Abstract: We report here additional remains referred to Egatochoerus jaegeri from the late Eocene locality of Ban Mark in the Krabi basin, Thailand. The new material described, comprising upper and lower cheek teeth, deciduous premolars and partial cranial remains, makes E. jaegeri the best-documented Eocene representative of Old World Suoidea at present. Detailed study and comparison of their cheek teeth structure reveal a homogeneity of the molar crest and groove patterns of Old World and New World Palaeogene suoids.
A lthough the Neogene record of Old World suoids (comprising pigs (Suidae) and peccaries (Tayassuidae)) is rather well documented (Pickford 1988, 1993; Made 1990, 2010; Liu et al. 2002; Harris and Liu 2007), these groups are poorly known from the Palaeogene. The earliest Suoidea are known from Asia with HuaxiachoerusLiu, 2001 and EocenchoerusLiu, 2001 from the middle late Eocene of China (Liu 2001), EgatochoerusDucrocq, 1994 from the late Eocene of Thailand (Ducrocq 1994), SiamochoerusDucrocq et al., 1998 known from the late Eocene of Thailand (Ducrocq et al. 1998) and from the middle late Eocene of China (Liu 2001), and OdoichoerusTong and Zhao, 1986 from the late Eocene or early Oligocene of China (and see Ducrocq 1994; Ducrocq et al. 1998; Liu 2003 for a discussion about the status and age of this genus). The basal relationships within suoids have long been problematic, and the familial status of the Eocene taxa is still uncertain (Harris and Liu 2007). Systematic reviews including Palaeogene suoids grouped Huaxiachoerus, Egatochoerus, Siamochoerus and Odoichoerus in ‘Suoidea indet.’ (Liu 2003) or ‘Primitive suoids’ (Harris and Liu 2007), and suoids basal relationships remained uncertain in recent phylogenetic analyses (Liu 2003; Orliac et al. 2010a). Made (2010) proposed placing Huaxiachoerus, Eocenchoerus and Odoichoerus in the Palaeochoeridae, Siamochoerus in the Suidae and rejecting the inclusion of Egatochoerus in Suoidea. This brief summary of recent papers illustrates the confusion that prevails in the systematics of early Suoidea. This lack of resolution primarily stems from the scarcity of Eocene suoid remains (most Eocene taxa are based on isolated teeth) and from the ‘Oligocene gap’ in the fossil record (Orliac et al. 2010a, b), which makes it difficult to elucidate the relationships between the Eocene and Neogene taxa.
The Krabi fauna in southern Thailand from which the new suoid material is described here includes about thirty distinct mammal species (Ducrocq et al. 1992, 1995, 1996, 1997a, 2006a, b; Ducrocq 1999a, b; Marivaux et al. 2000, 2006; Peignéet al. 2000; Métais et al. 2001, 2007a, b; Antoine et al. 2003) that includes two suoids, Egatochoerus jaegeriDucrocq, 1994 and Siamochoerus banmarkensisDucrocq et al. 1998. In the Krabi basin, dental and postcranial remains of E. jaegeri have been described from the localities of Wai Lek and Bang Pu Dam (Ducrocq 1994). We report here new dental and cranial remains from Bang Mark locality (Text-fig. 1) that can be referred to this taxon. The additional material, including milk dentition and partial cranial remains, provides better knowledge of this poorly known suoid taxon, so far documented by upper molars and a fragmentary mandible with badly preserved dentition. The mammal assemblage (Ducrocq et al. 1995) and magnetostratigraphic analyses (Benammi et al. 2001) indicate a late Eocene age for the Krabi fauna, making E. jaegeri the best-documented Old World Eocene suoid to date. This new material is therefore fundamental for the understanding of basal suoid morphology and relationships.
Institutional abbreviations. DMR, Department of Mineral Resources, Bangkok; IPS, Instituto de Paleontología Miguel Crusafont, Sabadell; KNM, Kenyan National Museum, Nairobi; MGZ, Museun of the Guangxi Zhuang Autonomous Region; MNHN, Muséum National d’Histoire Naturelle, Paris; NHM, Natural History Museum, London; SMN, Staatliches Museum für Naturkunde Stuttgart; UM2, Université Montpellier 2, Montpellier.
Holotype. TF 2681 (right lower jaw with c, p2-m3; Ducrocq 1994; Text-fig. 3); Bang Pu Dam, Krabi basin, late Eocene of Peninsular Thailand.
Referred material. Ban Mark (Krabi basin, late Eocene of Peninsular Thailand) – TF 2978 (fragmentary left lower jaw with dp3-4 and m1-m2, and right m1-m2); TF 3477 (fragmentary skull with jugal and a portion of maxilla with left DP4 and part of P4); TF 3478 (left M1); TF 3479 (left M1-M2); Wai Lek (Krabi basin, late Eocene of Peninsular Thailand) – TF 2674 (fragmentary right maxillary with P4-M1); TF 2672 (fragmentary right maxilla with M1-M2); TF 2826 (left DP4); TF 2827 (left M2); Ban Pu Dam (Krabi basin, late Eocene of Peninsular Thailand) – TF 2825 (fragmentary left lower jaw with p3-m3); TF 2826 (right DP4); TF 2672 (right M1/M2); TF 2673 (right M1); TF 2967 (left i2); TF 2968 (right i3); TF 2969 (right I1); TF 2813 (right astragalus); TF 2812 (right astragalus); TF 2814 (right calcaneus). All the specimens are curated at the Department of Mineral Resources, Bangkok.
Type locality and horizon. Lignite mine, Bang Mark pit, Krabi basin, southern Thailand (7°58′31′’N, 99°03′37′’E), upper level of the main lignite seam of Bang Mark pit (Formation B2, see Bristow 1991); faunal and magnetostratigraphic evidence suggests a late Eocene age (Ducrocq et al. 1995, 1997b; Benammi et al. 2001).
Emended diagnosis. Small-sized suoid with a short and straight jugal; upper molars with clear centroconule, sharp and complete centrocrista, ectometacristule, fused internal roots, system of grooves poorly expressed; lower jaw deep with a clear vascular incisura; lower canine vertical and with a triangular cross-section; premolars increasing in size distally; diastema between c and p2; no trace of p1; p4 with a clear metaconid; lower molars with a deep waist between the two lobes, with large and distinct mesoconulid and hypoconulid, with strong postectocristids on the mesial lobe lined by deep postfossids and without clear postentocristid; dp4 with only two root pillars.
Remarks. Distinguished from Siamochoerus by deeper main fossids and a simpler groove pattern in lower molars, higher cusps with more vertical lingual and buccal walls, more individualized mesoconulid, the presence of a small cuspulid on the buccal face of the metaconid (endometaconulid, also observed in Odoichoerus and Palaeochoerus). Differs from Odoichoerus by the presence of a metaconid on p4, less massive cusps of the lower molars, weaker endoprotocristid on m2, stronger postectometacristid and ectoentocristid and the presence of an ectostylid on m1-m2. Morphologically very close to Palaeochoerus from which it mainly differs by a smaller size, and stronger postectocristid on the mesial lobe of lower molars. P4 taller with more mesiodistally compressed and more centrally located protocone than that of Palaeochoerus; further differs from Palaeochoerus by the presence of a complete centrocrista on upper molars.
Cranial fragment and upper dentition. DMR TF 3477 is a skull fragment preserving the anterior part of the zygomatic arch and a portion of the maxillary with left DP4 and the buccal part of P4 still in its crypt (Text-fig. 2C, F). There is no trace of the M1 behind the DP4 (Text-fig. 2E). The remaining part of the zygomatic arch preserves the ventral edge of the orbit and part of the ventral contact between the jugal and the maxilla. This contact is bounded by the insertion for the masseter muscle and restricted to the mesial-most end of the jugal. The preserved anterior part of the zygomatic arch of this juvenile individual is thin, shallow, and the ventral edge is straight (Text-fig. 2D).
Only the mesiobuccal part of P4 can be directly observed: the paracone is low, and it displays a weak preparacrista connected to the mesiobuccal corner of the tooth. A tiny spur of enamel is situated very close to the apex of the cusp on the postparacrista; it might correspond to an incipient metacone. The buccal and mesial faces of the crown are straight and flat, and a small cingulum occurs mesially. The CT scanning of the specimen reveals the presence of major enamel knob lingual to the paracone that most probably corresponds to the protocone, but the finest crests of the latter cusp cannot be described (Text-fig. 2F). The specimen DMR TF 2674 from Wai Lek (Ducrocq 1994) preserves the lingual part of a P4, which allows the completion of this description. The protocone lobe is massive; the cusp is high and pointed and much taller than the protocone of M1. There is neither postprotocrista nor lingual cingulum. The preprotocrista is high, and it constitutes a transverse mesial cutting edge.
The DP4 of DMR TF 3477 is only slightly worn, and its mesiobuccal corner is missing (Text-figs 3E, 6A). The tooth exhibits a slightly trapezoidal occlusal outline with the lingual part shorter than the buccal one. The buccal cusps are larger than the lingual ones, and their crest pattern is simple. The protocone is mesiodistally compressed, and the anterior part of the paracone is broken. The paraconule is present but heavily worn; it is situated between the two mesial cusps. A strong and sharp postparacrista is mesiodistally oriented and connected with the premetacrista, thus forming a distinct centrocrista. A well-developed centroconule occurs in the transverse valley. The metacone bears two sharp crests: a premetacrista and a postmetacrista, the latter being connected to a weak metastyle on the distobuccal corner of the tooth. The metaconule is heavily worn, and it only preserves a short distally oriented endometacristule and a postmetacristule that merges into the distal cingulum. The mesial cingulum is strong and displays a large and distinct mesiostyle; the distal cingulum, inflated at the distostyle level, extends on the whole distal face of the tooth. There is no buccal cingulum; a clear entostyle occurs at the lingual end of the transverse valley.
The morphology of the upper molars is nearly identical to that of DP4 with the exception of their more square occlusal outlines and the reduction of the entostyle that merges into the ectometacristule. The centroconule is small and lingually shifted, and a small buccal cingulum runs from the metastyle under the metacone to the distal half of the paracone. The centrocrista is continuous and sharp (Text-figs 3F, 6A). The lingual roots are fused.
Mandible and lower dentition. The lower dentition of E. jaegeri was previously known from a right lower jaw on which molars were too badly preserved to allow proper description (type specimen, DMR TF2681). New material from Ban Mark includes a fragmentary juvenile mandible (DMR TF 2978) preserving unworn m1-m2 and dp3-dp4, which yields information on lower permanent and deciduous molar morphology. The specimen was scanned using X-ray microtomography (CT scan) to investigate hidden structures such as foramina and root morphology.
The lower jaw is broken just in front of dp3 and behind m2; there is no trace of m3 on the preserved part (confirmed by CT scan images). Four mental foramina can be observed on the mandible (Text-fig. 2A): one under the anterior root of dp3, one in a lower position under its posterior root, a third one under the anterior root of dp4 and a fourth one situated only a few millimetres below the second lobe of dp4. The dp3 is a very short biradicular (double-rooted) tooth, with a lateral triangular outline. The mesial face is somewhat distally curved, and only a very slight enamel spur occurs on the base of the mesiolingual corner of the crown. The buccal face of the tooth is slightly more convex than its lingual face, which distolingually bears a very slight enamel swelling. The distal side of the apex is worn and flat, but most of the posterior part of the crown is lacking (Text-fig. 3A, B).
The dp4 is typically three-lobed with two cusps on each lobe. The tooth has only two root pillars, with a mesial one lying under the mesial lobe and the mesial half of the second lobe, and a more massive distal one under both the distal half of the second lobe and the distal lobe as shown in Text-figure 2B. The tooth widens distally. The cusps are tall and pointed, separated by wide transverse valleys. There is a clear preparaconulid lying between the primonid (sensu Made 1996b) and the paraconid. The paraconid and the primonid bear sharp postectocristids joining the preprotocristid and the premetacristid, respectively. The cusps of the second lobe (protoconid and metaconid) are linked by their postcristids, thus forming a transverse crest. The metaconid bears a sharp and high postectocristid that connects to a sharp ectoendocristid. The latter exhibits a small stylid, also present on m1. The mesoconulid is small and has a slight lingual extension. The distal cusps are simple, and there is no trace of a transverse crest. The hypoconulid is large and distinct from the distal cingulid (Text-fig. 3A–C).
On all molars, the trigonid is higher than the talonid, and the system of grooves (‘Furchenplan’ of von Hunermann 1968) is deep. The buccal side of the molars is more waisted than their lingual face (Text-fig. 3B, C). The enamel is smooth to finely wrinkled. The m1 is morphologically similar to m2, with the exception of its smaller size. The m1 and m2 are elongated and bear four slender and high cusps. The crest pattern is rather simple. The protoconid only displays a short preprotocristid and a sharp and low postprotocristid. A low enamel shelf runs down from mid-height of the mesial face of the protoconid; it tapers at the base of the metaconid and partially closes the trigonid mesially. A low postectoprotocristid also occurs and connects to a small ectostylid at the buccal end of the transverse valley. The metaconid displays a short premetacristid and sharp postecto- and postmetacristids. The postmetacristid connects with the postprotocristid to form a continuous transverse structure that interrupts the sagittal valley. A tiny cuspulid is present on the buccal face of the metaconid and here recognized as an ‘endometaconulid’. The endo- and postfossids of the mesial cusps are deep. A well-developed and distinct accessory cusp (mesoconulid) occurs in the transverse valley, and it extends lingually to reach the mesial wall of the entoconid. The hypoconid and entoconid are of similar size. The hypoconid is linked to the mesoconulid by a short prehypocristid, and it also displays a distinct endohypocristid and a short posthypocristid, the latter being connected to a well-defined hypoconulid. The entoconid exhibits a sharp ectoentocristid lined by a deep ectoentofossid, and a short endoentocristid that meets the endohypocristid. A very low, narrow and short cingulid occurs on the mesial face of the crown. The distal cingulid is present on each side of the hypoconulid, and it extends under the whole distal face of the crown (Text-fig. 3B, C).
Comparison with other suoids from Krabi basin
The type specimen of E. jaegeri from Ban Pu Dam is larger than the specimen from Ban Mark described here (DMR TF 2978, Table 1). Similarly, the lower teeth of the holotype being inflated and corroded because of the bad preservation of their enamel, they appear larger than those of DMR TF 2978 and their strong wear makes morphological comparisons impossible. However, the upper molars from Ban Mark are almost identical in size and morphology with those of E. jaegeri from Bang Pu Dam and Wai Lek. The suoid remains from Ban Mark described here are therefore assigned to this species.
Table 1. Dental dimensions in mm of Egatochoerus jaegeri specimens from Ban Mark, Ban Pu Dam and Wai Lek, late Eocene, Krabi basin, Thailand.
The new material from Ban Mark is much smaller than the lower and upper teeth of Siamochoerus banmarkensis from the same locality. The estimated body weight of E. jaegeri, based on m1 dimensions (m1 length and m1 surface, equation for nonselenodont artiodactyls, Damuth 1990), is about 9 kg, equivalent to the size of a pygmy hog (Sus salvanius, Oliver and Roy 1993). Using the same method, the estimated body weight of S. banmarkensis is about 30 kg. The lower molar crowns of E. jaegeri are higher with straight lingual and buccal walls, whereas in S. banmarkensis, crowns are lower with more bulbous walls and more centrally located cups (Text-fig. 4A vs C). The mesoconulid of the lower molars of both species of Siamochoerus (S. banmarkensis and S. viriosus) is less individualized and merged into the prehypocristid. Their main fossids are shallower, and their groove pattern is more complicated. DMR TF 2978 further differs from the lower teeth of Siamochoerus by the presence of a small cuspulid on the buccal face of the metaconid (endometaconulid). The upper molars of Siamochoerus and Egatochoerus share a similar structure with a distinct and well-developed centrocrista. However, the upper molars of Siamochoerus are only known by two specimens, broken in their central part. The deformation most probably artificially confers to the upper molars wider crowns with more slanted lingual and buccal walls.
Comparison with Eocene suoids from China
Odoichoerus uniconusTong and Zhao, 1986 from the late Eocene or early Oligocene of Guangxi, China is known only by its p4-m3, and it is morphologically very close to E. jaegeri (Text-fig. 4A vs B). Both share the same general lower molar structure with large and distinct mesoconulid and hypoconulid. On the buccal face of the metaconid, the lower molars of Odoichoerus exhibit a bulge of enamel that most probably corresponds to the accessory cuspulid observed in E. jaegeri and interpreted as an endometaconulid. Odoichoerus can be distinguished by its larger size, more massive cusps, the presence of a stronger endoprotocristid on m2, weaker postectometacristid and ectoentocristid, and the absence of an ectostylid on m1-2. The estimated body weight of O. uniconus (based on m1 length and m1 surface, equation for nonselenodont artiodactyls, Damuth 1990) is about 10 kg, which is very close to the estimated size of E. jaegeri.
Eocenchoerus savageiLiu, 2001 from the late Eocene of China (Gongkang Formation, Bose Basin, Liu 2001) is only known by isolated M3 and P4. Its size is larger than that of E. jaegeri; the general dental proportions are markedly different from other known Eocene suoids in being much more elongated and waisted (Text-fig. 5C). Compared to E. jaegeri, the lingual cusps of Eocenchoerus are more mesially located than the buccal ones. Eocenchoerus also exhibits a distinct distocone that occupies a lingual position (which is very unusual in suoids where this cusp is always more buccally situated). The paraconule is mesially located and connected to the mesial cingulum, which bears a large mesiostyle. Contrary to other Eocene suoids, there is no clear centrocrista. In its proportions and the buccal extension of the centroconule, the morphology of Eocenchoerus is similar to that of Taucanamini (considered as Palaeochoeridaed by Made 2010) such as Taucanamo grandaevum (Text-fig. 5I). The P4 referred to Eocenchoerus by Liu (2001) is heavily worn. The shape of the worn protocone indicates that it might have had a postprotocrista, which is a potential synapomorphy with the Suidae (Orliac et al. 2010a).
Huaxiachoerus guangxiensis is known from the late Eocene of China from isolated upper molars (Liu 2001). The published material is heavily worn, and little can be said about the molar structure. A centrocrista seems to occur (Text-fig. 5D). It is noteworthy that Huaxiachoerus displays a M3 smaller than the M2, like the Eocene North American suoid Perchoerus.
An indeterminate suoid has been described by Métais et al. (2008) from the early middle Eocene locality of Shanghuang in eastern China. This form, represented only by a left m3, clearly differs from Egatochoerus by its much larger size and mainly by its crest pattern and by the absence of a mesoconulid. Recent restudy allows us to recognize it as a non-Indopakistani representative of the Raoellidae (Orliac and Ducrocq 2011).
Comparison with European Palaeogene suoids
With the exception of a recent description of early Oligocene suoids from the Bugti Hills, Pakistan (Orliac et al. 2010b), Oligocene Old World suoids are mainly known by the European Palaeochoeridae whose taxonomic content and familial status remain problematic (Orliac et al. 2010a). The Thai species E. jaegeri is morphologically very close to Palaeochoerus (Quercy localities of Pech Crabit MP23 and Pech Desse MP28, France). It mainly differs from the latter by a smaller size and stronger postectocristid on the mesial lobe (especially the postectoprotocristid, which is lacking in Palaeochoerus, Text-fig. 4G). The accessory cusplet located on the buccal face of the metaconid in E. jaegeri and Odoichoerus is also observed in Palaeochoerus (Text-fig. 4A, B vs G). The deciduous teeth also display morphological similarities in both genera. The dp4 of Palaeochoerus and Egatochoerus only has two roots, whereas three roots are usually present in artiodactyls, including Suidae. Both genera have a distinct preparaconulid and an incomplete mesial ‘lophid’ (Text-fig. 6F vs G). They are markedly different from that of basal suids like Listriodontinae (Text-fig. 6H), which present an even more complicated structure with a clear mesial lophid, and a small third root pillar on the buccal side.
The P4 of Egatochoerus (DMR TF 2674) differs from that of Palaeochoerus by its taller and more mesiodistally compressed protocone that occupies a central position on the crown, whereas it is more mesial in Palaeochoerus. The distal part of the tooth is clearly more developed in Palaeochoerus. However, both taxa lack postprotocrista. The upper molars of Egatochoerus and Palaeochoerus share a very similar morphology (Text-fig. 5A vs H). The main difference lies in the lack of complete centrocrista in Palaeochoerus. In the latter, postparacrista and premetacrista are separated by the buccal end of the transverse valley. However, the post- and prefossae of the centrocrista are almost connecting in the Palaeochoerus from Pech Crabit (MP 23). These crests are farther apart in specimens from younger localities such as Pech Dess and Pech du Fraysse (MP 28). The centrocrista of Eocene suoids is lined buccally by clear fossae, which are reduced or absent in Palaeochoerus specimens. In Palaeochoerus, the centroconule merged into the premetacristule unlike Egatochoerus where the centroconule is isolated. In the latter, the lingual and buccal walls of the upper molars are more slanted than in Palaeochoerus, and the cusps are slightly more centrally located. The surface involved in the chewing motions is thus reduced in Egatochoerus compared to Palaeochoerus in which cusps are farther apart. As for lower molars, morphological similarities observed in upper permanent molars are even more striking in deciduous molars (Text-fig. 6A vs B), which mainly differ by their crown outlines. The groove and crest pattern of the DP4 of Palaeochoerus and Egatochoerus are almost identical, with an individualized entostyle and single lingual root pillar. Both genera exhibit a small entostyle at the lingual end of the transverse valley.
The morphology of the anterior part of the zygomatic arch of E. jaegeri is similar to that of Palaeochoeridae. The dorsal edge is only slightly curved, the mesial-most part being almost straight (Filhol 1882, MNHN Qu1). The contact with the mesiolateral wing of the squamosal occurs anterior to the distal bar of the orbit and is therefore more mesial in E. jaegeri than in Palaeochoeridae. The same morphology is observed in the alleged palaeochoerid Taucanamo from the early Miocene of Els Casots (MN4, Spain, IPS 9097) and the middle Miocene of Steinheim (MN7, Germany, SMN 9901).
Comparison with New World suoids
The dentition of extant New World suoids, or tayassuids, is highly derived, with the multiplication and reduction of secondary structures (Text-figs 4F, 5G), making comparison with Eocene taxa difficult. New World suoids appear in North America during the late Eocene (middle Chadronian, Stucky 1992). We follow here the systematic classification proposed by Wright (1998) and consider the material of Perchoerus probus from the White River beds (late Eocene–early Oligocene, Scott 1940) to be representative of the genus Perchoerus for comparison purpose. Further comparisons are made with Cynorca from the early Miocene of the John Day Formation (Woodburne 1969). The lower molar crest pattern of Perchoerus resembles that of eastern Asian Eocene suoids Egatochoerus and Odoichoerus by connected precristids and postcristids and clear postectocristids lined by internal grooves on the mesial cuspids (Text-fig. 4A, B vs D). Comparison with the upper molars of Perchoerus is more problematic as the homology of cusps is not obvious. In Perchoerus, like in Egatochoerus and Siamochoerus, the centrocrista is complete and buccally lined by deep grooves. In Perchoerus and Cynorca, the ectometacristule, present in upper molars of all suoid, is distinct from a second mesiodistal crista, lingual to the premetacristule (Text-fig. 5E, F). In Perchoerus, the ectometacristule is connected to a small style that could correspond to the vestige of a hypocone (Scott 1940, pl. 52, figs 2, 3); in Cynorca, this structure is not present and the lingual crest that we recognize as an ectometacristule lies on the lingual side of the metaconule. In Tayassu (Text-fig. 5G), the structure is simpler and the crista lingual to the premetacristule is not developed.
The morphology of the anterior part of the zygomatic arch of Egatochoerus is very close to that of Perchoerus (Pearson, 1923; Scott 1940, pl. 52, fig. 1). The ventral edge of the jugal is straight in both taxa, but the curvature of the dorsal edge is slightly more important in Perchoerus. The contact with the mesiolateral wing of the squamosal is distal to the orbital bar in Perchoerus, suggesting that the mesial part of the zygomatic arch of Egatochoerus is rather, short like in Diseohyus (Woodburne 1969, pl. 48, fig. 9); this could, however, be attributed to the young individual age of the specimen. The shape of the mandible of Egatochoerus (Ducrocq 1994, fig. 3) is also similar to that of Perchoerus, both taxa exhibiting a deep vascular groove on the distoventral edge of the horizontal ramus and inflection of the dorsal part of the horizontal ramus anterior to P4. The angulus mandibulare of Egatochoerus is rounded but not down-turned as is the case in Perchoerus.
Egatochoerus as a suoid
Egatochoerus was initially referred to as a tayassuid by Ducrocq (1994). However, the scarcity of available material and the lack of resolution of basal phylogenetic relationships within Suoidea make the position of Eocene suoids problematic (Orliac et al. 2010a). Indeed, the taxonomic status of Egatochoerus has been debated, although the genus has been recently placed in ‘primitive Suoidea’ (Harris and Liu 2007). In a still more recent revision of suoid systematics, Made (2010) included Odoichoerus, Huaxiachoerus and Eocenchoerus within the Palaeochoeridae. Based on the height and bulkiness of the horizontal ramus of the mandible, the reduction of the p1, and on the development of the metaconid and the reduced hypoconid of the p4, Made (2010) challenged the attribution of Egatochoerus to the Suoidea. According to Made (2010), these characters are observed in later Suidae and Dicotylidae but are not likely to be observed in Eocene suoids. However, the picture changes a lot if the polarization of characters is not based on comparison with Eurasian Miocene taxa only. Indeed, the height of the mandible of the type specimen – most probably artificially increased by the bad state of preservation of the material – does not differ much from that of Perchoerus (Scott 1940, pl. 52, fig. 1), which also presents a reduced p1 and a distinct metaconid on the p4. A well-developed metaconid on p4 (also maybe artificially displaced lingually by postmortem deformation in the specimen from Krabi) is observed in several Eocene–Oligocene cetartiodactyls (heloyids, cebochoerids and dichobunids) indicating that this feature, if not plesiomorphic, is likely to be highly homoplastic. Made (2010, p. 113) makes no further comment about the systematic status of the upper molars of Egatochoerus that are ‘compatible with an assignment to the Suoidea’.
The additional material of Egatochoerus from Ban Mark and the thorough description of fine molar structures clearly indicate that this genus belongs to the Suoidea. Egatochoerus is now the best-documented Eocene suoid in the Old World. Comparison with other Eocene taxa both allows the refinement of characters common to all suoids and allows this group to be defined more accurately. By means of a formal phylogenetic analysis, Orliac and Ducrocq (in press) indicate that suoid lower molars differ from the primitive artiodactyl lower molar pattern by: (1) the absence of paraconulid; (2) the presence of precristids on the mesial cuspids; (3) the presence of internal cristids on the distal cupids (hypolophid) recognized as endohypocristid and endo- or preentocristid; (4) a larger entoconid; and (5) a more reduced central part of the molars occupied by a central accessory cusp. Within Suoidea, the Asian Eocene genera (Egatochoerus, Odoichoerus, Siamochoerus and Perchoerus) and Palaeochoerus display a very similar structure of their lower molars (Text-fig. 4): the buccal face is more waisted than the lingual one, the endometacristid bears a small accessory cuspid (except for Siamochoerus), the postcristids of the mesial lophid form a continuous transverse structure and the entoconid has no postcristid. Perchoerus, Siamochoerus, Egatochoerus and Palaeochoerus have a metaconid on their p4, which suggests that it might have been primitively present in Suoidea. Concerning the upper molars, Egatochoerus presents an ectometacristule that encompasses the remnant parts of the lingual cingulum, running from the top of the metaconule to the basis of the protocone. This structure is observed in all Suoidea. Egatochoerus also exhibit a continuous centrocrista on its upper molars, a plesiomorphic structure observed in Eocene suoids.
As to cranio-mandibular characters, the mandible of Egatochoerus exhibits a peculiar concavo-convex outline of the cheek tooth row in lateral view, also observed in Perchoerus and Palaeochoerus. This morphology is most probably constrained by the vertical orientation of the canines (Herring 1972). The straight and rather short anterior part of the zygomatic arch is also observed in Perchoerus and in the Palaeochoeridae sensu lato (including Taucanamo). The morphology of the zygomatic arch and of the mandible of Egatochoerus is similar to that of both Perchoerus and Palaeochoerus, suggesting that these features are likely to be plesiomorphic among Suoidea.
Two astragali and one calcaneum have been found in association with dental remains of Egatochoerus and were referred to this taxon. The astragalus of Egatochoerus presents the diagnostic features of suoids (medio-laterally convexe sustentacular facet, proximal and distal trochlea not parallel, no lateral connection between the distal extremity of the medial trochea and the proximal extremity of the distal trochlea; Orliac et al. 2006) but differs from other suoids by its general proportions, the bone is more slender medio-laterally and antero-posteriorly, and by a distal trochlea medio-laterally more slender than the proximal one. In lateral view, the distal part of the lateral border of the sustentacular facet is straight and anteriorly inclined as in Taucanamo and Tayassu (and not convex as in Doliochoerus, suids and sanitheres; Orliac et al. 2006; Ducrocq 1994). The calcaneum of Egatochoerus is of typical suoid morphology, and the sustentaculum tali is short with a ventro-laterally sloping posterior side. Like the astragalus, the antero-posterior width of the articular part of the calcaneum is slender antero-posteriorly. The postcranials of Egatochoerus are very different from other suoids and might represent the primitive morphology.
If Egatochoerus is clearly a representative of the Suoidea, this taxon exhibits an association of plesiomorphic features (adult dentition and milk dentition), which hinders us from proposing close affinities to one family or another, although placement within Suidae can be clearly ruled out. The lack of a third lingual pillar on the dp4, also observed in Palaeochoerus, could suggest close affinities to Palaeochoeridae, but observations on suoid milk dentition remain scarce and the distribution of this character needs further investigation. Egatochoerus is a non-suid stem-suoid.
Affinities of Eocene suoids
Recent classifications have left open the question of Eocene taxa (Liu 2003; Harris and Liu 2007), and most authors (for example Pickford 1988, 1993; Made 1996a) have previously pointed out that the dental distinction between suids and tayassuids was unclear, especially when primitive taxa are taken into account. Parallel to the lack of resolution concerning the Old World Suoidea relationships, recent phylogenetic analyses raised the question of the monophyly of New World Tayassuidae (Geisler 2001; Geisler and Uhen 2003, 2005; Theodor and Foss 2005; Spaulding et al. 2009; Orliac et al. 2010a). Those two problems are obviously related. Indeed, whereas the content of the Suidae is rather stable, Palaeogene suoids are drifting from Tayassuidae to Palaeochoeridae. However, the latter has primarily been named for grouping Oligocene suoids of Europe (e.g. Palaeochoerus) and America (e.g. Perchoerus, Matthew 1924) and is also regarded as Old World tayassuids. Made (2010) refers the Eocene taxa Huaxiachoerus, Odoichoerus and Eocenchoerus to the Palaeochoeridae mainly on the basis of fused roots on the lower and upper molars and of a preprotoconule fused to the protocone and not to the mesial cingulum (the only characters that can be observed on these poorly known taxa). These two characters are also observed in the alleged tayassuid Perchoerus from the early Oligocene of the White River Formation (Scott 1940) and are most probably symplesiomorphies at the Suoidea level. To sum up, the problem is basically because of the lack of synapomorphies defining Tayassuidae and Palaeochoeridae (Orliac et al. 2010a).
The new material described here and our comparison of fine dental structures among Eocene suoids indicate that Perchoerus, Odoichoerus, Egatochoerus and the European palaeochoerids share a very similar molar pattern. The main difference between Odoichoerus and other Eocene taxa lies in the absence of a metaconid on p4. Given the presence of a metaconid in other Eocene taxa, the lack of this cusp suggests a close relationship between Odoichoerus and Taucanamini, the latter also lacking a metaconid on p4. The lower molars of Taucanamo from the Early Miocene of Europe differ from those of Odoichoerus by a more complicated structure of the distal cuspids (presence of an ectohypocristid and a postentocristid; Text-fig. 4H). Taucanamini first occur in the early Oligocene deposits of the Bugti Hills in Pakistan, with Pecarichoerus sp. (Orliac et al. 2010b), suggesting that this group most probably differentiated during the Eocene. Unfortunately, the generic features described by Colbert (1933) on the upper molars of Pecarichoerus cannot be contrasted with Odoichoerus (whose upper dentition is unknown). However, Odoichoerus might be the earliest representative of the Taucanaminae, which most probably have a long Asian history poorly documented yet.
Among Eocene taxa, Siamochoerus is the most morphologically divergent taxon. Liu (2001) first assigned the genus to the Palaeochoeridae on the basis of the upper molar width, the lack of talon on M3 and lack of clear accessory cusps. Based on the results of a phylogenetic analysis, the same author then referred Siamochoerus to a basal indeterminate suoid family (Liu 2003). Made (2010) considered Siamochoerus as a representative of the Suidae, based on separate roots of the m2 of S. viriosus. The lower molars of Siamochoerus exhibit low cuspids with wrinkled enamel and a complex groove pattern close to that of the Sanitheriidae/Suidae (Text-fig. 4C vs I and J), which is congruent with the conclusions of Made (2010). However, Siamochoerus still presents a continuous centrocrista that highlights the primitive morphology of its upper molars.
Sanitheriidae and Suidae differ from other Palaeogene suoids by a weaker waisting of the lower molar buccal wall and the presence of a postcristid and/or postectocristid on the entoconid (Text-fig. 4I, J), and by a more complicated crest pattern of the buccal cusps of the upper molars (Text-fig. 5J, K). Marked differences are also present on the milk dentition: in Sanitheriidae and Suidae, the DP4 has two distinct lingual roots, the parastyle is located in the inflated mesiobuccal corner of the tooth (Text-fig. 6A, B vs E, D), and there is no entostyle. Note that the overall morphology of the DP4 of extant pecaries is more similar to that of Eocene suoids than to that of Suidae (no inflation of the mesiobuccal corner, presence of an entostyle; Text-fig. 6C). On the other hand, the dp4 of Neogene suoids is more complicated, with marked external grooves, and it lacks the paraconid (Text-fig. 6F, G vs H).
Egatochoerus jaegeri is now known from the localities of Ban Mark, Wai Lek and Ban Pu Dam in the Krabi basin, Thailand. It is documented by upper and lower, deciduous and permanent dentition as well as by fragmentary mandibular and cranial remains and is the best-documented representative of the Old World Eocene Suoidea at present. The dental and cranial features of this taxon unambiguously refer it to the Suoidea. However, the affinities of E. jaegeri, initially referred to as a member of the Tayassuidae (Ducrocq 1994), remain unclear. Indeed, the comparative study of the cheek teeth of Eocene suoids indicates that Egatochoerus, Perchoerus, Odoichoerus share a very close molar pattern of both deciduous and permanent dentition with the Oligocene European palaeochoerids. Palaeogene suoids also share cranio-mandibular characters, such as a concavo-convex outline of the tooth row and a straight and rather short anterior part of the zygomatic arch. The morphological homogeneity and the scarcity of Palaeogene suoid material make it difficult to clarify their phylogenetic relationships. ‘Palaeogene suoids’ clearly differ from more derived groups such as sanitheres and suids by the crest-and-groove pattern of their upper and lower permanent molars. Marked differences are also observed on the milk dentition, which highlights the potential phylogenetic importance of the deciduous teeth.
The co-occurrence of two morphologically divergent suoids –Siamochoerus banmarkensis and Egatochoerus jaegeri– in the late Eocene of Thailand (Ducrocq 1994; Ducrocq et al. 1998) and of three additional, distinct suoid genera in coeval deposits of China (Siamochoerus, Eocenchoerus and Huaxiachoerus; Liu 2001, 2003; Harris and Liu 2003) demonstrates the significant role of South-east Asia in the emergence and early evolution of the group. This diversity shows in turn that the differentiation between Suidae and Tayassuidae most probably occurred well before the end of the Eocene.
Acknowledgements. We thank J. Sudre and P.-O Antoine for their comments and L. Meslin for drawings. Many thanks to the two anomymous reviewers for their useful comments. Fieldwork in Thailand was supported by the Mission Paléontologique Française en Thaïlande (Ministère des Affaires Etrangères) and the Agence Nationale de la Recherche, project ANR-09-BLAN-0238. This work was also supported by the ANR, ERC Palasiafrica (ANR-08-JCJC-0017). This is ISE-M publication 2011–122.