Coppens et al. (1978), Tobien (1971, 1978), el−Khashab (1979), and Tassy (1994a). Within these genera, species have been recognized primarily by size variation (e.g., Matsumoto 1924). While some features may be too variable intragenerically to be diagnostic ( Coppens et al. 1978; Tassy 1994a), the specimens listed in the hypodigm above manifest a preponderance of features traditionally used to identify Phiomia , such as molar trilophodonty (including M3/m3), bunodonty, elongation of the symphysis, and presence of large accessory central conules in at least the first transverse valley, supporting their allocation to that genus.

aff. Palaeomastodon sp. nov. A

Figs. 13 and 14 , Table 2.

Referred specimens: CH 3−V−62, partial left M1?; CH 14−V−12, left M3 ( Fig. 13A ); CH 35−V−23, worn right M3 ( Fig. 14A ); CH 71−16, left distal femur; CH 71−20, worn right M3.

Age and distribution: Late Oligocene, ca. 28–27 Ma. Only known from the Upper Guang and Gahar Valley Sections, Chilga region, northwest Ethiopia.

Description.—These teeth are palaeomastodont in morphological grade. Several M 3 specimens are larger than those in Fayum species of Palaeomastodon ( Fig. 12 ), but otherwise are similar in their incomplete trilophodonty.

Based on the size of other elephantiform molars from Chilga, one specimen, CH 3−V−62, is probably an M1 ( Table 2). The posterior two−thirds of its crown is all that now remains, and is composed of a metaloph worn into a complete loop and a poorly formed tritoloph. On the pretrite side of the tritoloph, the conelet is worn into an enamel loop set posteriorly oblique to the long axis of the crown; it is partnered on the posttrite side by a single, small conelet .

Each M3 also has only two and one−half lophs, no accessory central conules, and exhibits wrap−around cingular shelves (most prominent lingually) ( Figs. 13A and 14A ). A single diminutive conelet occupies the posttrite side of the tritoloph; it is smaller and lower than the pretrite half−loph. With the exception of the last posttrite conelet, the other half−lophs may be superficially subdivided into two or three apical digitations. Slight crests descend posteriorly from the outer posttrite conelets of the proto− and metalophs into the transverse valleys. The mesoconelet of the second pretrite half−loph is set anterior to the corresponding mesoconelet of the posttrite side, and the second pretrite half−loph is angled posteriorly obliquely to the long axis of the crown. Specimen CH 35−V−23 also exhibits choerolophodonty on its less worn posttrite side.

Comments.—These specimens appear to belong in Palaeomastodon . They differ from corresponding molars of Phiomia major sp. nov. by the incomplete formation of their last lophs, and by the presence of weak posterior posttrite cristae. These features have been regarded as characteristic of Palaeomastodon ( Matsumoto 1924; Tobien 1978; Tassy 1994a). Indeed, these Chilga M 3 specimens are close to the paratype M3 of Palaeomastodon intermedius (AMNH 13449, from the Fayum; see Osborn, 1936: figs. 93, 94) in the disposition of their cusps and conelets, especially the obliquity of the second pretrite half−loph ( Figs. 13A, B , 14A ). Furthermore, the incomplete development of the tritoloph, and its intimate connection with the posterior cingulum are identical in AMNH 13449 and CH 35−V−23. This morphology is not observed in molars of Phiomia from Chilga. However, some Fayum molars attributed to Phiomia have posttrite cristae ( Tassy 1994a) or are incompletely trilophodont (for example, M 3 specimens AMNH 13492 and AMNH 13493; Matsumoto 1924: figs. 29 and 31). Thus, while it is clear that multiple new palaeomastodont species are represented in the Chilga assemblage, it is best to defer naming these few morphological outliers until additional diagnostic evidence has been obtained to refine their taxonomy.

aff. Palaeomastodon sp. nov. B

Fig. 14 , Table 2.

Referred specimens: CH 14−11, right M3 ( Fig. 14B ); CH 25−16, left M2.

Age and distribution: Late Oligocene, 28–27 Ma. Only known from the Upper Guang and Gahar Valley Sections, Chilga region, northwest Ethiopia.

Description.—Specimen CH 14−11 is severely step−fractured and distorted. A broad fissure runs through the metaloph, substantially lengthening the crown. The specimen has since been restored close to original shape and size by re−aligning cast sections across their breaks ( Fig. 14B , Table 2). This tooth is a low−crowned M3 with two and one−half lophs, and is ringed by a cingular shelf that is most prominent anteriorly and lingually. Half−lophs are each formed of a large outer cusp and a smaller, slightly lower adaxial conelet, with the exception of the posttrite side of the tritoloph, which is formed simply of an enlarged tubercle on the buccal side of the posterior cingulum. Enamel is thick and very rugose or “ptychodont”. There is no cementum.

The protoloph of CH 14−11 is transversely straight and has a strong crescentoid descending postero−medially from the outer pretrite cusp into the first transverse valley. It also has a posterior crest on the outer posttrite cusp. A very low tubercle occludes the floor of the first transverse valley on the posttrite side. The pre− and posttrite sides of the metaloph are transversely offset (the pretrite half is more anterior). A small anterior accessory central conule projects from the outer pretrite cusp of the metaloph into the first transverse valley. The corresponding outer posttrite cusp has a more pronounced posterior crest than in the protoloph. The third pretrite half−loph dwarfs the single conelet on the posttrite side, and has a low posterior accessory central conule that connects it with the posterior cingulum.

A moderately worn upper molar ( CH 25−16), probably M2, with large anterior and small posterior interproximal facets, is morphologically and dimensionally conformable with CH 14−11 and is therefore also included in the hypodigm of this taxon.

Comments.—Specimen CH 14−11 defies precise taxonomic allocation. Its occlusal configuration generally resembles M3 of Palaeomastodon , including aff. Palaeomastodon sp. nov. A ( Figs. 13A, B , 14A ). However, it differs from Palaeomastodon M3 by its more rectangular occlusal outline, greater enamel rugosity, more pronounced development of posterior posttrite cristae and cingula, and by its considerably greater size (it is approximately half again larger than M 3 in Fayum palaeomastodonts, and is about 20 percent larger than CH 14−V−12 and CH 35−V−23; Fig. 12 and Table 2; Andrews 1906; Matsumoto 1924; Lehmann 1950). While sexual dimorphism could account for size variation among CH 14−11, and CH 14−V−12, CH 35−V−23, and CH 71−20, their morphological contrasts are great enough to warrant allocation to different species.

In the ways that CH 14−11 differs from or is more pronounced than Palaeomastodon molars, it anticipates the dental morphology of early Miocene mammutids. The oldest recognized mammutid is Eozygodon morotoensis , dated to ca. 23 Ma at Meswa Bridge, Kenya ( Bishop et al. 1969; Pickford and Tassy 1980; Pickford and Andrews 1981; Tassy and Pickford 1983; Pickford 1986b) and>20.6 Ma at Moroto, Uganda ( Gebo et al. 1997; but see Pickford et al. 1986 and Pickford et al. 1999, who feel that Moroto I and II are ca. 17–15 Ma based on biochronological comparison). Its similarity to CH 14− 11 in cingular, posterior crescentoid, and posterior posttrite cristae (“zygodont crest”) morphology is conspicuous. For this reason, prior to its restoration CH 14−11 was initially considered mammutid ( Sanders and Kappelman 2001). However, in Eozygodon and Zygolophodon , another mammutid which first occurs in the early Miocene (in Europe and North Africa), molar tritolophs are complete and (in M3) associated with a pronounced posterior cingulum, meta− and tritolophs are transversely straighter and their half−lophs are not offset, and crown height and molar size are greater ( Fig. 12 ; Tobien 1975, 1996; Tassy and Pickford 1983; Tassy 1985; Göhlich 1998, 1999; Sanders and Miller 2002).

Other aspects of crown organization in CH 14−11 (ptychodonty, transverse offset of pre− and posttrite half−lophs, oblique orientation of pretrite half−lophs, and anterior advancement of pretrite mesoconelets relative to posttrite half−lophs) more closely resemble molar morphology in the oldest known choerolophodont, Afrochoerodon kisumuensis , documented from the early−mid Miocene sites of Wadi Moghara, Egypt ( Sanders and Miller 2002), and Cheparawa and Maboko, Kenya ( MacInnes 1942; Tassy 1977 a, 1985, 1986; Pickford 2001). This species is possibly synonymous with Choerolophodon palaeindicus , presumably from the early Miocene levels of the Bugti Beds, Pakistan ( Cooper 1922; Raza and Meyer 1984; Tassy 1985, 1986; see Welcomme et al. 1997, 2001). As with mammutids, however, archaic choerolophodonts also have M3s of larger size and that are more advanced in the greater development of their tritolophs and posterior cingula than CH 14−11. While CH 14−11 approximates M3 of Hemimastodon (also presumably from the early Miocene levels of the Bugti Beds, Pakistan) in size ( Fig. 12 ), it differs from that obscure elephantoid as well in lacking full trilophodonty and in the offset of its half−lophs ( Pilgrim 1912; Tassy 1988).

Because of its primitive loph formula and lack of further evidence with which to more comprehensively assess its affinities, for now we prefer to conservatively refer CH 14−11 to Palaeomastodon . Nonetheless, the intriguing morphological and temporal intermediacy of CH 14−11 between palaeomastodonts and these early Miocene taxa suggests that recovery of additional fossil material of aff. Palaeomastodon sp. nov. B might prove useful for refining current hypotheses (e.g., Tassy 1994a, 1996b; Shoshani 1996) about sistergroup relationships between particular palaeomastodont and elephantoid taxa.

Figures

Fig. 7. Dental specimens of Phiomia major sp. nov. Anterior is to the left. A. Left P3, CH25−5, in buccal (reversed) (A1) and occlusal (A2) views. B. Right M1, CH3−53, in occlusal view; note piece of fossil wood along midline of crown. C. Left M3, CH22−1, in buccal (C1) and occlusal (C2) views. Scale bar 1 cm.

Fig. 8. Dental specimens of Phiomia major sp. nov. Anterior is to the left. A. Left P4, CH4−2, in lingual (A1) and occlusal (A2, A3) views. B. Left M2, CH9−1, in occlusal view. C. Right I2, CH17−1, part of holotype, in medial (C1) and lateral (reversed) (C2) views.

Fig. 9. Dental specimens of Phiomia major sp. nov. A. Right p3, CH4−4, in lingual (A1) and occlusal (A2, A3) views. B. Left m1, CH25−1, in buccal (B1) and occlusal (B2, B3) views. C. Right m3, CH9−2, in occlusal view. Anterior is to the left. Broken areas indicated by small dotted lines and cross−hatching. Scale bar 1 cm.

Fig. 10. Dental specimens of Phiomia major sp. nov. A. Left p4, CH5−15, in occlusal (A1) and buccal (A2) views. B. Left m2, CHS4−1, in buccal (B1) and occlusal (B2, B3) views. Anterior is to the left. Broken areas indicated by small dotted lines and cross−hatching. Scale bar 1 cm.

Fig. 11. Phiomia major sp. nov., parts of holotype, CH17−1. A. Mandibular symphysis and right i2 projecting from alveolus, in dorsal view. B. Left i2 in medial (reversed) view. Anterior is to the left. Scale bar 3 cm.

Fig. 12. Bivariate plots of natural log−transformed M3, m3, and m2 crown length versus width in Chilga and other fossil elephantiforms. Comparative dimensions are from Fourtau (1920), Cooper (1922), Osborn (1936), Bergounioux and Crouzel (1959), Arambourg (1961), Hamilton (1973), Gaziry (1976, 1987), Tassy (1977b, 1983a, 1983b, 1985), Tassy and Pickford (1983), Gentry (1987), Roger et al. (1994), Göhlich (1998), and Sanders and Miller (2002). A. M3. B. m2. C. m3.

Fig. 13. Dental specimens of Chilga palaeomastodonts, aff. Palaeomastodon. A. aff. Palaeomastodon sp. nov. A., left M3, CH14−V−12, in lingual (A1) and occlusal (A2, A3) views. B. Diagrammatic representation of occlusal view of right M3, AMNH 13449, paratype of Palaeomastodon intermedius (after Osborn 1936: figs. 93, 94). Anterior is to the left. Scale bar 1 cm.

Fig. 14. Dental specimens of Chilga palaeomastodonts, aff. Palaeomastodon. A. Right M3, CH35−V−23, aff. Palaeomastodon sp. nov. A. in occlusal view. B. Right M3, CH14−11, aff. Palaeomastodon sp. nov. B. Original specimen in buccal (B1) and occlusal (B2) views; note step fractures and separation of crown segments, indicated by dotted lines. Reconstructed cast specimen in occlusal view (B3, B4). Anterior is to the left. Broken areas indicated by dotted lines. Scale bar 1 cm.