Xujiayao 14 is a right mandibular ramus with a largely complete lateral surface but a partial medial surface (Fig. 1). The condylar neck is complete, but the condyle is absent. The coronoid process is largely complete. The mandibular notch is intact, with minor damage to the anterior edge just below the coronoid tip but continuing to the original edge of the condyle. The tip of the coronoid process was reduced antemortem, such that the original height was probably a few millimeters higher and the highest point was more anterior. The posterior ramal margin is intact from the condylar neck around gonion and the medial pterygoid insertion. The inferior margin is present laterally but not medially. The anterior margin of the ramus is present to the level of the distal M₃ alveolus, but there is no trace of the M₃ socket or of the porous retromolar surface. The medial surface is missing bone along the inferior alveolar nerve canal, preserving the floor of the canal by the mandibular foramen; it lacks the anterosuperior canal margin and the lingula, but the superior 14.5 mm of the posteroinferior margin is intact.

The mandible (Institute of Vertebrate Paleontology and Paleoanthropology (IVPP) PA 1497) was excavated in 1977 at the Xujiayao (Hsuchiayao) site, Locality 74093 in Houjiayao village, northwestern Nihewan Basin, Hebai Province, northern China (40° 06′ 02′′ N, 113° 58′ 39′′ E) (Wu, 1980). The deposits consist of open-air fluviatile and lacustrine deposits, with erosional surfaces present within stratigraphic layers of sandy or silty clay. The 1976, 1977 and 1979 excavations yielded thousands of lithic artifacts, abundant faunal remains, and 17 fragmentary human remains from a sloping layer of yellowish-green clay, that was between 8 and 12 m below the modern surface (Chia and Wei, 1976 and Chia et al., 1979). The faunal assemblage is dominated by late Middle and earlier Late Pleistocene species (Chia et al., 1979) and includes both colder and more temperate species; a cooler climatic inference is supported by palynological remains (Yan et al., 1979).

A precise radiometric age for the archeological level has remained elusive. Uranium-series dating on Equus sp. and Coelodonta antiquitatis tooth enamel provided mean ages between ∼104 ka BP and ∼125 ka BP (Chen et al., 1984 and Chen et al., 1982), within Marine Isotope Stage (MIS) 5. The deposits are above a paleomagnetically reversed sequence, below ∼15 mm, which has been interpreted as the early MIS 5 Blake Excursion (Liu et al., 1992), 119–126 ka BP (Lund et al., 2006). Preliminary optically stimulated luminescence (OSL) dating of the archeological horizon provided late MIS 4 ages (60 ± 8 and 69 ± 8 ka BP) (Nagatomo et al., 2009). The Xujiayao human remains therefore derive from early Late Pleistocene (MIS 5 to 4) deposits. Morphologically, the full human fossil sample represents late archaic humans and are distinct from H. erectus and early modern humans (Chia et al., 1979, Wu, 1980, Wu, 1986 and Wu and Poirier, 1995). In particular, the Xujiayao 6 and 12 occipital bones, the Xujiayao 15 temporal bone and the Xujiayao 1 dentition lack distinctive external morphological features of the Neandertals.

The Xujiayao 14 mandible is assessed using the one standard linear measurement preserved (minimum ramal breadth) and a set of discrete traits. For internal visualization, the mandible was μCT scanned using an industrial CT scanner (225 KV μCT, made by the Institute of High Energy Physics CAS) in the IVPP (tube voltage 150 kV; tube current 110 μA; pixel size: 50.5 μm).

The Xujiayao 14 mandibular ramus is scored for the six discrete traits (Fig. 2). They include: •

mandibular notch asymmetry;

The first two features appear to be unrelated to craniofacial spatial or functional considerations, although the second has been associated with hypertrophy of the sphenomandibular ligament (Richards et al., 2003 and Smith, 1978). The second pair of traits may be related to mediolateral proportions of the ramus relative to the cranial base and neurocranial vault, in that the crest to condyle position may reflect the mediolateral positioning of the crest and/or the condyle on the ramus, and gonial eversion may relate to masseteric hypertrophy and zygomatic positioning. An enlarged superior medial pterygoid tubercle, in the context of smaller more inferior ones, should reflect differential hypertrophy of those more superior fibers, which may relate to craniofacial proportions (Richards et al., 2003). And retromolar space presence is a product of the interrelationship of mandible length, ramal breadth and dental arcade length (Franciscus and Trinkaus, 1995) and appears to be allometrically scaled with superior mandibular length (Rosas and Bastir, 2004). In addition, as an indication of ramal size, the minimum ramus breadth [Martin #71a (Bräuer, 1988)] of Xujiayao 14 is compared across the Pleistocene Homo samples. The Xujiayao 14 ramus also has a distinct pit within the planum triangulare, an unusual feature briefly considered with respect to Pleistocene Homo rami. These features, most of which can be seen to be direct or indirect products of craniomandibular hypertrophy and/or proportions, can nonetheless serve to assess late archaic and early modern human mandibular ramal variation across the Old World.

The Xujiayao 14 mandible and Pleistocene Homo samples are compared to each other and to recent humans using alternative character state frequencies. Rami are bilateral, and right-left configurations are not necessarily independent; sample sizes therefore refer to individuals, and the frequencies employ bilateral averages, counting each side as 0.5 in cases of bilateral asymmetry and assuming symmetry when only one side is preserved for the trait. When character states divide a continuous morphology, intermediate forms are scored as 0.5 for each configuration.

The first comparative sample consists of other East Asian Pleistocene human mandibular rami, listed individually in Table 1. The other such samples consist of Pleistocene Homo mandibles from the western Old World (Africa, Southwest Asia and Europe) (Table 2 and Appendix) and two recent (late Holocene) modern humans samples. These samples are composed so as to provide comparative distributions through the Pleistocene and take into account the Late Pleistocene differences between Neandertal and early modern human craniofacial morphology. In addition, given the emergence of “Neandertal” craniofacial and dental morphology in western Eurasia during the Middle Pleistocene (Arsuaga et al., 1997, Martinón-Torres et al., 2013, Stefan and Trinkaus, 1998a and Trinkaus, 1987), the Middle Pleistocene sample is separated into an African one and a western Eurasian one.

The first “western” sample therefore consists of Late Pleistocene western Eurasian late archaic humans (Neandertals), from MIS 6/5 to 3. They are joined by a sample of early modern humans from western Eurasia and north Africa, from MIS 5 to early MIS 2; there is little difference in ramal morphology between the MIS 5 and the MIS 3-2 early modern humans (Trinkaus and Rougier, 2013) and the samples are therefore pooled. The earlier samples consist of a Middle Pleistocene African one, a Middle Pleistocene “Neandertal lineage” one from western Eurasia, and a small Early Pleistocene sample from southwest Asia and Africa. Data for late Holocene recent humans derive from modern Chinese and Euroamericans.

The paleontological observations are from the original specimens (when possible), published descriptions, and/or high quality casts. The latter two sources apply particularly for the lost Předmostí and Zhoukoudian specimens. For traits that do not appear to change during development (mandibular notch asymmetry and mandibular foramen bridging), older juvenile and adolescent specimens are included in the samples; the other features are known to change with facial growth or may well do so (Coqueugniot, 1999, Rhoads and Franciscus, 1996 and Richards et al., 2003), and they are only scored on late adolescent and adult mandibles.

The Xujiayao 14 has a minimum ramus breadth of 44.5 mm, which is the largest Pleistocene one known from East Asia (Fig. 3), falling above the two Middle Pleistocene values from Zhoukoudian and the early modern human values from Minatogawa, Moh Khiew, Tianyuan and Upper Cave; it is probably also modestly above that of the Early Pleistocene Sangiran 21 (Kaifu et al., 2005). It is at the top of the Late Pleistocene western samples, exceeded only by those of the Nazlet Khater 2 and Oase 1 early modern humans and the Shanidar 4 Neandertal. It falls in the middle of the pooled western Middle Pleistocene sample and below the two Early Pleistocene rami (Dmanisi D2600 and KNM-ER 992). It therefore has a ramal breadth most similar to those of western Middle Pleistocene mandibles. Nonetheless, although there is a general trend for rami to become narrower through Pleistocene Homo (ANOVA P = 0.004 across the comparative samples), there is substantial intergroup overlap.

It is possible to quantify the shape of the mandibular notch (Rak et al., 2002), but the taphonomic fragility of the region means that few Pleistocene Homo mandibles preserve the notch and coronoid process intact, thereby minimizing the number of individuals providing data (and hence the perceived variation). It is therefore scored as symmetrical (the lowest point in the middle third) versus asymmetrical (the lowest point in the posterior third) (Stefan and Trinkaus, 1998b), thereby permitting the inclusion of less complete rami. On this basis, the mandibular notch of the Xujiayao 14 is asymmetrical, in that the lowest point of the notch is posterior to the middle third of the distance from the condyle to the coronoid process. An asymmetrical notch is present in Sangiran 21 and Zhoukoudian G2 (as well as the juvenile Zhoukoudian C1), but it is absent from all East Asian early modern humans, all but two of the western early modern humans (Muierii 1 and Ohalo 2), and most recent humans (Table 2). Asymmetrical notches reach the highest frequency among the Neandertals (72.2%), but they are reasonably common among Early and Middle Pleistocene Homo. This feature of Xujiayao 14 contrasts principally with the early and recent modern humans, and it is generally similar to archaic Homo across the Old World.

The complete bridging of the mandibular foramen by the lingula, known as “lingular bridging” and the “horizontal-oval foramen” (Kallay, 1955 and Yamano and Yamaguchi, 1976) (it is distinct from and independent of bridging of the mylohyoid groove), consists of lingular enlargement and its fusion to the posterior margin of the foramen (Jidoi et al., 2000, Kallay, 1955, Richards et al., 2003 and Smith, 1978). It is rare among recent humans (Table 2; Jidoi et al., 2000, Smith, 1978 and Yamano and Yamaguchi, 1976), but it has been frequently noted on Middle and Late Pleistocene Neandertal lineage mandibles since Gorjanović-Kramberger (1906) and McCown and Keith (1939). It is absent from Xujiayao 14, as is indicated by the intact posteroinferior margin of the mandibular foramen (Fig. 1). It occurs in one late Middle Pleistocene archaic specimen (Table 2), two early modern humans [Oase 1 (unilaterally) and Upper Cave 104 (bilaterally)], and 38.5% of the Neandertals.

In most recent humans (all but one early modern human, Muierii 1), and most non-Neandertal Early to Middle Pleistocene humans (all but Sangiran 21 and Tighenif 2), the crest of the mandibular notch meets the condyle laterally, or within the lateral third of the condyle (Table 1 and Table 2). Among the Middle and Late Pleistocene Neandertal lineage fossils, however, in a majority of them (60.0%; N = 35), the crest meets the middle third of the condyle. Although this variation is seen in terms of the position of the crest relative to the condyle, it probably relates principally to the positioning of the condyle on the condylar neck. Xujiayao 14 and all of the eastern Eurasian mandibles except Sangiran 21 conform to the majority pattern.

In posterior view, the gonial region of Xujiayao 14 is distinctly everted, relative to the condylar neck and the lateral plane of the ramus (Fig. 1). In this feature, it is similar to the earlier Zhoukoudian remains. However, there is considerable variation within the fossil and recent samples (Table 1 and Table 3). The Early to Middle Pleistocene samples are too small to be distributionally meaningful, the Neandertal mandibles are mostly inverted at gonion, the early modern humans ones are mostly straight, the recent human ones are mostly everted, and all of the forms occur in reasonable frequencies in each of the sufficiently large samples. Moreover, gonial profiles are asymmetrical in 7.1% of the pooled recent human sample (N = 140) and in two Pleistocene humans (Tighenif 3 and Dolní Vĕstonice 16). The Xujiayao 14 gonial eversion is therefore unexceptional in a fossil or recent human context, contrasting principally with the Neandertals.

Neandertals have been noted (Rak et al., 1994) for the frequent differential enlargement of the superior medial pterygoid tubercle on the medial ramus, tubercle #6 or the “medial pterygoid tubercle” (MPT) of Richards et al. (2003); following them and Creed-Miles et al. (1996), it is considered equivalent to the tuberculum pterygoideum inferius of Weidenreich (1936). More specifically, it is scored as “present” if this tubercle is strongly developed in contrast to the smaller ones that occur more inferiorly on the gonial margin (cf. Rak et al., 1994). A general enlargement of all of the medial pterygoid tubercles around gonion, occasionally seen in recent human mandibles, does not qualify as “present” for this trait. Because the tubercle is frequently differentially enlarged on immature recent human mandibles (Creed-Miles et al., 1996 and Richards et al., 2003), it is scored only on adolescent and adult mandibles.

On the Xujiayao 14 mandible, the superior tubercle is relatively and absolutely prominent (Fig. 1 and Fig. 4). There are five small ones inferiorly, each extending ∼6 mm from the gonial margin and minimally raised from the medial surface. The superior tubercle, above these smaller ones, rises to a sharp point, extends 2 mm medially from adjacent surface bone and 8 mm anteriorly from posterior margin, and is ∼12 mm high. Enlarged superior medial pterygoid tubercles, ones that are similarly enlarged both absolutely and relative to the more inferior ones, are occasionally present in other human mandibles (Table 2); they are evident in Minatogawa 1 and 5 and possibly ATD6-96 (Carbonell et al., 2005 and Suzuki, 1982). They are found in elevated frequencies principally in the Neandertal lineage, occurring in 74.2% of the mandibles. This feature therefore aligns Xujiayao 14 predominantly, but not exclusively, with the roughly contemporaneous Neandertals.

Xujiayao 14 retains all of the anterior ramal margin and a small portion of the superolateral corpus margin anterior of the ramus, but there is no trace of either the M₃ alveolus or of the retromolar alveolar bone (Fig. 1). There is a portion of the sulcus anterior of the crista endocoronoidea and torus triangularis. Internally, there is no evidence of trabecular bone or alveolar margin in the region in which an M₃ distal root socket should have occurred (Fig. 4). Comparisons with robust Late Pleistocene mandibles, with and without a retromolar space, indicate that evidence of the M₃ and/or retromolar alveolus should be on the preserved portion if a retromolar space was absent. In addition, the large dimensions of the Xujiayao 14 ramus, in a Pleistocene archaic human context, also suggest that a retromolar space should have been present (Rosas and Bastir, 2004). It therefore appears likely that the distal M₃ was mesial of the anterior ramal margin in norma lateralis [or that there was a retromolar space sensu stricto (Franciscus and Trinkaus, 1995)]. For these reasons, Xujiayao 14 is tentatively scored as having a retromolar space, bearing in mind the damage to the region.

As noted above, retromolar spaces are not features by themselves but are the results of the relative dimensions and positions of more primary aspects of the mandible (Franciscus and Trinkaus, 1995 and Rosas and Bastir, 2004). They are absent in earlier Homo mandibles but are common among Neandertal lineage specimens and occasionally present among early modern humans (Table 2). If Xujiayao 14 possessed such a mandibular gap, it would suggest a relatively long mandible, given its wide ramus (Fig. 3) and the large dental dimensions of the Xujiayao 1 maxilla and the Xujiayao 2, 13 and 17 isolated molars (Chia et al., 1979 and Wu, 1980). Among the ten Late Pleistocene mandibles with ramal breadths approaching or exceeding that of Xujiayao 14 (all ≥ 43 mm), half of them (all Middle Paleolithic in association) exhibit retromolar spaces.

On the medial aspect of the Xujiayao 14 superior ramus, between the crista endocoronoidea and the crista endocondyloidea, and hence within the planum triangulare, there is a distinct depression, with a small foramen in its anterior portion (Fig. 1). Among Pleistocene Homo (and recent human) mandibles, this region is generally flat or modestly concave, and it may be slightly convex. It rarely exhibits a distinct pit. In the pooled sample of Pleistocene Homo mandibles (N = 69), small depressions or pits are evident in only six mandibles, one of them unilaterally. They are distributed across most of the comparative samples [1 eastern early modern human (N = 6); 2 western early modern humans (N = 23); 1 Neandertal (N = 22); 1 western Middle Pleistocene human (N = 8); 1 Early Pleistocene mandible (N = 3)]. Moreover, in none of them does the size or depth of the depression approach that of Xujiayao 14. In eastern Eurasia, it is present only unilaterally on the Moh Khiew 1 mandible (Matsumura and Pookajorn, 2005) and on Xujiayao 14. Its significance is unclear, and it does not morphologically distinguish the comparative samples.