Fossil suids were first found at Bolt’s Farm in the 1930’s (Broom, 1937, 1948; Shaw, 1938) and have been collected sporadically since then (Cooke, 1993; Pickford & Gommery, 2016). Four suid species have been recorded from the Bolt’s Farm Palaeokarst System (Table 1) comprising two assemblages separated in time: a Late Pliocene fauna with Notochoerus capensis and Potamochoeroides hypsodon (ca 3.0-2.7 Ma) and an Early Pleistocene one with Metridiochoerus andrewsi and Phacochoerus modestus (ca 1.8 Ma).

Recent excavations at Aves Cave I have yielded additional suid specimens, several of juvenile individuals, all of which can be attributed to Potamochoeroides hypsodon. Older deposits at Brad Pit ‘A’ have yielded specimens of Gerontochoerus koobiforaensis and younger deposits at Milo ‘A’, Bridge Cave and Alcelaphine Cave (Pickford & Gommery, 2016) have yielded remains of the large suid with hypsodont molars, Metridiochoerus andrewsi.

The aim of this contribution is to describe and interpret the new suid fossils from the Bolt’s Farm Palaeokarst System, to document the juvenile dentition and to refine the biostratigraphic implications of the fossils. Detailed morphometric comparisons are made with the Makapansgat Suidae. A discussion on the systematic status of the genus Notochoerus is extended (adding to the recent work of Pickford, 2013), which indicate that it is a suine rather than a tetraconodont as thought by most authors since the late 1950’s (Leakey, 1958; Ewer, 1958a; Cooke, 1976; Harris & White, 1979; Geraads & Bobe, 2017; Lazagabaster et al. 2018). The new analysis indicates that Potamochoeroides and Notochoerus are potentially synonymous.

The Bolt’s Farm Palaeokarst System (Pickford & Gommery, 2016) is part of a regional karst network developed in dolomitic country rock, outcrops of which are widespread in Gauteng, South Africa (Sénégas et al., 2002; Thackeray et al., 2008). The Gauteng karst infillings are known for the diverse palaeoanthropological discoveries which have been made in them (Herries et al. 2013), and for other mammals found alongside the hominids (Sénégas & Avery, 1998). Fossil suids are reasonably well represented, and have proven to be useful in two domains: biochronology and palaeoenvironmental studies (Pickford & Gommery, 2016).

The Bolt’s Farm fossil record, and that of Gauteng in general, indicates that karst processes were not uniform in intensity or locale through geological time, but were more intense during some periods, less active during others, and, as is usual in karst systems, the zones undergoing karst processes shifted laterally and vertically as well as partly reworking deposits in the pre-existing networks (Fig. 1). In the last case, deposits may contain composite faunas from two time periods, as for example at Brad Pit. The fossiliferous deposits that accumulated in the karst network can thus be highly complex, with cut-and-fill structures, decalcification of breccias, re-cementing of decalcified breccias, and the juxtaposition of young and old breccias is common. Differences of opinion in the literature reveal that sorting out the sequence of events has been arduous and often controversial, but biochronology has provided valuable constraints, and among the fossil groups that have yielded important sequence and age data are the Suidae (Cooke, 1993, 1994, 2005; Pickford, 2006, 2012, 2013a, 2013b; Pickford & Gommery, 2016).

The oldest recognised endokarst deposits in the Bolt’s Farm Palaeokarst System occur at Waypoint 160, aged ca 4.5 Ma (Sénégas, 2000, 2004; Sénégas & Avery, 1998) followed by Brad Pit ‘A’ aged ca 3.7 Ma, followed by Aves Cave I and related deposits (Pit 14) aged ca 2.6 Ma - 3 Ma, followed, after a break in deposition, by deposits aged ca 1.8 Ma (Pit 1, Pit 3, Milo ‘A’, Bridge Cave Panthera Spot). There are younger deposits in the same region such as the Rising Star Cave infillings (ca 0.25 Ma) (Dirks et al. 2017) close to Bolt’s Farm, but over most of the karst system, there is little depositional activity occurring today (an exception comprises the deeper parts of Baboon Cave where karst sedimentation processes are still active). A summary of the chronology of the sites is provided in Figure 1.

AC - Aves Cave I; AL - Alcelaphine Cave; BC - Bridge Cave; BPA - Brad Pit ‘A’; BPI - Bernard Price Institute, University of the Witwatersrand, Johannesburg (Now ESI); DNMNHP- Ditsong National Museum of Natural History, Pretoria; ESI - Evolution Studies Institute, University of the Witwatersrand, Johannesburg; MA - Milo ‘A’; NHMUK - Natural History Museum of the United Kingdom, London; PEM - Port Elizabeth Museum; TM - Transvaal Museum, Pretoria (Now DNMNHP).

The dental nomenclature employed in this paper is based on Hünermann (1968) modified by Pickford (1988) and Pickford (2013a).

Family Suidae Gray, 1821

Genus Potamochoeroides Dale, 1948

Type species.- Potamochoeroides hypsodon Dale, 1948

Species Potamochoeroides hypsodon Dale, 1948

Holotype.- ESI (BPI) M 303, right maxilla containing P3/-M3/.

Type locality and age.- Makapansgat, South Africa, Late Pliocene.

New material.- Excavations at Aves Cave I between 2015, 2017 and 2018, yielded a number of suid fossils, dominated by isolated teeth and postcranial bones, but also an adult maxilla containing three cheek teeth. Among this material, there are several deciduous teeth, especially incisors (9 specimens: Table 2), a canine, a D2/ two D3/s and a D4/, which provide precious information concerning the deciduous dentition in Potamochoeroides, revealing that it was similar to Potamochoerus Gray, 1854, and Dasychoerus Gray, 1873, in retaining the full complement of deciduous incisors and canines, and rather different from Phacochoerus in which the upper deciduous and permanent dentition lacks the lateral incisors (Fig. 2). The deciduous incisor battery of Metridiochoerus has not been reported, but the adults possess lower incisors like those of Phacochoerus (which often shed the permanent teeth (Pickford, 2013a).

The list of referred material is provided in Table 2.

Pickford & Gommery (2016) described a juvenile maxilla (AC 231) with the milk teeth and first permanent molar in occlusion and the M2/ incompletely formed within its crypt, indicating an individual of about 8 months of age (using an extant Sus scrofa developmental clock (miniature pig) Ide et al., 2013). They also described a lower milk incisor (AC 359) and fourth deciduous molar (AC 1) of an individual aged ca 5 months when it died.

The new fossils from Aves Cave I include several deciduous upper incisors and a canine as well as two examples of D3/ (Table 3) one of which is lightly worn at the cusp apices suggesting an individual of ca 3 months age, the other is deeply worn, indicating an older subject ca 11 months of age. There are also several fragments of upper and lower canines, some of which were incompletely formed at the time of death. An adult maxilla in the new collections (AC 515) is from an individual about 2 years old, and a heavily worn M2/ belongs to an even older individual, probably 2.5 to 3 years of age.

Pickford & Gommery (2016) attributed a complete upper canine (AC 96) to Potamochoeroides hypsodon. Several other upper canines have been collected from the Aves Cave deposits including a complete specimen and several fragments (Fig. 11). Apart from specimen AC 96, all the other canines from the Aves Cave complex are compatible with each other in dimensions and morphology and clearly represent a single species. These specimens are morphometrically close to abundant material from Makapansgat, the type locality of Potamochoeroides hypsodon, and we accordingly attribute them to this species. The Makapansgat sample of this species is comprehensive and even though there is a significant range of metric variation, best interpreted in terms of sexual bimodality and individual variation within a single species, none of the specimens is as large as AC 96 which has twice the height of the other complete canine from Aves Cave I (AC 371) (Fig. 11) as do specimens from Makapansgat attributed to Notochoerus capensis (Ewer, 1958). Furthermore, the upper canines attributed to Potamochoeroides have an almost straight medial part in the distal profile of the tooth, whereas the canine AC 96 is evenly curved thoughout its distal profile (Fig. 11). We therefore incline to the view that AC 96 probably belongs to Notochoerus capensis rather than to an abnormally large male of Potamochoeroides hypsodon. In metric terms this re-attribution is compatible with the other dental specimens of Notochoerus capensis from the site, which are somewhat larger than specimens of Potamochoeroides hypsodon, and smaller than specimens of Metridiochoerus andrewsi (Pickford & Gommery, 2016). The ventral part of the canine (AC 96) is covered in enamel which is patterned by coarse longitudinal ridges as in Potamochoerus and Potamochoeroides.

A second specimen from Aves Cave I, AC 530 from Block N° AC 11-15, confirms the presence of Notochoerus at the site. It is the radicular part of a right upper canine, rather poorly preserved but showing the shallow but broad posterior groove, while the anterior surface is covered in coarsely ridged enamel as in specimens from Makapansgat. The section of the tooth measures 27.7 mm antero-posterior diameter x ca 37 mm dorso-ventral diameter which is considerably larger than canines of Potamochoeroides hypsodon (Table 3).

Genus Metridiochoerus Hopwood, 1926

Type species.- Metridiochoerus andrewsi Hopwood, 1926

Species Metridiochoerus andrewsi Hopwood, 1926

Holotype.- NHMUK M 12805, upper third molar.

Type locality and age.- Homa Mountain (Kanjera) Kenya, Middle Pleistocene.

Bridge Cave (entrance dump): A left upper canine of a large suid (in a block labelled BC 09-18) was found in stratified red breccia at Bridge Cave (Fig. 12). The antero-posterior diameter at the open radicular end is 34 mm and the total height of the tooth from root to apex is 20 cm (measured in a straight line). The distal profile of the tooth is uniformly curved as in canines of Metridiochoerus and Notochoerus, unlike the interrupted curvature of canines of Potamochoeroides. The specimen is somewhat smaller than upper canines of Metridiochoerus andrewsi from Gondolin (Pickford, 2013b) (antero-posterior diameter at radicular end: 50 mm, height from root to apex ca 25 cm), but it probably belongs to this species.

Milo ’A’: An upper canine of a large suid was found at Milo ‘A’. The fossil is still in breccia, so cannot be measured, but its overall dimensions indicate that it represents Metridiochoerus andrewsi. This site also yielded a few molar fragments, one of which is probably an m/2 (MA 3) with a bucco-lingual diameter of 19.5 mm, compatible with Metridiochoerus andrewsi. A left i/2 from Milo ‘A’ (MA 44) measures 7.0 mm mesio-distal diameter by 6.8 mm labio-lingual diameter. It has an elongated wear facet along the distal margin of the crown which has removed most of the surface morphology which renders identification difficult.

Alcelaphine Cave on Bolt’s Farm yielded a moderately worn, incomplete right m/2 (AL 4). Of the anterior lophid, only the distal parts of the protoconid and metaconid are preserved, along with the median accessory cusplet. The entoconid and hypoconid are almost vertical, but the hypoconulid (posterior accessory cusplet) leans distally, producing a prominent overhang in the distal part of the crown. The bucco-lingual diameter is 16.0 mm and the occlusal diameter is estimated to have been about 35 mm, similar to a specimen from Kromdraai ‘A’ (KA 2857 + KA 95-4 (m/2 length x breadth: 36.0 x 15.0 mm) Pickford, 2013a). The tooth is much bigger than m/2s of Potamochoeroides hypsodon (length x breadth: 26.1 x 15.4 mm; Pickford & Gommery, 2016) its crown is taller and the distal overhang is more exaggerated than in the latter species. AL 4 is much larger than corresponding teeth of Phacochoerus modestus (m/2 length ca 25 mm; Pickford, 2013a).

Genus Phacochoerus Cuvier 1826

Type species.- Aper aethiopicus Pallas 1767

Species Phacochoerus modestus ( Van Hoepen & Van Hoepen, 1932)

Bridge Cave Panthera Spot yielded associated right talus and cuboid, and an associated set of left metatarsals and pedal phalanges (Tables 5-6; Figs 13-16). The material is well-preserved and warrants close examination. The Mt III is similar in dimensions and overall aspect to AC 360 from Aves Cave I, the main difference being its slightly more slender appearance, partly due to the fact that the proximal end is missing a sliver of bone from the medial side. It is also close in dimensions and morphology to extant Potamochoerus porcus and Phacochoerus aethiopicus. Detailed comparisons of the fossils were made with the corresponding bones of Potamochoeroides hypsodon, Phacochoerus aethiopicus and Potamochoerus porcus. All three of these taxa are approximately the same size, but the detailed layout of the articular facets is different. The comparisons revealed closer morphological similarities between the fossils and bones of the Desert Wart Hog rather than with the other two taxa. The specimens are therefore attributed to Phacochoerus modestus.

The talus (BC 985) is slightly smaller than the corresponding bone in Potamochoeroides hypsodon from Aves Cave I, and the internal ridge on the margin of the calcanear facet has a greater offset from the sustentacular surface. The angulation between the cuboid and navicular facets is sharper than in Potamochoeroides and the angulation between the proximal and distal articulations is less marked. All these features indicate that the Bridge Cave suid was more cursorially adapted than the suid from Aves Cave, in much the same way that the Wart Hog is a faster runner than the Bush Pig (personal observations MP).

The right cuboid from Bridge Cave Panthera Spot (BC 884) is robust with a well-developed calcanear facet. The proximal processes on the cranial and caudal ends of the talar facet are strong, the cranial one in particular, translating into a well-stabilised ankle joint. It fits well with the talus described immediately above, and probably represents the same individual.

The Mt II has a medio-laterally slender diaphysis which is dorso-plantarly shallow at the proximal end, deepening distally and curving gently medially towards the distal epiphysis. The proximal epiphysis is small with a clear facet for the Mt III, a minute proximal facet, and a lateral facet for the middle cuneiform (Flower, 1876; Plug, 2014). The distal epiphysis is swollen, with a prominent internal ridge on the plantar side which blends into the dorsal surface of the articulation for the first phalanx. The medial part of the articular surface flares medially but is low dorsally.

The Mt III has a large concavo-convex facet for the external cuneiform, and a sloping medial facet which contacts the middle cuneiform. The medial part which articulates with the Mt IV is broken off which gives the impression that the bone is more slender than AC 360. There is a clear facet for the Mt II which is contiguous with the medial facet but is oriented vertically to it. The plantar process is prominent and has a facet where it contacts the plantar process of the Mt IV. The diaphysis is robust, flattened internally where it lies against the Mt IV, rounded medially and generally broadening gently towards the distal epiphysis. The distal articular surface has a prominent central ridge which extends from the plantar to the dorsal sides of the bone, but not extending onto the diaphysis. The part of the distal articulation close to the Mt IV is more distally positioned than the part which is close to the Mt II, as is usual in suoid metatarsals.

The morphology of the Mt IV is typically suid, with an almost flat proximal facet for the cuboid, and a well-developed plantar process which articulates with the plantar process of the Mt III (Flower, 1876, fig. 120). At the proximal end of the diaphysis on the internal side (which contacts the Mt III) there is a low ridge-like process which slots into a depression in the diaphysis of the Mt III. There is also a small facet laterally for the Mt V. The diaphysis and distal epiphysis are basically mirror images of those of the Mt III. The Mt IV is slightly longer than the Mt III as is usual in suids.

The augmented sample of suids from the Bolt’s Farm Palaeokarst System, improves the basis for understanding the range of metric variation in the species and thereby permits a more informed comparison between the suids from this site and those from Makapansgat, which is the type locality of Potamochoeroides hypsodon (Fig. 17). Potamochoeroides shawi, also defined at Makapansgat, is considered to be a synonym of Potamochoeroides hypsodon (Pickford & Gommery, 2016). One of the outcomes of this comparison is that an upper canine (AC 96) from Aves Cave I previously attributed to Potamochoeroides is transferred to Notochoerus capensis and a lower p/3 (AC 258) previously identified as Notochoerus capensis is transferred to Potamochoeroides hypsodon. A second outcome of this revision is that a talus from Makapansgat previously interpreted to belong to Notochoerus capensis by Pickford & Gommery (2016) is far too large to belong to it, but is more likely to represent the huge tetraconodont suid Gerontochoerus scotti.

Three species of suids are present at Makapansgat: Potamochoeroides hypsodon, Notochoerus capensis and Gerontochoerus scotti. There are four species of suid at Bolt’s Farm, but not all from the same time span: Potamochoeroides hypsodon and Notochoerus capensis from Aves Cave I, and Metridiochoerus andrewsi and Phacochoerus modestus from younger deposits on the farm. The lower third molar of Potamochoeroides hypsodon from Bolt’s Farm (Cooke, 1993) appears to be narrower than the equivalent teeth from Makapansgat, but this is undoubtedly due to the fact that the unerupted m/3 in the mandible is incompletely formed, and would have been 4-5 mm broader if growth had been complete.

From this data it is inferred that the fossiliferous breccias at Makapansgat and Aves Cave I are close to each other in terms of geological age, but are older than the deposits that yield Metridiochoerus andrewsi, a species which is not reliably recorded from Makapansgat.

Kullmer (2008) described fossil suids from Malawi attributing them to six taxa - Notochoerus jaegeri, Notochoerus euilus, Notochoerus scotti, Metridiochoerus andrewsi/compactus, Potamochoerus porcus and Phacochoerus aethiopicus, the first four of which would be called « giant » according to the criteria of Broom (1925) and are thus of interest to compare with the large fossils from Bolt’s Farm.

The data presented in Fig. 17 suggests that Notochoerus capensis may be present in Malawi, along with Gerontochoerus scotti and a species of Metridiochoerus, probably large specimens of Metridiochoerus andrewsi as inferred by Kullmer (2008). In addition, some incomplete specimens (not plotted) can be attributed to Potamochoeroides hypsodon. The upper third molar from Malawi attributed to Nyanzachoerus jaegeri plots close to the range of variation of Notochoerus capensis (black line). The data suggests that the Bolt’s Farm succession of breccias spans a similar time period to that in the Lake Malawi Basin (Fig. 18).

The sole suid specimen from « Rhino Site » at Ekuma, in the Etosha National Park, Namibia (Pickford et al. 2016) indicates the presence of Gerontochoerus scotti, on the basis of which a Middle Pliocene correlation is proposed for the deposits. The tooth is too long to be attributed to Notochoerus capensis (Fig. 18).

Brad Pit ‘A’: Parts of an upper canine of a large suid (BPA 152, BPA 79+13) were collected at Brad Pit ‘A’. The dorso-ventral diameter measured some distance from the radicular end is 29 mm and the antero-posterior diameter is 45.1 mm. This specimen is compatible in cross-sectional shape and dimensions to material of Gerontochoerus koobiforaensis. This specimen and other suid fossils from the site will be published in a separate paper.

The word « giant » applied to the holotype of Notochoerus capensis Broom, 1925, has proven over the long term to be a source of confusion. When first described, the specimen from the Vaal River Gravels at Longlands, South Africa, was the largest known fossil suid tooth from the continent (indeed the World) with a length of ca 72-81 mm (minimum and maximum estimates made by Van der Made, 1998, depending on how the broken fossil is reconstructed) being much longer than upper third molars of the extant Giant Forest Hog (Hylochoerus meinertzhageni Thomas, 1904) which range in length from 38-48 mm (Figs 19-20). As such the adjective was appropriate when first applied to the species, but discoveries of much bigger material rapidly followed with teeth over 100 mm long reported in the literature (Hopwood, 1926; Broom, 1928; Van Hoepen & Van Hoepen, 1932; Shaw, 1938; Dietrich, 1942; Arambourg, 1943, 1947; Leakey 1943, 1958; Ewer, 1958a; Harris, 1983; Harris & White, 1979; Harris et al. 1988; Bender, 1990, 1992; Cooke, 1993; Van der Made, 1998; Kulmer, 2008). These reports soon rendered the word ‘giant’ somewhat inappropriate when applied to the specimen from Longlands. The writings of several authors, starting with Shaw (1938) reveal that they were so influenced by Broom’s (1925) usage of the word « giant » that they erroneously attributed much larger teeth to the species capensis which adversely influenced their systematic and taxonomic interpretations. In the full panoply of suids now known from the continent, the Longlands tooth would today qualify for the word « large » rather than « giant ». This confusion is ongoing, partly because the very large suid specimens from Makapansgat which belong to Gerontochoerus scotti have usually been attributed to Notochoerus capensis (cf Cooke & Wilkinson, 1978; Harris & White, 1979, fig. 48, 49; Bender, 1990, 1992) which in turn has influenced identifications of very large teeth from sites in Malawi (Kullmer, 2008) and Namibia (Pickford et al. 2016).

The suid tooth from Longlands described by Broom (1925) would have had an original length of ca 72 mm (see Van der Made, 1998, for a discussion about the possible original dimensions of the tooth) which pales when it is compared with the dimensions of teeth of the species Gerontochoerus scotti which has lower third molars nearly 72% longer (114.8 mm long for the holotype of Notochoerus harrisi Van der Made, 1998, a synonym of Gerontochoerus scotti, Leakey, 1958) and 116.5 mm long for a specimen from Makapansgat (ESI M 2077). A tooth from Malawi is 110.2 mm long (Kullmer, 2008) and one from Namibia is over 97 mm long (Pickford et al. 2016).

Over the years since the initial description of Notochoerus capensis, more than 230 suid third molars have been found in African Plio-Pleistocene deposits which are equal in length and/or breadth to, or are longer or broader than, the type specimen, attributed by their describers to a bewildering variety of taxa, many of which are synonyms (Harris & White, 1979) (Nyanzachoerus jaegeri Coppens 1971, Nyanzachoerus pattersoni Cooke & Ewer 1972, Gerontochoerus eulius (Hopwood, 1926), Gerontochoerus koobiforaensis Pickford 2013c, Gerontochoerus scotti Leakey 1943, Gerontochoerus harrisi (Van der Made, 1998), Kolpochoerus heseloni (Leakey, 1943), Notochoerus paiceae Broom, 1931, Kolpochoerus sinuosus Van Hoepen & Van Hoepen, 1932, Kolpochoerus olduvaiensis (Leakey, 1942), Metridiochoerus andrewsi (Hopwood, 1926), Metridiochoerus hopwoodi Leakey, 1958, Metridiochoerus meadowsi (Broom, 1928), Stylochoerus compactus Van Hoepen & Van Hoepen, 1932, Synaptochoerus hieroglyphicus Van Hoepen & Van Hoepen, 1932, Afrochoerus nicoli Leakey, 1958, Notochoerus serengetensis Dietrich 1942, Pronotochoerus nyanzae Leakey, 1958, Orthostonyx brachyops Leakey, 1958, among others (Cooke, 1949)). Bivariate plots of the length and breadth of these teeth suggest that significant numbers of specimens identified in the literature as lower third molars are likely to be upper third molars, but because few of the specimens have been illustrated it is not possible to make informed revisions without access to the original fossils. It is also likely that different ways of measuring teeth may be contributing to the uncertainty.

The mental gymnastics associated with the discussion of large African hypsodont suids, is evident in their taxonomic history, as shown by frequent name changes, shuffling of specimens between taxa, problems with determining the side or whether the teeth were upper or lower and so on (Pickford, 2013c; Geraads & Bobe, 2017). This uncertainty influences down-line analyses which flow from the basic data, such as biochronology, meaning that a thorough and general revision of these suids is required.

Ewer (1958a) listed 18 fossils from Makapansgat (plus 8 doubtful specimens) which she attributed to Notochoerus (Gerontochoerus) euilus and wrote «The specimens about to be described so closely resemble the Omo material that there can be little doubt that they belong to the same species. A comparison with the type specimen of Notochoerus capensis shows that the differences are considerable - so great indeed that they appear to warrant more than specific separation. The Makapan material is therefore referred to Notochoerus euilus, and the name Gerontochoerus has been revived as a subgenus of Notochoerus to emphasise the differences». In fact, the 26 fossils from Makapansgat attributed to this species by Ewer (1958a) comprise two taxa, one a suine (Notochoerus capensis), the other a tetraconodont (Gerontochoerus scotti). Despite the observations made by Ewer (1958) concerning the major differences between fossils attributed by her to Notochoerus (Gerontochoerus) euilus and the holotype of Notochoerus capensis, subsequent authors (Cooke & Wilkinson 1978; Harris & White, 1979; Cooke, 2005) have amalgamated them into a single taxon, Notochoerus capensis, thereby producing a chimaera.

Bender (1990) did not enter into details concerning problems related to convergence and parallelism in the Makapansgat giant suids when he wrote «Notochoerus from the Limeworks had previously been classified as Notochoerus capensis by a number of researchers ( Cooke and Wilkinson 1978; Harris and White 1979). The classification was based on the twenty one specimens studied in the B.P.I. collections. Due to its fragmentary nature, it is not possible to revise the taxonomic position of the Notochoerus collection, the few comparisons made with East African specimens indicated a similarity with Notochoerus capensis. On the basis of the antero-posterior length of the lateral wall of the lingual face in M3/, it appears that Notochoerus capensis is distinct from Notochoerus euilus and Notochoerus scotti ( Cooke and Wilkinson 1978). Despite the difference in the M3/s of Notochoerus capensis on the one hand and of Notochoerus euilus and Notochoerus scotti, on the other, he continued «Cooke and Wilkinson (1978) and Harris and White (1979) constructed very similar phylogenies for Notochoerus, in which it is derived from Nyanzachoerus. Notochoerus euilus is considered generally more primitive and ancestral to Notochoerus capensis, which in turn is regarded as ancestral to Notochoerus scotti».

All the suid upper third molars from Bolt’s Farm with the exception of that of Notochoerus capensis, plot within the « grazers » oval (Fig. 19) well away from the « omnivores » oval. From this it is inferred that the vegetation in the Gauteng region during the Plio-Pleistocene was not suitable for omnivores (Potamochoerus porcus for example) but that grass supplies were sufficiently reliable for « grazers » to survive and thrive. From the dental morphology (hypsodonty, strongly infolded enamel outlines, presence of cementum on the molars) it is inferred that all four suid taxa known from Bolt’s Farm (Potamochoeroides hypsodon, Notochoerus capensis, Phacochoerus modestus and Metridiochoerus andrewsi) were grazers, a possibility borne out by stable isotope studies (Harris & Cerling, 2002) but in addition, it seems clear that the geologically younger taxa (Early Pleistocene Phacochoerus modestus and Metridiochoerus andrewsi) were dentally better adapted to grazing than were the earlier taxa (Pliocene Potamochoeroides hypsodon and Notochoerus capensis) which could have been « mixed feeders » at least on a seasonal basis. Examination of the lower third molars reveals a similar pattern (Fig. 20).

In African Plio-Pleistocene strata, suids are well-represented, and have proven to be useful for coarse-focus biostratigraphy, not only in East Africa where many records have been calibrated by radio-isotopic dating of subjacent volcanic deposits (Cooke & Maglio, 1972) but also in Southern African karst-related deposits (Harris & White, 1979) where there are no volcanic deposits associated with the strata.

Fossil suids are represented in many of the karstic infillings in the Gauteng Cradle of Humankind, including the Bolt’s Farm Palaeokarst System. This paper documents the fossil suids from several karst deposits in the latter system. There were three phases of deposition in which suids have been found, the earliest one corresponding to the Middle Pliocene at Brad Pit ‘A’ (ca 3.7 Ma), a second late Pliocene assemblage aged ca 2.6-3.0 Ma (Aves Cave I, Bolt’s Pit 14) and a third one correlating to the Pleistocene, ca 1.8-2.0 Ma (Milo ‘A’, Alcelaphine Cave and Bridge Cave). Older deposits at Waypoint 160 (ca 4.5 Ma) have not yet yielded suids.

New fossil suid material from Aves Cave I comprises several juvenile individuals, which collectively reveal that the deciduous and permanent dental formula of Potamochoeroides hypsodon was close to those of Dasychoerus verrucosus (Müller & Schlegel, 1845) (the extant Warty Pig) and Potamochoerus porcus (Linnaeus, 1758) (the extant Bush Pig), but different from that of Phacochoerus aethiopicus (Pallas, 1767) (the Wart Hog) which has suppressed two of the upper incisors. Unfortunately, the deciduous dentition of derived species of Metridiochoerus such as M. andrewsi, is poorly represented in the fossil record so it is not possible to make detailed comparisons. Recall that Harris & White (1979) considered that the species here referred to as Potamochoeroides hypsodon, was an early form of Metridiochoerus andrewsi lineage (in their scheme, Phase I of a 3-phase evolutionary sequence), but Pickford (2013a) showed that, even though they are possibly related forms, the skull morphology, the form and penetration of the lower canine within the mandible and other features indicated that Potamochoeroides and Metridiochoerus are better classified in distinct genera (see below).

The prevalence of juveniles at Aves Cave I, even of unweaned infants suggest that Potamochoeroides may have been breeding near the sites, perhaps even sheltering within the cave entrances, but they could also represent the remains of prey items carried into the caves by predators.

A well preserved distal radius and ulna from Aves Cave I indicates that Potamochoeroides hypsodon was probably a carpograde suid, like the Wart Hog and Metridiochoerus andrewsi (Pickford, 2013b). In a previous study (Pickford & Gommery, 2016) the only available distal ulna was crushed and its bone surface was flakey, which obscured details of the functional features related to carpogrady, on which basis it was erroneously deduced that the genus probably did not practice this particular form of locomotion. That can now be rectified, and it would appear that carpogrady was a feature of the Middle Pliocene suid Potamochoeroides hypsodon. Two distal radii from Makapansgat (BPI M 10883, M 4804) indicate the same thing.

It was hypothesised by White & Harris (1977) that Potamochoeroides hypsodon was a basal member (stage I) of a tripartite Metridiochoerus andrewsi lineage but Pickford (2012) resurrected the genus Potamochoeroides because of significant differences in cranio-dental anatomy between these suids. The distinction between these genera is underlined by the new discoveries at Aves Cave I and by study of the Gondolin suids (Pickford, 2013b). A major difference between these genera concerns the lower canines: in Potamochoeroides, the radicular part of the tooth extends beneath the premolars, well into the horizontal ramus, as in Dasychoerus, for example, whereas in Metridiochoerus the lower canine root terminates in the symphyseal area, and does not extend distally into the horizontal ramus. Furthermore, the canines in advanced species of Metridiochoerus such as M. andrewsi are considerably larger than those of Potamochoeroides hypsodon and they extend only a short way into the mandible, their radicular ends not extending beyond the symphysis. As a consequence, the mandibular canal in Metridiochoerus is voluminous and extends as far anteriorly as the rear of the symphysis, differing from the condition in Potamochoeroides in which the mandibular canal is narrow in the zone of the premolars and diastemata. In addition, the lower canines of Metridiochoerus compactus are endowed with a core of osteodentine (Harris & White, 1979), whereas those of Potamochoeroides hypsodon are not. Nevertheless, these two extinct genera appear to be related to each other (Pickford, 2013b) as well as to the Wart Hog (Phacochoerus).

The presence of Notochoerus capensis at Aves Cave I seems to be assured on the basis of the M3/ and a p/4 described by Pickford & Gommery (2016). The attribution to Notochoerus capensis of a large upper canine from Aves Cave I, that was previously identified as Potamochoerides hypsodon is done on the basis that there appear to be only two taxa of suids at the site, the smaller Potamochoeroides hypsodon, and the larger Notochoerus capensis. If this reattribution is correct, then it suggests that Notochoerus may be closely related to Potamochoeroides than to Tetraconodontinae. However, until more complete remains of Notochoerus capensis are recovered it will be difficult to explore the possibility that Notochoerus and Potamochoeroides could be synonyms. As it stands, the morphology of the Aves Cave I suid canine, the M3/ and the p/4 and attributed to Notochoerus capensis recall those of Potamochoeroides hypsodon in several respects, in its known parts the former being an upscaled version of the latter, and both taxa differing in similar ways from Metridiochoerus andrewsi.