Although the suid genus Kolpochoerus is well known from the Plio-Pleistocene of Africa, the evolutionary history of one of its constituent species, K. majus, remained obscure until substantial fossil evidence accumulated during the last 20 years, largely from sites in Ethiopia. Here, we describe Kolpochoerus phillipi sp. nov., based on a fairly complete skull and the remains of additional individuals from ∼2.5 Ma deposits at Matabaietu, in the Middle Awash study area of Ethiopia. Based on a phylogenetic analysis, we suggest that K. phillipi sp. nov. belongs to a clade of “bunolophodont suines” including K. majus and the extant giant forest hog Hylochoerus meinertzhageni. Within this clade, K. phillipi sp. nov. likely represents a potential ancestor of K. majus, based on its morphology and stratigraphic position.
In Sub-Saharan Africa, wild pigs (family Suidae) are represented by five extant omnivorous and herbivorous species, which have been the focus of research dealing mostly with biodiversity conservation (Oliver 1993; d'Huart and Yohannes 1995; Randi et al. 2002; Grubb and d'Huart 2010). Although recent molecular studies have shown extant African suids to form a monophyletic group, with Hylochoerus and Phacochoerus forming a clade to the exclusion of Potamochoerus (Gongora et al. 2011), relatively little is known about their origin and evolutionary history, owing to their patchy fossil record and the uncertain taxonomic and phylogenetic status of most of their fossil relatives. Potamochoerus in particular has traditionally been used as a wastebasket taxon for many archaic-looking specimens characterized by simple, short, brachyodont and bunodont teeth, resulting in general confusion regarding the antiquity of this genus (White 1995; Bishop 2010). The fossil record of Hylochoerus is also very poor, with the only securely-dated and well-identified fossils coming from Member III of the Kibish Formation, ca. 0.1 Ma (Assefa et al. 2008).
Our current knowledge of the origins and evolution of extant African suids is highly reliant on abundant Plio-Pleistocene fossils. Specifically, Hylochoerus has been suggested to descend from the extinct genus Kolpochoerus (Harris and White 1979; Geraads 2004), whereas Potamochoerus might either be the sister group of Kolpochoerus (Cooke 1997; Geraads 2004), or a descendant of the Pliocene K. afarensis (White 1995). The genus Kolpochoerus itself is most famous for the rich fossil record and biostratigraphic value of its best represented lineage, K. limnetes/K. olduvaiensis (Cooke 1978b; Harris and White 1979). Most, but not all, authors recognize the following species: Kolpochoerus afarensis, K. cookei, K. deheinzelini, K. limnetes (also called K. heseloni following Cooke 1997), K. majus, K. olduvaiensis, K. paiceae, and K. phacochoeroides (White and Harris 1977; Cooke and Wilkinson 1978; Harris and White 1979; Hendey and Cooke 1985; White 1995; Brunet and White 2001; Geraads 2004; Geraads et al. 2004a). In addition, Haile-Selassie and Simpson (2013) recently described K. millensis, a putative chronospecies intermediate in morphology and time between K. deheinzelini and K. afarensis.
Amongst the recognized species of Kolpochoerus, K. afarensis, the most abundantly represented early member of the genus, is considered the ancestor of the younger K. limnetes (White 1995), and it is sometimes assigned to the genus Potamochoerus (Cooke 1997; Bishop 2010). Kolpochoerus olduvaiensis is a derived chronospecies of K. limnetes (Gilbert 2008), and is sometimes included within the latter species (Harris and White 1979). This is also the case for K. paiceae, an apparent South African endemic similar to K. olduvaiensis (Harris and White 1979). By contrast, K. phacochoeroides is endemic to northern Africa (Geraads 1993, 2004), and sometimes interpreted as conspecific with either K. limnetes from eastern Africa (e.g., Pickford 1994), or K. olduvaiensis (e.g., White 1995).
Kolpochoerus majus was first described on the basis of a left mandibular fragment preserving p3−m1 and the alveolus for p2 from Bed IV of Olduvai Gorge (Hopwood 1934), and is characterized by brachyodont, mesiodistally short and simple third molars with a bunolophodont wear pattern resulting from mesiodistally compressed lateral pillars (Leakey 1942, 1958; Harris and White 1979). The species has been reported from various Pleistocene localities dating from ca. 1.9 Ma to ca. 0.5 Ma, including the Daka Member of the Bouri Formation and the Upper Bodo beds (Middle Awash), Member L of the Shungura Formation, the Busidima Formation (Asbole), the Melka Kunturé Formation (Garba IV), and the Konso, Olduvai, Kapthurin, Nachukui, Olorgesailie, and Nyabusosi formations (Hopwood 1934; Leakey 1942; Harris and White 1979; Harris et al. 1988; Pickford 1994; White 1995; Suwa et al. 2003; Geraads et al. 2004a, b). Despite this abundance of sites, the evolutionary history of the K. majus lineage has remained largely unknown, owing to a lack of adequate material predating the Pleistocene specimens-although some authors have hypothesized that the species may have evolved from an early form of K. limnetes ∼2.5 Ma, before the latter acquired a more complex dental morphology (White 1995; Geraads et al. 2004a).
List of the specimens described as belonging to Kolpochoerus phillipi sp. nov. from the late Pliocene “W” Formation of the Middle Awash. For the specimens collected by the RVRME, the equivalent localities of the Middle Awash research project are indicated.
Here, we describe fossil material that is dated around 2.5 Ma and comes from the Matabaietu area in the Middle Awash study area of Ethiopia, and assign them to a new, archaic species of Kolpochoerus. The latter displays strong dental similarities to K. majus, and may help to shed light on the transition from K. afarensis (or from an early K. limnetes morph) to K. majus.
Institutional abbreviations.—ARCCH, Authority for Research and Conservation of Cultural Heritage, Addis Ababa, Ethiopia; KNM, Kenya National Museums, Nairobi, Kenya; MNHN-ZM-AC, Collection Zoologie et Anatomie Comparée, Muséum National d'Histoire Naturelle, Paris, France; MRAC, Muséum Royal d'Afrique Centrale, Tervuren, Belgium; NHMUK, Natural History Museum, London, United Kingdom.
Other abbreviations.—A.L., “Afar Locality”, prefix of locality numbers in the Hadar Formation, Ethiopia; ER, “East Rudolf”, prefix of fossil specimens from the Koobi Fora Formation, Kenya; KL, “Kalb Locality”, prefix of locality numbers defined by the RVRME (Rift Valley Research Mission in Ethiopia) in the Middle Awash study area, Ethiopia; L, Omo Shungura Formation, Ethiopia, American locality number; MAT-VP-, Middle Awash vertebrate paleontological locality Matabaietu, Ethiopia; OMO, Omo Shungura Formation, Ethiopia, French locality number; WIL-VP-, Middle Awash vertebrate paleontological locality Wilti Dora, Ethiopia.
The new material of Kolpochoerus described here (Table 1) was recovered by the RVRME (prior to 1980), and subsequently the Middle Awash research project from the area of Matabaietu in the Middle Awash study area, Afar Depression, northern Ethiopia. Sediments in the local catchment range from the middle Pliocene to the Middle Pleistocene, with all of the present material coming from the late Pliocene “W” Formation of the Middle Awash (White et al. 1993), estimated to be roughly equivalent to the Hatayae Member of the Bouri Formation exposed on the western side of the study area (de Heinzelin et al. 1999). The age of these deposits has been estimated at ∼2.5 Ma, based notably on biochronological studies of fossil bovids (Kalb et al. 1982b; Vrba 1997; White et al. 2005; Bibi 2011). However, age assessments of this part of the Middle Awash are still a work in progress.
Material and methods
We compared our new material to several species of Kolpochoerus, as well as specimens of extant African suids (Potamochoerus and Hylochoerus) held at ARCCH, KNM, MNHN-ZM-AC, MRAC, and NHMUK. Comparisons with the extant taxa in particular allowed us to evaluate the presence of both sexual dimorphism and geographic variation in the fossil record, and assess the effects of the latter on systematics.
Teeth were measured using a digital caliper (precision of 0.1 mm), following Harris and White (1979). To quantify differences in skull morphology, we developed a new measurement protocol (Table 2), adapted in part from the literature (Cooke and Ewer 1972; Driesch 1976). Using a 3D-digitizer (microscribe G2X, Immersion Corporation), each measurement was taken between two anatomical landmarks. Whenever both sides were preserved, measurements were averaged. To estimate the degree of repeatability of our measurements, the latter were performed ten times on two skulls (one of Hylochoerus and one of Potamochoerus), which resulted in per-measurement standard deviations averaging close to 1%. All data are reported in Tables 3 and 4, as well as the Supplementary Online Material (SOM available at http://app.pan.pl/SOM/app60-Souron_etal_SOM.pdf).
In order to determine the phylogenetic position of the new species described here, we performed a parsimony-based phylogenetic analysis in TNT (Goloboff et al. 2008). Our matrix (Appendix 1) included 22 characters (Appendices 1–4), coded for nine taxa. Some of the characters (0, 4, 7, 10–12, 18, and 21) were adopted from Geraads (2004). All characters were defined and coded based on direct observation (no taxa were coded from the literature), taking into account sexual dimorphism, geographic, and temporal variation. We focused our analysis on species of Kolpochoerus and two extant genera of African suids, Potamochoerus and Hylochoerus. Within Kolpochoerus, we restrained our analysis to species from eastern Africa for which sufficient complete cranio-mandibular material is available. A recent phylogenetic analysis revealed the Eurasian genus Sus to be the sister group of African suids, and Babyrousa to be the basalmost taxon within the subfamily Suinae (Gongora et al. 2011). We therefore used Babyrousa as outgroup, and included Sus scrofa in our analysis. We analyzed our data using an exhaustive search with all characters left unordered and unweighted. Bremer indices were calculated for each node to estimate branch support.
Cranio-mandibular measurements used in the paper, based partly on Cooke and Ewer (1972) and Driesch (1976). Table 3. Cranium measurements of the holotype skull of Kolpochoerus phillipi sp. nov. (MAT-VP-1/5) compared to various suids. Measurements to nearest mm. Measurements are displayed as follows: mean ± SD (n) (min; max); e, estimate.
Cranium measurements of the holotype skull of Kolpochoerus phillipi sp. nov. (MAT-VP-1/5) compared to various suids. Measurements to nearest mm. Measurements are displayed as follows: mean ± SD (n) (min; max); e, estimate.
Class Mammalia Linnaeus, 1758
Order Cetartiodactyla Montgelard, Catzeflis, and Douzery, 1997
Family Suidae Gray, 1821
Subfamily Suinae Gray, 1821
Genus Kolpochoerus van Hoepen and van Hoepen, 1932
Type species: Kolpochoerus sinuosus van Hoepen and van Hoepen, 1932. Holotype specimen from the Pleistocene of Uitzoek, near Cornelia, South Africa.
Etymology: Named in honor of the late and great South African paleoanthropologist Phillip V. Tobias for his support of the Middle Awash research project, and in recognition of his work on African fossil hominids never losing sight of the importance of associated vertebrate faunas and their environmental context (White 2012).
Type material: Holotype: MAT-VP-1/5 (housed at the ARCCH), partial cranium with alveoli for I1-I3, isolated I1 and I2, partial C, and P1-M3; associated mandible with i1, i2, alveoli for i3, partial c, and p2-m3) (Figs. 1–3). Paratype: MAT-VP-5/1 (housed at the ARCCH), left mandible fragment with p4, m2, m3 and alveoli for m1 (Fig. 3B). Type locality: Matabaietu Vertebrate Paleontology Locality One, Middle Awash study area, Afar Depression, Ethiopia.
Type horizon: “W” Formation, Unit 9 sandstone (in situ), Plio-Pleistocene, ca. 2.5 Ma.
Material.—Seven isolated third molars, here referred to Kolpochoerus cf. phillipi sp. nov. (Table 1, Fig. 3C-I): KL 234-1, KL 234-3-5, MAT-VP-5/2, Matabaietu area, Middle Awash study area, Ethiopia: KL 10-3, KL 24-1 b, KL 39-4, KL 73-1, Wilti Dora area, Middle Awash region, Ethiopia; “W” Formation, Plio-Pleistocene, ca. 2.5 Ma. Diagnosis.—Differs from all species of Potamochoerus and Hylochoerus, but resembles Kolpochoerus, in the presence of a ventrally extended and laterally expanded zygomatic arch, as well as a marked angle separating the rostrum and the elevated neurocranium in lateral view. Differs from all species of Kolpochoerus except K. majus in the presence of mesiodistally compressed lateral pillars of the molars, and zygomatic arches oriented at a 40° angle to the sagittal plane in dorsal view in males; differs from K. deheinzelini, K. millensis, and K. cookei in having much larger and more complex teeth; differs from K. afarensis in having a dorsoventrally thickened mandibular symphysis, a more swollen mandibular body, less developed lower incisors, a less acute lower incisor arcade, more developed upper and lower canines, a less constricted parietal roof, and an anteroposteriorly elongated supra-canine flange (with the latter only being present in males); differs from K. limnetes in having a relatively shorter cranium, a rostrum retaining a constant width throughout, a narrower mandibular symphysis, and less symmetric and less complex m3 talonids; differs from K. olduvaiensis, K. paiceae, and K. phacochoeroides in having much shorter, lower and simpler third molars, unreduced premolars, an orbit located below the dorsal surface of the frontal, and a well-developed supra-canine flange in males; differs from K. majus in having a narrower mandibular symphysis, a more constricted parietal, near-parallel cheek tooth rows, and a strongly developed supra-canine flange in males.
Description.—The holotype cranium preserves the complete rostrum, most of the parietal roof, and the incomplete left zygomatic arch (Fig. 1), but lacks the entire occipital bone and the ventral portion of the braincase. An isolated piece of the basicranium was found, but is poorly preserved and cannot be attached to the rest of the skull. Judging from the eruption of all of the teeth and the presence of heavy wear on the first two molars, the specimen represents an adult. It can furthermore be identified as a male, based on the presence of well-developed lower and upper canines, a strong supra-canine flange, a laterally inflated zygomatic arch, a strongly swollen mandibular body, and a long mandibular symphysis. Sexual dimorphism in canine size is well documented in the Suidae (Harris and White 1979; Groves 1981; Made 1991; Cooke 2007), while the lateral expansion of the zygomatic arch in males occurs in several species of living and fossil African suids, including Hylochoerus, Kolpochoerus, and Metridiochoerus (Harris and White 1979; Geraads 2004). Similarly, a dorsoventrally developed, crest-like supra-canine flange characterizes males of Potamochoerus and some species of Sus, such as S. celebensis (Hardjasasmita 1987), and is supposedly linked to the protection of the rhinarium muscles in species that fight using snout boxing (Kingdon 1979).
In terms of overall size, the skull is close to the average size of Hylochoerus, slightly smaller than female specimens of K. majus, and much smaller than K. limnetes and the only known male specimen of K. majus (BOU-VP-25/107; Gilbert 2008; Table 3). As in other species of Kolpochoerus, the cranium is relatively tall compared to Potamochoerus and Hylochoerus, and further resembles K. majus and early specimens of K. limnetes, but not K. olduvaiensis, in being relatively short anteroposteriorly (Fig. 4D). The postorbital constriction is comparable to the average K. limnetes, being weaker than in most specimens of Potamochoerus and K. olduvaiensis, but stronger than in K. majus and Hylochoerus (Fig. 4B, Table 3). The dental rows are almost parallel, unlike in K. majus in which the latter diverge anteriorly (Fig. 4C). The premaxilla is relatively wide transversely and inclined ventrally in lateral view, as in K. majus and K. limnetes. By contrast, the premaxilla is relatively narrow and more horizontal in Potamochoerus and Hylochoerus (Fig. 4A). In cross section, the rostrum is square, with vertical edges as seen in Potamochoerus and early specimens of K. limnetes (Fig. 1C). The transverse width of the rostrum remains constant along its entire length, unlike in K. limnetes in which the rostrum is widest somewhat posterior to the supra-canine flange and tapers anteriorly. As in Potamochoerus and early K. limnetes, the nasals overhang the maxilla in anterior view (Fig. 1C). In contrast to K. olduvaiensis and K. paiceae, the orbit is located below the dorsal surface of the frontal, as in K. afarensis, K. majus, and K. limnetes. In lateral view, the dorsal surfaces of the frontal and parietal are arched dorsally, more so than in K. afarensis, but similar to K. majus and early K. limnetes. The supra-canine flange is elongated anteroposteriorly and developed dorsoventrally, resembling the condition seen in males of early Kolpochoerus limnetes, although in the latter the dorsal edge of the flange is less rounded in lateral view. By contrast, the ventral part of the flange is short anteroposteriorly in males of Potamochoerus, and greatly reduced with only a weak lateral crest and a shallow gutter in K. olduvaiensis, K. majus, and Hylochoerus (Gilbert 2008). The zygomatic arch strongly projects ventrally, and is oriented at a 40° angle to the sagittal plane in dorsal view. In the latter feature, the present material resembles females of K. majus from the Daka Member of the Bouri Formation (ca. 1 Ma; Gilbert 2008) and the site of Asbole (Busidima Formation, 0.8–0.6 Ma; Geraads et al. 2004a), although it differs from males of this species in its slightly more obliquely oriented and less laterally inflated zygomatic arch (see above; Gilbert 2008). By contrast, the angle of the zygomatic arch to the sagittal plane is close to 90°, and the zygomatic knob is usually covered with rugose bone in male specimens of K. afarensis, K. limnetes, and K. olduvaiensis. As in K. afarensis and early K. limnetes, the insertion scars for rostral muscles are relatively well marked in K. phillipi sp. nov., although less so than in Potamochoerus. In Hylochoerus and other species of Kolpochoerus, the muscle scars are reduced.
Upper dentition.—The upper dental formula is 126.96.36.199. The right I1 and I2 were found isolated (Fig. 3A), and resemble those of K. limnetes and K. majus. The upper canine is slightly triangular in cross section, much larger than in Potamochoerus, and much smaller than in males of K. limnetes and K. majus (Table 3). There are no marked grooves as in the canines of male K. limnetes, K. majus, and Hylochoerus, and the ventral side of the tooth displays a strongly ribbed enamel band. As in Hylochoerus, the canine is oriented slightly anterodorsally, at an angle of approximately 45° to the sagittal plane of the cranium (in dorsal view). The total length of the cheek tooth row is similar to that of the largest specimens of Potamochoerus, K. afarensis, and the smallest specimens of K. limnetes and K. majus (Table 3). The diastema between the upper canine and the premolars is relatively long compared to Potamochoerus, but within the range of K. majus, and relatively shorter than in the average K. limnetes and Hylochoerus (Table 3). The upper premolars are similar in proportion and size (Table 5) to those of the smallest specimens of K. limnetes and K. majus, but relatively wider, with a higher width/length ratio, than those of K. afarensis (especially P3). P1 is small, and separated from P2 by a small diastema (Fig. 3A). The ratio between the premolar and molar row lengths is relatively high (C16/C17*100 = 50), similar to Potamochoerus (average C16/C17*100 = 56; n = 63) and K. majus (average C16/C17*100 = 54; n = 12), and relatively higher than in the average K. limnetes (average C16/C17*100 = 44; n = 20). Both M1 and M2 are heavily worn, and fall within the range of K. afarensis, K. limnetes, and K. majus in terms of their size and proportions (Table 5). Finally, M3 comprises two pairs of lateral pillars, followed by a strong terminal pillar and several small ectostyles (Fig. 3A), and resembles the corresponding teeth of K. afarensis, K. majus, and early K. limnetes from members B and C of the Shungura Formation (Fig. 5B). The wear pattern of the lateral pillars is simple, and similar to specimens of K. majus.
Mandible.—The holotype mandible is essentially complete (Fig. 2). As in other species of Kolpochoerus and Hylochoerus, the cheek tooth rows are nearly parallel, in contrast to the more anteriorly convergent condition seen in Potamochoerus. The mandibular symphysis is dorsoventrally thick as in K. majus and K. limnetes, but unlike Potamochoerus and K. afarensis. As in K. majus, the mandibular symphysis is thinnest close to its anterior extremity (Fig. 6D). The symphyseal area is relatively narrow, with its width/length ratio (M4/M3*100 = 75) falling within the range of Potamochoerus (59–85; n = 51) and K. afarensis (77–82; n = 2), while being narrower than in K. majus (93–114; n = 7) and males of K. limnetes (75–90; n = 3) (Fig. 6B, C). In female specimens of K. limnetes (recognized by their small lower canines and short mandibular symphysis), this region is comparatively narrow (ratio of 61–74; n = 3). In K. phillipi sp. nov., the posterior edge of the symphysis is fairly vertical, as in K. majus, but unlike the inclined condition seen in K. limnetes and Potamochoerus (Fig. 6D). The postcanine constriction is strong, although it should be noted that this character is related to sexual dimorphism (the differential development of canines in males and females), and hence highly variable within each species. The incisor arcade is relatively straight, as in Hylochoerus and K. majus (Fig. 6C). It is possible that this character may be highly variable and related to sexual dimorphism as in Potamochoerus, K. afarensis, and K. limnetes, with relatively straight and angled incisor arcades characterizing males and females, respectively.
As in Hylochoerus and K. majus, but unlike in Potamochoerus, the width across both mandibular bodies at the level of the third molars is relatively large compared to the width of the mandible across the mandibular condyles (Table 4). The ascending ramus is relatively short, and similar to that of other species of Kolpochoerus. The mandibular body is relatively high compared to the overall height of the mandible, and marked by a prominent swelling at the level of the molars. The latter feature, typical of Kolpochoerus, is to a lesser degree also present in Potamochoerus (Fig. 6A). The main anterolateral mental foramen is located relatively more dorsally than in Potamochoerus and K. limnetes, and more ventrally than in Hylochoerus and K. majus. In lateral view, the coronoid process terminates below the level of the dorsal border of the mandibular condyle, unlike in Potamochoerus, Hylochoerus, and other species of Kolpochoerus.
Lower dentition.—The mandibular dental formula is 188.8.131.52, with no evidence for p1. As in Kolpochoerus limnetes and K. majus, the lower incisors are reduced compared to K. afarensis and Potamochoerus. The enamel of the cheek teeth is relatively thick, as in K. afarensis, K. majus, early K. limnetes, and Potamochoerus. The lower canine is verrucose (the distal face being smaller than the labial one), as in every species of Kolpochoerus. The canine is relatively wider than in K. afarensis, resulting in a relatively open, triangular cross section as seen in K. majus and K. limnetes; however, the height of the canine (measured from the alveolar rim to the crown apex) is smaller than in the latter two species.
The total length of the cheek tooth row is similar to K. afarensis, somewhat smaller than in K. limnetes and K. majus, and larger than in average specimens of Potamochoerus (Table 4). The diastema between the lower canine and the premolars is relatively short compared to the average K. limnetes and K. majus, and relatively longer than in K. afarensis and Potamochoerus. The premolars are similar in proportion and size to the smallest specimens of K. limnetes and K. majus. The ratio of premolar row to molar row length (M14/M15*100 = 49) resembles that of Potamochoerus (average M14/M15*100 = 52; n = 62), early K. limnetes (for example, L 36–27 from Member D of Shungura Formation has a ratio of 46) and K. majus (average M14/M15*100 = 49; n = 7) (Fig. 5D). Both the m1 and m2 of the holotype mandible are heavily worn, and within the range of K. afarensis, K. limnetes, and K. majus in terms of their size and proportions (Fig. 5C; Table 5). The m2 of the paratype mandible is slightly smaller than that of the holotype, similar in size to the specimens of K. afarensis (Fig. 5C).
Judging from the isolated m3s referred to Kolpochoerus cf. phillipi sp. nov. (Fig. 3C–G), it seems that they were smaller on average, than those of K. majus and early K. limnetes (Member C of the Shungura Formation). The holotype m3 is one of the largest specimens, which is congruent with its identification as a male (Fig. 5A; Table 5), and comprises three pairs of lateral pillars, followed by a small terminal pillar. The labial pillar of the third pair is more developed than the lingual one. In terms of its size and proportions, the m3 of K. phillipi sp. nov. is largely indistinguishable from that of K. afarensis, early K. limnetes (e.g., from members B and C of the Shungura Formation), and K. majus (Fig. 5A). However, the m3 talonid is more complex than in K. afarensis, which is characterized by two pairs of lateral pillars, followed by a single, well-developed terminal pillar. By contrast, the talonid of K. phillipi sp. nov. is less complex than that of contemporary specimens of K. limnetes from Member C of the Shungura Formation (ca. 2.7–2.5 Ma), which have a symmetric third pair of lateral pillars and a well-developed terminal pillar (Fig. 3M, N); however, older representatives of K. limnetes from Member B of the Shungura Formation (ca. 2.9 Ma) resemble K. phillipi sp. nov. in terms of m3 talonid complexity (Fig. 3K, L). In Middle Pleistocene specimens of K. majus, there are three pairs of lateral pillars and a well-developed terminal pillar (Fig. 3P-S). With the exception of the smaller size of the terminal pillar, the holotype m3 thus resembles that of K. majus, having well-separated and mesiodistally compressed lateral pillars and a simple, rounded wear pattern, resulting in a bunolophodont aspect (Fig. 3A). Both of the holotype m3s are too worn to determine whether there are one or two median pillars between the second and third pairs of lateral pillars. Several authors consider this character to be diagnostic of the genus Kolpochoerus (van Hoepen and van Hoepen 1932; Geraads 2004), although it is absent in K. deheinzelini, K. millensis, K. cookei, and usually also K. afarensis. In K. cf. phillipi sp. nov., the median pillar is single, or faintly divided (Fig. 3B–G), whereas there is a double median pillar in the earliest K. limnetes (Fig. 3K, L) and K. majus (Fig. 3P–S).
Stratigraphic and geographic range.—Known only from ca. 2.5 Ma old deposits in the Matabaietu and Wilti Dora areas, Middle Awash study area, Afar Depression, northern Ethiopia.
Mandible measurements of the holotype skull of Kolpochoerus phillipi sp. nov. (MAT-VP-1/5) compared to various suids. Measurements to nearest mm. Measurements are displayed as follows: mean ± SD (n) (min; max).
Dental measurements (in mm) of the holotype skull of Kolpochoerus phillipi sp. nov. (MAT-VP-1/5). Abbreviations: e, estimate; L, length; P, p, premolar; M, m, molar; W, width.
Evolutionary relationships.—The ventrally and laterally expanded zygomatic arches, together with the dorsally expanded neurocranium, the characteristic dental morphology, and the inflated mandibular body firmly place the new species within the genus Kolpochoerus. This assignment was corroborated by the results of our cladistic analysis (Fig. 7), which resulted in a single most parsimonious tree of 35 steps (Consistency index = 0.69; Retention index = 0.75). Twelve characters (3, 5, 6, 9, 11, 12, 14, 16, 17, 19, 20, and 21) were free of homoplasy.
African suids form a monophyletic group (node A) including Potamochoerus, Hylochoerus, and a paraphyletic Kolpochoerus. This clade is supported by three unambiguous synapomorphies: a transverse zygomatic arch in dorsal view in males (ch. 1); a dorsoventrally high supra-canine flange (ch. 11); and a strongly swollen mandibular body (ch.15). A clade comprising Kolpochoerus + Hylochoerus (node B) is also supported by three unambiguous synapomorphies, including: a ventrally extended zygomatic arch in males (ch. 0); a mediolaterally thickened zygomatic arch in males (ch. 2); and the separation of the rostrum and neurocranium by a distinct angle in lateral view (ch. 4). Characters describing the morphology of the premaxilla (ch. 8 and 9) could not be coded for K. afarensis, as the premaxillae were either missing or distorted in all available specimens, including the holotype A.L. 147-10 (Cooke 1978a). However, based on the morphology of the mandibular symphysis, which bears an infradentale located distinctly ventral to the level of the lower cheek teeth, it is likely that the premaxilla of K. afarensis was strongly inclined so as to allow contact between the lower and upper incisors. If confirmed by new fossils, this would make the presence of a relatively wide and ventrally inclined premaxilla (ch. 8 and 9) a potential synapomorphy for Kolpochoerus. In addition, the symphyseal morphology of K. millensis is very similar to that of K. afarensis (Haile-Selassie and Simpson 2013: fig. 4), and might therefore support the inclusion of K. millensis (and, indirectly, its ancestral chronospecies K. deheinzelini) within the genus Kolpochoerus.
The clustering of advanced species of Kolpochoerus (K. limnetes, K. olduvaiensis, K. majus, and K. phillipi sp. nov.) and Hylochoerus to the exclusion of K. afarensis (node C) is supported by two unique, unambiguous synapomorphies: a straight lower incisor arcade in males (ch. 16); and reduced lower incisors with i2 being shorter and oriented obliquely to i1 (ch. 19). The presence of an anteroposteriorly elongated supra-canine flange (ch. 12) is an ambiguous synapomorphy that could support either this clade (node C) or the clade comprising Kolpochoerus + Hylochoerus (node B). This ambiguity stems from the fact that the morphology of the supra-canine flange is unknown in K. afarensis.
Within the clade comprising advanced Kolpochoerus species and Hylochoerus, K. limnetes, and K. olduvaiensis form a sister-group relationship (node D) supported by three unique, unambiguous synapomorphies: a relatively long cranium (ch. 3); a relatively low rostrum (ch. 5); and molars tending to display a flat wear pattern of the lateral pillars (ch. 21). Closely related to this group is a clade comprising the “bunolophodont suines” K. majus, K. phillipi sp. nov., and Hylochoerus, supported by two unambiguous synapomorphies: an obliquely oriented zygomatic arch in dorsal view in males (ch. 1); and molar teeth displaying mesiodistally compressed lateral pillars (ch. 20), resulting in a characteristic bunolophodont appearance (Fig. 3A–J, P–S).
Within the “bunolophodont suines”, Kolpochoerus majus is the sister taxon of the extant Hylochoerus meinertzhageni (node F), to the exclusion of K. phillipi sp. nov. This relationship is supported by four unambiguous synapomorphies: a weakly constricted parietal roof (ch. 6); reduced scars for the attachment of the rhinarium muscles (ch. 7); the absence of a crest-like supra-canine flange (ch. 10); and a dorsoventrally low supra-canine flange (ch. 11). Two characters (14 and 17) describing the mandibular morphology of female specimens could not be coded for K. phillipi sp. nov., and thus represent ambiguous synapomorphies that could support either the clade comprising K. majus + Hylochoerus clade, or the “bunolophodont suines” as a whole.
The only previous cladistic analysis of Kolpochoerus found all of the dentally derived species (K. olduvaiensis, K. paiceae, K. phacochoeroides, and Hylochoerus) to form a clade based on the reduction of the premolars, and the presence of longer and more complex third molars (Geraads 1993, 2004). However, these characters are notorious for being prone to convergence (Harris and White 1979), and appear multiple times during the Plio-Pleistocene in different genera of African suids (Kolpochoerus, Metridiochoerus, Nyanzachoerus/Notochoerus), probably as an adaptation to a more specialized herbivorous diet. We therefore chose not to include characters dealing with the length of the third molars in our analysis. In addition, we found that the reduction of the upper premolars convergently occurs in at least two independent groups (K. limnetes/K. olduvaiensis and Hylochoerus), according to our results.
In our cladogram, Kolpochoerus phillipi sp. nov. is positioned as an early offshoot of the “bunolophodont suines”. However, given their respective ages, their general morphological resemblance (especially as regards the morphology of the molars), the apparent absence of autapomorphies in the new species, and the fact that phyletic change is a well-documented phenomenon in various lineages of Plio-Pleistocene suids (Harris and White 1979), we consider it likely that K. phillipi sp. nov. and K. majus form a chronospecies pair analogous to K. limnetes/K. olduvaiensis, with K. phillipi sp. nov. representing an earlier, ancestral form of K. majus. Clarification of this question will likely have to await the description of abundant material of K. majus from Konso, southern Ethiopia, dated to around 1.9-1.4 Ma (currently under study by Gen Suwa and colleagues). Given the limited fossil record of Hylochoerus and our poor knowledge of its evolutionary history, the potential status of H. meinertzhageni as a highly derived species of Kolpochoerus is intriguing (e.g., Cooke and Wilkinson 1978; Harris and White 1979; Geraads 2004), and we suggest that it may have originated sometime during the Pleistocene from a K. majus-like ancestor. Our ongoing studies of numerous K. majus specimens from the Middle Awash area spanning the time period from 1.0–0.1 Ma may help to resolve this issue.
Biochronology.—Kolpochoerus phillipi sp. nov. closes the morphological gap between the small, brachyodont and bunodont Pliocene K. afarensis and the larger, bunolophodont Middle Pleistocene K. majus. Although the lower third molars of K. phillipi sp. nov. are somewhat shorter and less complex than that of K. majus, the similarities in dental morphology and wear pattern are striking, and draw attention to the potential use of cranio-mandibular anatomy as a biostratigraphic tool within a potential K. phillipi sp. nov. + K. majus lineage. Especially interesting features include the supra-canine flange, the zygomatic arch, and the mandibular symphysis.
The Konso Formation of Ethiopia and the Nyabusosi Formation of Uganda have been dated to around 1.9–1.4 Ma and 1.8 Ma, respectively, and have yielded numerous dental specimens assigned to K. majus (Pickford 1994; Suwa et al. 2003). In terms of their morphology, these specimens are less complex than those of Middle Pleistocene K. majus, and include m3s with three well-developed pairs of lateral pillars, followed by a small terminal pillar. While resembling the condition seen in K. phillipi sp. nov., the terminal pillar appears somewhat more developed than in the latter, and thus possibly represents an intermediate morphological stage. More complete cranio-mandibular material from Konso will enable us to test this hypothesis. Harris et al. (1988) also attributed a single m3 from the upper part of the Lomekwi Member of the Nachukui Formation (dated to around 2.6 Ma) to K. majus. Although its age makes it possible that this specimen may in fact belong to K. phillipi sp. nov., no cranio-mandibular remains were found in association.
Paleoecology.—Suid paleodiets are generally interpreted based on the strong differences between the three extant African genera Potamochoerus, Hylochoerus, and Phacochoerus. Potamochoerus has short, brachyodont and simple third molars, as well as well-developed lower and upper incisors and well-marked rhinarium muscle scars. By contrast, Phacochoerus has very long, complex, and hypsodont third molars with numerous pillars packed within cement, greatly reduced incisors, and reduced rhinarium muscle scars. These characteristics are thought to be adaptations to omnivorous (with a strong rootling behavior) and grazing diets (e.g., Kullmer 1999), respectively. Hylochoerus has moderately hypsodont and relatively complex bunolophodont cheek teeth that are thought to reflect its diet of soft browse (Harris and Cerling 2002), although some studies also indicate a more opportunistic diet (d'Huart 1978; Viehl 2003; Cerling and Viehl 2004).
Considered within this framework, Kolpochoerus phillipi sp. nov. would definitely fall within the Potamochoerus-like omnivorous suids, based on its well-developed rhinarium muscle scars, square cross-section of the rostrum, and relatively unspecialized dentition. Kolpochoerus majus also retains an unspecialized dentition (although there is a trend towards a somewhat more complex and hypsodont m3), but the cross-section of its rostrum is less square and the rhinarium muscles scars are reduced, suggesting an omnivorous diet with a reduced degree of rootling.
Based on our results, two different groups of Kolpochoerus coexisted during the Pleistocene: the K. limnetes/K. olduvaiensis phyletic lineage, and the newly described clade of “bunolophodont suines.” Whereas the former displays clear, progressive adaptations to grazing, as shown by the increased length, hypsodonty, and complexity of the third molars, as well as the reduction of the premolars (Harris and White 1979), K. phillipi sp. nov. and K. majus retained their overall unspecialized dental morphologies from the apparent origin of the lineage ∼2.6 Ma up to the last known occurrence of K. majus around 0.1 Ma (Kalb et al. 1982a; Gilbert 2008; Middle Awash research project unpublished data). During this period, lateral pillars apparently increased in height relative to crown length, and enamel became more rugose as an anchor for thicker cementum, phenomena also observed in the living Hylochoerus. A bunolophodont morphology associated with an otherwise unspecialized dentition and skull was therefore a hallmark of the lineage leading to K. majus, in contrast to the obviously specialized cranio-mandibular and derived dentition of Hylochoerus. Future paleoecological reconstructions using dental microwear, stable isotopes, and morpho-functional analysis should help to shed light on the relationships between the cranio-mandibular morphology, dental morphology, and diet of these “bunolophodont suines”.
We describe the new species Kolpochoerus phillipi sp. nov. and suggest that it may either represent an early member of a clade of “bunolophodont suines”, or, more likely, an ancestral chronospecies of the younger K. majus. In order to better understand the interrelationships, as well as the mode and tempo of evolutionary change within this lineage, additional efforts are needed in several respects. More collection is planned for the Matabaietu area to increase the number of specimens assignable to the new species. In addition, ongoing stratigraphic and geochronological studies on both sides of the Awash River in the Middle Awash will further refine the placement of specimens of both K. phillipi sp. nov. and K. majus. We judge it likely that in time, either here or elsewhere in Africa, there will be transitional forms found that may either confirm the status of K. phillipi sp. nov. as an independent species, or unite the latter and K. majus as chronospecies within a single phyletic lineage. We are of the opinion that the creation of yet another chronospecies for such yet undiscovered fossils-to fill such temporal gaps along a phyletic lineage (as was the case for the recently proposed K. millensis; Haile-Selassie and Simpson 2013)-is a classificatory practice that over-parses the fossil record and unnecessarily inflates the roster of Plio-Pleistocene suids. Only additional fossils will clarify these matters.
Lineage differentiation within the Kolpochoerus clade has been clarified by the recognition of K. phillipi sp. nov. as a close relative and potential ancestor of K. majus, but timing the origin of this species lineage remains difficult. Likewise, the K. limnetes lineage also continues to be plagued by a paucity of adequate cranial material, and its earliest members are usually only represented by dental remains. Obviously, by the time K. phillipi sp. nov. appears in the African fossil record at 2.5 Ma, there has been cladogenesis within Kolpochoerus, presumably arising from a common ancestor such as K. afarensis. Recent discoveries in the 3.0–2.5 Ma intervals of the Omo Shungura Formation of southern Ethiopia and the Middle Awash of the Afar Rift promise to resolve the details of this cladogenesis, and clarify the lower bounds of the lineages leading to K. phillipi sp. nov. and K. limnetes, pending completion of the chronostratigraphic and anatomical studies currently underway.
We are grateful to the staff of the institutions, where we spent much time studying collections of extant and fossil suids: Joséphine Lesur and Christine Lefèvre (Muséum National d'Histoire Naturelle, Paris, France), Wim Wendelen and Emmanuel Gilissen (Musée Royal d'Afrique Centrale, Tervuren, Belgium), Roberto Portela, Richard Sabin, Andy Currant, and Jerry Hooker (all Natural History Museum, London, UK), Jara HaileMariam, Yonas Desta, Mamitu Yilma, Dawit Tibebu, Tomas Getachew, and Andergachew Girmaye (all Authority for Research and Conservation of Cultural Heritage, Addis Ababa, Ethiopia), Emma Mbua, Mary Muungu, and Ogeto Mwebi (all Kenya National Museums). We also thank the Omo Group Research Expedition for access to its specimens. We are grateful to Zeresenay Alemseged (California Academy of Sciences, San Francisco, USA) and Denis Geraads (Muséum National d'Histoire Naturelle, Paris, France) for access to the Asbole suids. Preparation work of the holotype specimen of the new species was done with much help by Wogen Amarga (Middle Awash research project). We also thank Patrick Vignaud, Ghislaine Florent, and Carine Noël (Institut de Paléoprimatologie, Paléontologie Humaine: Évolution et Paléoenvironnements, Université de Poitiers, France) for their support, the editor Felix Marx (University of Bristol, Bristol, UK), John Harris (Natural History Museum of Los Angeles County & George C. Page Museum, Los Angeles, USA), Ray Bernor (Howard University, Washington, DC, USA), and an anonymous reviewer for their reviews that helped to improve our manuscript, and Gen Suwa (University of Tokyo Museum, Tokyo, Japan) for fruitful discussions. AS was supported by a European Union SYNTHESYS grant (GB-TAF 886) for his work at the NHMUK, and also by the Agence Nationale pour la Recherche (project ANR-09-BLAN-0238, Michel Brunet) for other collections. For funding of the Middle Awash research project, we thank the NSF (this material is based on work supported by grants 8210897, 9318698, 9512534, 9632389, 9729060, 9727519, 9910344, and 0321893 (HOMINID-RHOI), the Institute of Geophysics and Planetary Physics of the University of California at Los Alamos National Laboratory (LANL), and the Japan Society for the Promotion of Science. Desmond Clark (University of California, Berkeley, USA) inspired this effort and conducted Middle Awash laboratory and/or field research. We thank the Ministry of Tourism and Culture and the Authority for Research and Conservation of the Cultural Heritage for permissions and facilitation; and the Afar Regional Government, the Afar people of the Middle Awash, and many other field workers who contributed directly to the research effort and results.
List of characters used in the cladistic analysis.
0. Anterior part of male zygomatic arch in lateral view: (0) sub-horizontal; (1) strongly drooping.
1. Orientation of zygomatic arch in dorsal view in males: (0) oblique (angle <90°); (1) transverse (angle ca. 90°).
2. Development of zygomatic arch in males: (0) thin; (1) latero-medially thickened zygomatic; (2) rounded knob.
3. Relative cranium length: (0) relatively short (average 100*C10/C1>30); (1) relatively long (average 100*C10/C1<30).
4. Angulation between rostrum and neurocranium: (0) absent; (1) present.
5. Relative height of rostrum: (0) relatively high (average 100*C14/C15>50); (1) relatively low (average 100*C14/C15<50).
6. Constriction of parietal roof posteriorly to orbital processes: (0) constricted (average 100*C11/C10<45); (1) little constricted (average 100*C11/C10>45).
7. Development of rhinarium muscle scars: (0) well-marked scars; (1) reduced scars.
8. Premaxillary proportions and inclination in males: (0) relatively narrow (average 100*C5/C4<85) and sub-horizontal (average angle C19>150°); (1) Relatively wide (average 100*C5/C4>85) and inclined ventrally (average angle C19<150°).
9. Premaxillary proportions and inclination in females: (0) relatively narrow (average 100*C5/C4<85) and sub-horizontal (average angle C19>150°); (1) relatively wide (average 100*C5/C4>85) and inclined ventrally (average angle C19<150°).
10. Crest-like supracanine flange: (0) present; (1) absent. 11. Supracanine flange dorso-ventral development: (0) low; (1) high.
12. Supracanine flange antero-posterior development: (0) short; (1) elongated.
13. Mandibular symphysis proportions in males: (0) relatively narrow (average 100*M4/M3>85); (1) relatively wide (average 100*M4/M3<85).
14. Mandibular symphysis proportions in females: (0) relatively narrow (average 100*M4/M3>85); (1) relatively wide (100*M4/M3<85).
15. Mandibular corpus swelling in males: (0) weak (average 100*M20/M8>36); (1) strong (average 100*M20/M8<36).
16. Lower incisor arcade angle in males: (0) sharp incisor arcade (average angle M19>40°); (1) flat incisor arcade (average angle M19<40°).
17. Lower incisor arcade angle in females: (0) sharp incisor arcade (average angle M19>40°); (1) flat incisor arcade (average angle M19<40°).
18. Development of upper premolars: (0) P1 usually present and well-developed.; (1) P1 usually absent, P2 well-developed; (2) P2 reduced or even absent.
19. Development of lower incisors: (0) i1 and i2 very long and parallel, forming a comb; (1) i2 smaller and oblique to i1.
20. Mesio-distal compression of lateral pillars in molars: (0) not compressed; (1) compressed.
21. Wear pattern of lateral pillars (labial and lingual walls): (0) always rounded; (1) tend to be flat.