Holotype: BSPG 1926 X, right maxilla with P3, M1–2, currently lost (Gertrude Röβner, personal communication 2009). Figured in Stromer 1926: pi. 40: 17; also figured in Pickford 1984: fig. 1.

Type locality and horizon: Langental, Namibia, Early Miocene.

Material.—CGM 82975, right mandible fragment with p4—m2; WM 97–697, right mandible fragment with m2–m3; WM 05–50, palate with left and right P2—M3 (Fig. 1A); WM 06–49, left mandible with p4—m3 (Fig. 4A); WM 06–14, left mandible with dp4—m2 (Fig. 4F); WM 05–21, right M2 (Fig. 4B); WM 05–48, right m3 (Fig. 4C); WM 06–55, right mandible with worn m3 (Fig. 4G); WM DEC06-11, right ml (Fig. 4D); WM DEC06-025, m3 (Fig. 4C); DPC 17688, palate with roots of left and right P2-M3 (Fig. 2); DPC 12599, left mandible, il—ml (Fig. 3A); DPC 14581, mandible with symphysis, roots of right il—p2, left il—p4 (Fig. 3B); DPC 8997, right mandible, m1–2, p4 erupting (Fig. 1C); DPC 6469, juvenile mandible with dp4, m1—2 (Fig. 1D); DPC 6618, mandible with m2–m3 (Fig. 1B); DPC 12942, left maxilla containing P4—M2.

Diagnosis.—Differs from Sanitherium in having less molarized premolars, and in having anterior and posterolingual cusps less developed; P4 with two main cusps and two subsidiary ones; metastylid prominent on unworn specimens; m1–m3 ca. 40 mm (range 37.5 to 42.5 mm) (after Pickford 1984).

Description.—Palate and upper dentition: WM 05–50 is the most complete sanithere palate known, all other material being distorted or broken (Fig. 1). The specimen is of a young adult, with the M3s beginning to erupt, and M1s in medium wear. Lingually the cheek tooth rows are parallel to each other (distance between the P2s and the M2s is 17 mm). In this young individual the palatine foramen is opposite the rear of Ml (Fig. 1A₂) and the posterior choanae invaginate the palate as far forwards as the front of the M3. However, this position would probably have changed had the individual lived to maturity, as revealed by the position of the palatine foramen in DPC 17688, which lie opposite the middle of M2 (Fig. 2B). The rear of the zygomatic process of the maxilla is located above the rear of M2, and anteriorly it rises from the facial surface at about the level of P4—M1. In lateral view, the occlusal surface of the cheek teeth in WM 05–50 appears convex from front to back, a condition that is commonly observed in sanitheres but which is unlike that observed among suids, in which the tooth row is concave (Pickford 1986, 2004). A second specimen, DPC 17688, preserves the snout but with much worn cheek teeth, and also shows a convex tooth row (Fig. 2C). In this latter specimen, the infraorbital foramen opens above P3 and is preceded by a large depression, which extends anteriorly as a broad groove that rises and narrows above the canine jugum, as in a sanithere specimen KNM-BG 41925A from Aka Aiteputh, Kenya (Pickford and Tsujikawa 2005). Although a Pl crown is not represented in the collection, the roots of this tooth are present in DPC 17688, and reveal a biradicular condition. On WM 05–50 (Fig. 1A), P2 has a prominent paracone and metacone, forming a continuous buccal wall. Lingually, the protocone is close to the paracone and is followed distally by a depression that is full of wrinkled enamel. This depression is bordered lingually by a prominent puffy cingulum. P3 is constructed along the same lines as the P2, but the cusps are more isolated from each other. In particular, the protocone is separated from the paracone and is oriented more obliquely. P4 is essentially a larger version of P3, but with a better developed, broader, distal part, and the paracone and metacone are separated from each other by a shallow valley. M1 is comprised of four sub-equal cusps, and anterior and buccal cingula. The protocone sends a strong crest antero-buccally where it joins the pre-crista of the paracone, the junction lying distal to the anterior cingulum. The protocone is broader than the paracone. The hypocone has a sloping lingual surface, so that its apex is close to the midline of the crown. The metacone has an almost circular enamel outline. There is a low cingular remnant at the lingual end of the median transverse valley. M2 differs from M1 in being slightly larger in size, and in having a more prominent cingular cusplet at the lingual end of the transverse valley. Because the M2 crown is less worn than that of M1, the wrinkling in the enamel is more in evidence, and the grooves on the buccal cusps can be discerned. The orientation of these grooves is typical of Sanitheriidae, and quite different from that seen in suids. The buccal cingulum is narrow with a beaded edge, it is continuous from front to back, and it extends onto the distal surface of the metacone and hypocone. There is a posterior accessory cusplet in the midline of the tooth, between the metacone and hypocone, which joins the distal cingulum. M3 is similar to M2, save that it sports a small talon cusp in the disto-lingual extremity of the crown, imparting a trapezoidal occlusal outline to the tooth. Measurements of the upper and lower teeth from Moghra are presented in Table 1. Tables S1 and S2 (see Supplementary Online Material at  http://app.pan.pl/SOM/app55-Pickford_etal_SOM.pdf) provide measurements for the Moghra sanitheres compared with a large sample of other African and Eurasian sanitheres.

Mandible and lower dentition: DPC 12599 is a left mandible with the two central incisors, the left i2-p2 and p4 as well as the roots of p3 and anterior root of m1 (Fig. 3A). The individual was very old when it died, the crowns of the teeth being deeply worn. The central and second incisors in particular show secondary dentine associated with the pulp canal, leaving little of the crown to describe. What remains are oval occlusal surfaces worn almost flat. The il and i2 are subequal in dimensions, whereas the i3 is appreciably smaller than the i2. There are extremely short gaps anterior and distal to the i3. The canine is small with two fused roots, and is in contact with the pi, which also has fused roots. The i3 is slightly compressed labio-lingually, and shows a blunt distal crest. The canine is oval with a posterior crest bordered by shallow grooves. Its morphology and dimensions indicate that the individual was a female (Pickford 2006). The pi is deeply worn, but shows evidence of a distal cusplet separated from the main cusp by shallow buccal and lingual grooves. The p2 is in medium wear and reveals that the crown is comprised of a main cusp positioned above the gap between the two roots. From the apex of this cusp there is an anterior crest that curves lingually near the anterior end of the tooth, and there is a stronger posterior crest and postero-lingual cusplet. The p3 crown is missing. The p4 is damaged distally, but it has a main cusp above the gap between the two roots. There is a prominent lingual cuspid attached to the main cusp which broadens the centre of the tooth. An anterior crest curves lingually close to the anterior margin of the crown. Distally, there is a broad crest descending from the apex of the tooth towards the distal edge but the cusplet at its disto-lingual margin is damaged. In lateral view, the most striking aspect of this mandible is the curvature of the occlusal surface of the incisor battery. This curvature is enhanced by the advanced wear stage of the teeth, but nevertheless, there can be little doubt that the occlusal surface of the incisors curves strongly ventrally starting at the level of the canines. There are four mental foramina, three large ones at mid-height of the jaw, respectively beneath the anterior root of the p3 and the two roots of the m1, and a smaller one slightly lower down, beneath the anterior root of p4. There are also two prominent nutritive foramina in the anterior part of the symphysis in line with the root of i2. In lingual view the symphysis is long and robust, terminating distally beneath the level of the p3. The same applies to an edentulous mandibular symphysis (DPC 14581) (Fig. 3B), in which the two halves of the jaw are solidly fused to each other, a condition that may be age-related, although it seems to be observed more often in sanitheres and other suoids than in anthracotheres.

WM 06–14 and WM 06–49 are two mandible fragments (Fig. 4A, F). The former specimen is a juvenile with the rear two thirds of the dp4 preserved, distal to which ml is fully erupted and m2 is in its crypt. The latter specimen is fully adult, with p4—m3 fully erupted. The rear of the symphysis in WM 06–49 lies opposite the middle of p3, and a break in the specimen anterior to the p3 alveolus reveals no sign of an enlarged canine alveolus, indicating that this individual was probably a female. There are mental foramina beneath the junction between p2—p3 and p4—m1, at about half the depth of the jaw. The mandible is slender and the tooth row is straight except where the m3 curves slightly buccally. The rear two pairs of cusps of the dp4 in WM 06–14 are in medium wear and have lost most details of crown morphology. What remains looks like a small version of the ml. The p4 in WM 06–49 has a large protoconid and a prominent metaconid almost as tall as the protoconid, but is separated from it apically. Distally there is a low hypoconid from which a distal cingulum descends lingually. Buccally there is a low cingulum between the hypoconid and the protoconid and at the antero-buccal corner of the crown. The protoconid has prominent pre- and postcristids oriented antero-posteriorly, the pre-cristid curving lingually as it descends towards the cervix. There are two roots. The ml in the adult mandible is in light wear with dentine exposed on three cusps (not on the entoconid). There is a well developed, beaded, buccal cingulum which curves anteriorly and posteriorly at the ends of the tooth. The lingual cusps are transversely compressed and are thus narrower than the buccal cusps. The small cuspid immediately distal to the metaconid is just visible, although heavily worn. The protoconid has pre- and post-cristids that curve lingually, but not so far as to enclose the metaconid. Between the rear parts of the protoconid and metaconid there is a small but tall cuspid. The hypoconid also has pre- and postcristids that curve lingually, but only reach the midline of the tooth, the precristid blocking the median transverse valley.

The m2 is a larger version of the m1, but being less worn, it shows the structures of the crown better. In particular, the small cuspid just distal to the metaconid is clearly distinguishable, its apex being separated from that of the metaconid. The m3 is like the m2 save for the presence of a large, almost centrally positioned hypoconulid distal to the second lophid of the tooth. Measurements of the upper and lower teeth from Moghra are presented in Table 1, and Tables S1 and S2 (see Supplementary Online Material at  http://app.pan.pl/SOM/app55-Pickford_etal_SOM.pdf) provide dental measurements for the Moghra specimens compared with a large sample of other African and Eurasian sanitheres.

Remarks.—New sanithere specimens from Wadi Moghra increase our knowledge about the family Sanitheriidae, most notably concerning the form of the palate and lower incisors, which were absent or poorly preserved in previously available samples. The palate is remarkably uniform in breadth from P2 to M3, the inner margins of the upper cheek teeth being almost parallel. The posterior choanae invaginate anteriorly to the level of distal M2, but this may have changed had the individual lived to maturity. The morphology and dimensions of the mandible and upper and lower cheek teeth fall within the range of variation for Diamantohyus africanus (Stromer 1922, 1926; Pickford 1984). The upper premolars of the more derived species Diamantohyus nadirus are more complex, with better developed cutting edges in the anterior cusps (Pickford and Tsujikawa 2005; Pickford 2006). The Moghra material differs from Eurasian Sanitherium leobense (de Bonis et al. 1997) and Sanitherium schlagintweiti (Koufos 2007) in its narrower molars and simpler premolar morphology.

Geographic and stratigraphic range.—Namibia (Langental, Fiskus, Bogenfels); Kenya (Rusinga (Hiwegi, R1a, R1, R3, R45, RVII, R91, Kaswanga, Kulu, Gumba), Locherangan, Kalodirr, Karungu, Kajong, Majiwa, Chianda Uyoma); Uganda (Bukwa, Napak); Egypt (Wadi Moghra); Libya (Gebel Zelten), early Miocene.

Type species: Nguruwe kijivium (Wilkinson, 1976), early Miocene, Uganda.

Holotype: NAP I'64, left maxilla fragment with M1–3, figured in Wilkinson 1976: pl. 2: g; Pickford 1986: fig. 6.

Type locality and horizon: Early Miocene, Napak I, Uganda.

Material.—WM 97-715, left m3; WM DEC06-10, worn left m3 (Fig. 5C); DPC 4385, right mandible fragment with m2–3 (Fig. 5A); DPC 14564, right m3 (Fig. 5B).

Diagnosis.—A small genus of Kubanochoerinae in which Il is labiolingually compressed, not meeting interproximally; P4 with two main cusps and complete cingulum; molars with thick enamel, inflated main cusps, closed lingual notches; simple talon/id in upper and lower third molars; occlusal outline of m3 symmetrical; p3 with wide distal lingual cingular cusp or platform; p4 with metaconid almost completely suppressed; dm4 buccolingually inflated; lower canine scrofic; upper canine with dorsal cement cover (Pickford 1986).

Description.—DPC 4385, a right mandible fragment, retains m2 and m3 in medium wear (Fig. 5A). The m2 has four main cusps arranged in two lophids and there are small anterior, median and posterior cusplets in the center line of the crown. The median accessory cusplet is antero-posteriorly compressed, transversely elongated, and at the wear stage in which the dentine lake extends over about half the width of the crown. The median transverse valley is extremely narrow from just above cervix level to the occlusal surface of the tooth, as in other specimens of Nguruwe. The m3 is similar to the m2 save for the presence of a large talonid cusp positioned slightly buccal to the mid-line of the tooth. The posterior transverse valley of this tooth possesses low tubercles buccally and lingually and there is a low tubercle at the buccal extremity of the median transverse valley.

Lower third molars of N. kijivium from Wadi Moghra are simple, bunodont teeth with a small almost centrally positioned talonid (Fig. 5C). WM DEC06-55 shows a medium amount of wear, with dentine showing at the apices of the two anterior cusps, and all evidence of the furchen abraded away. The enamel is thick as in specimens of this genus from East Africa and Namibia (Pickford 1986, 1987, 1995, 1997, 2003, 2004). The hypoconulid is located slightly to the buccal side of the central line, and it has cingular crests running antero-buccally and antero-lingually. The anterior cingulum is narrow and extends slightly onto the buccal side of the protoconid. The anterior accessory cusplet is deeply worn, being barely perceptible. The median accessory cusplet is heavily worn showing a transversely elongated wear facet occupying about half the extent of the transverse valley. This is similar to the situation observed in East African and Namibian specimens of the species that are worn to a comparable stage. The posterior accessory cusplet lies anterior to the hypoconulid in the midline of the tooth. Cingular remnants block the buccal ends of the median and posterior transverse valleys, but there is no cingulum at the lingual end of the median transverse valley. The dimensions of these teeth are close to those of Nguruwe kijivium from Eastern and Southern Africa (Table 2). In addition, Tables S3 and S4 (see Supplementary Online Material at  http://app.pan.pl/SOM/app55-Pickford_etal_SOM.pdf) compare M3 and m3 dimensions of Kubanochoerinae from Moghra with a large sample of other African and Eurasian kubanochoerines.

Remarks.—Fossils from Moghra attributed to Nguruwe kijivium are similar to specimens of this taxon from East Africa and Namibia. This species ranges in age from about 20–18 Ma in East Africa, and up to about 17.5 Ma in Namibia. It is thus a useful marker for the upper part of the early Miocene, and possibly the base of the middle Miocene. The deposits where this species has been found are usually correlated to MN 3 and MN 4 of the European mammal zones.

Geographic and stratigraphic range.—Uganda (Napak, I, IV, V); Kenya (Songhor, Koru, Mteitei, Mfwangano, Rusinga (Hiwegi)); Egypt (Wadi Moghra); Namibia (Arrisdrift), early to ?middle Miocene.

Type species: Libycochoerus massai Arambourg, 1961, early to middle Miocene, Libya.

Holotype: Left mandible with p2–m3, No. 1961-5-8 MNHN Paris.

Type locality and horizon: Terminal early Miocene to basal middle Miocene, Gebel Zelten, Libya.

Material—WM DEC06-09, partial right M3 (Fig. 5D); CGM 30791, right talus; DPC 17744 canine and DPC 14565 canine are provisionally referred to this species.

Diagnosis.—Large kubanochoere, about 10–20% smaller than Kubanochoerus robustus; upper molar series about 100 mm long (Pickford 1986).

Description.—WM DEC06-09, a partial upper third molar is unworn but is missing part of the paracone and the buccal edge of the metacone (Fig. 5D). However, enough remains of the tooth to reveal its main characters. The protocone, metacone and hypocone are well preserved conical bunodont cusps with coarse but shallow furchen and wrinkles. The anterior and median accessory cusplets are also conical and voluminous, being about 2/3rd the height of the main cusps. The anterior accessory cusplet is located well anterior to the line joining the apices of the paracone and protocone, and it is fused to the beaded anterior cingulum. The hypocone has a cingular remnant on its antero-lingual corner, partly blocking off the transverse valley. The median accessory cusp is relatively small and low. The hypoconule is located in line with the protocone and hypocone, and it sends a descending beaded cingulum antero-buccally. This tooth is remarkably similar to material of Libycochoerus massai from Gebel Zelten (see specimens illustrated by Van der Made 1996: pl. 10: 7, 8). In terms of size, WM DEC06-09 plots out at the small end of the range of variation of L. massai from Gebel Zelten (Fig. 7), but we have little hesitation in attributing this tooth to the species. Dental dimensions for the Moghra specimen are provided in Table 2 and Table S4 (see Supplementary Online Material at  http://app.pan.pl/SOM/app55-Pickford_etal_SOM.pdf) provides comparative dental data from a large sample of other African and Eurasian kubanochoerines.

CGM 30791 is a right talus from a large suid the dimensions of which are close to those of LBE 505, a right talus from Gebel Zelten attributed to L. massai (Orliac 2007: fig. 66.2). Dimensions of the Moghra and Zelten tali are provided in Table S5 (see Supplementary Online Material at  http://app.pan.pl/SOM/app55-Pickford_etal_SOM.pdf).

Geographic and stratigraphic range.—Wadi Moghra, Egypt; Gebel Zelten, Libya (Arambourg 1961, 1963), early to middle Miocene.

Locality and horizon: Moruorot and Kalodirr are aged ca. 17.2 Ma, whereas Rusinga and Karungu are 17.8 Ma (Drake et al. 1988).

Material.—DPC 8947, left mandible fragment with wind-abraded ml to m3.

Description.—DPC 8947, a sand-blasted left mandible with part of the right symphysis preserved, contains worn and damaged m1–m3 and roots of p4 (Fig. 6). The symphysis is solidly fused and very thick dorso-ventrally, its posterior extremity lying beneath about the p3, but the specimen is so damaged that little can be said about its morphology. It is possible to see however, that the junction between the two halves of the symphysis forms a raised ridge lingually, suggesting that the individual was aged when it died, a possibility supported by the heavy wear on the m3. The buccal surface of the jaw has been destroyed leaving nothing to be observed concerning the position of the mental foramen or the dimensions of the canine. The sublingual fossa is large and extends anteriorly as far as the ml, forming a deep recess beneath m3. The teeth are deeply worn occlusally and on their buccal sides have suffered greatly from wind-driven sand. Lingually the teeth are better preserved, and they reveal that the median transverse valleys in the m2 and m3 were broad near the cervix, as in the genus Libycochoerus (Pickford 1986). The median accessory cusps extend across less than half the bucco-lingual breadth of the tooth. This is clearly visible in the m2, in which the remnant of this accessory cusplet shows an almost circular outline, markedly divergent from the bucco-lingually elongated and mesio-distally compressed outline that occurs in teeth of Nguruwe species. The ends of the median transverse valleys do not possess any tubercles or cingular remnants. Although accurate measurements are difficult to take on account of the abrasion of the specimen, DPC 8947 is clearly larger than known specimens of Kenyasus rusingensis (Fig. 7), and is about the same size as Libycochoerus jeanneli and L. anchidens (Table 3).

Discussion.—The restricted amount of morphology that is discernible in this wind-abraded fragment is not compatible with Listriodontinae or Hyotheriinae, but recalls Kubanochoerinae. The dimensions of DPC 8947 are comparable with both Libycochoerus anchidens and Libycochoerus jeanneli, and both species are known from East Africa in deposits dated to between ca. 17.8 and 17.2 Ma (Pickford 1986, 2007). Miocene suid taxa tend to be extremely widespread, so the presence of one or the other of these species at Moghra would not be surprising. The species Nguruwe kijivium is equally widespread, being reported from Namibia, East Africa and Moghra. The possible presence of Libycochoerus anchidens or Libycochoerus jeanneli at Moghra would provide additional support for the view that the stratigraphic succession in the region includes deposits spanning the end of the early Miocene.

Geographic and stratigraphic range.—Libycochoerus jeanneli is known from Moruorot (type locality) and Kalodirr, Kenya (Arambourg 1933; Pickford 1986; Orliac 2007): Libycochoerus anchidens is known from Rusinga (type locality) and Karungu, Kenya (Pickford 1986; Van der Made 1996), early Miocene.