Qianzhousaurus sinensis Lü et al., 2014:figs. 1–3 (original description).

Holotype—Nearly complete skull and almost complete left mandible (GM F10004-1), described in detail herein, along with associated postcranial skeleton (GM F10004-2-8).

Locality and Horizon—Nanxiong Formation (Upper Cretaceous, Maastrichtian), Longling Town, Nakang, Ganzhou City, Jiangxi Province, China.

Updated Diagnosis—Long-snouted tyrannosaurid tyrannosauroid dinosaur with following autapomorphies of the skull: large fenestra on the ascending ramus of the maxilla; extremely reduced premaxilla; laterally straight, rod-like nasal element that does not expand laterally around the naris region; anteroventrally oriented ventral process of lacrimal; gracile and shallow dorsal prong of anterior ramus of squamosal; anteroposteriorly oriented ventral process of squamosal; anterior edge of dentary smoothly merges with ventral dentary margin; highly concave dorsal margin of supradentary. Potential but problematic autapomorphies (see text): an ovoid foramen on the medial surface of prearticular; limited medial exposure of the coronoid in medial view. See Lü et al. (2014) for one additional diagnostic character of the postcranium (absence of pronounced vertical ridge on lateral surface of illium).

Description

Premaxilla—The premaxilla is not preserved in the Alioramus remotus and Alioramus altai holotypes, making Qianzhousaurus the first (and thus only) alioramin for which this bone can be described (Figs. 1, 2). The premaxilla is preserved on both sides of the skull, and is a small bone compared with the rest of the skull, as in tyrannosauroids generally (Brusatte et al., 2014:character 826). It is remarkably small in Qianzhousaurus, however, measuring only 2.2% of the overall basal skull length, compared with ∼4–10% in other tyrannosauroids (Lü et al., 2014). Thus, despite the longirostrine snout of Qianzhousaurus, the premaxilla is not elongated, like the maxilla and nasals are (it is largely the increased length of these bones that make the premaxilla so proportionally small compared with other tyrannosauroids).

The dorsal process of the premaxilla (Fig. 2) contacts the anterior end of the nasal, dorsal to the external naris, permitting both bones to form the dorsal rim of the naris. Contact with the maxilla is via a small, tapering process that contacts or nearly contacts the nasal underneath the floor of the naris (see below) (Fig. 2).

The external naris has the characteristic ‘tear-drop’ morphology as in most tyrannosaurids (Fig. 2), although the elongate condition of the dorsal ramus of the premaxilla extends further posteriorly along the naris rim than in Tyrannosaurus (Brochu, 2003), Tarbosaurus (Maleev, 1974), and Gorgosaurus (Lambe, 1917). The elongate condition of the dorsal and ventral processes of the premaxilla act to orient the external naris more horizontally than in large derived taxa such as Tyrannosaurus (Brochu, 2003) and Tarbosaurus (Maleev, 1974; Hurum and Sabath, 2003), which exhibit an external naris that descends anteroventrally at a higher angle. Unlike in Tarbosaurus, which has an almost horizontal premaxilla-nasal suture that is immediately dorsal to the most anterior point of the external naris (Maleev, 1974; Hurum and Sabath, 2003; SLB pers. obs. PIN 4216/3), the same contact in Q. sinensis is positioned further posteriorly at around the mid-point of the dorsal surface of the external naris. The Tarbosaurus condition is also characteristic of Tyrannosaurus (Osborn, 1912), whereas albertosaurines (Carr, 1999; Currie, 2003) and the more basal large-bodied tyrannosauroid Bistahieversor (Carr and Williamson, 2010) are more similar to Qianzhousaurus. Anterior to the external naris, the lateral surface of the premaxilla is pierced by an accessory foramen (Fig. 2). This foramen is present on both the left and right premaxillae, although on the right bone the foramen is split into two smaller circular features, whereas it is a single, larger, ovoid foramen on the left. This feature is present in Tyrannosaurus (Brochu, 2003), Tarbosaurus (SLB pers. obs. PIN 4216/3), and Gorgosaurus (Lambe, 1917), for example.

The premaxilla houses four alveoli that are oriented mediolaterally, forming a wide U-shaped snout in which the teeth are all visible in anterior view, as is characteristic of tyrannosaurids (Brusatte et al., 2010; Carr and Williamson, 2010).

Maxilla—Both the right and left maxillae are present (Fig. 1) and well-preserved and consist of three processes: the main body (Fig. 2), the ventral ramus (Fig. 2) and the posterior ramus (Fig. 2), the latter two of which diverge from the body to form the anterior margin of the antorbital fenestra (Fig. 1).

The maxillae are elongate and low relative to most other tyrannosaurids, as is the case in A. altai (Brusatte et al., 2012:figs. 6, 7) and in the reconstructed cranium of A. remotus (Kurzanov, 1976). It is the elongate maxilla, plus nasal and anterior palate, that largely define the longirostrine skull morphology of these taxa. Juvenile tyrannosaurids generally have longer and lower maxillae than adults, the latter of which have deeper craniofacial regions (e.g., Carr, 1999).

The main body of the maxilla (Fig. 2) is convex along the tooth row, as in derived large-bodied tyrannosauroids (e.g., Sereno et al., 2009; Brusatte et al., 2010; Carr and Williamson 2010; Brusatte and Carr, 2016), although it is considerably less convex than in Tyrannosaurus and Tarbosaurus. The lateral surfaces of both left and right maxillae are rugose, more so than in A. altai (Brusatte et al., 2012) (Fig. 2), but not to the same degree as in Tyrannosaurus and Tarbosaurus (e.g., Brochu, 2003; Hurum and Sabath, 2003). Minor rugosity exists towards the ventral margin of the maxilla and is accompanied by two rows of neurovascular foramina; one that follows the ventral margin of the tooth row and one that traverses centrally within the main body (Figs. 1, 2). There is an accumulation of these foramina slightly anterior to the maxillary fenestra, as is the case in Tyrannosaurus and Tarbosaurus (Hurum and Sabath, 2003).

The remaining two processes, the dorsal and ventral rami, are elongate, and extend posterodorsally and posteriorly, respectively. The dorsal ramus (Fig. 2) is bordered dorsally by the nasal for most of its length, and contacts the lacrimal posteriorly. In this way, the dorsal ramus of the maxilla forms the anterior portion of the dorsal margin of the antorbital fenestra.

The ventral process (Fig. 2) forms the posterior end of the maxillary tooth row before continuing as a thin projection that contacts the jugal at approximately the midpoint of the ventral rim of the antorbital fenestra. As the ventral ramus extends posteriorly, it loses the convex profile that begins on the ventral margin of the main body and becomes shallowly concave towards its posterior tip.

The maxillary fenestra (Figs. 1, 2) is located between the body of the maxilla and the ascending process. The fenestra is large and elongate, in the shape of an oval, which is also the case in Alioramus altai (Brusatte et al., 2009, 2012), but which differs from the proportionally much less anteroposteriorly ovoid, and more circular, shape in other tyrannosaurids. Anterior to the maxillary fenestra of Qianzhousaurus is a depression, which houses a small promaxillary fenestra (Figs. 1, 2), which is visible in lateral view and not concealed by the ascending ramus. This fenestra is smaller, and located much closer to the maxillary fenestra, than in Alioramus altai (Brusatte et al., 2012). There is no accessory depression located posterodorsal to the promaxillary fenestra; this additional opening is autapomorphic of Alioramus altai (Brusatte et al., 2009, 2012).

The ascending ramus of the maxilla is pierced, on both sides of the skull, by an elongate, ovoid accessory pneumatic fenestra, posterodorsal to the maxillary fenestra (Figs. 1, 2). This is not present in Alioramus altai, which instead possesses a shallower depression in this area, nor in any other tyrannosauroids, and thus was considered autapomorphic for Qianzhousaurus (Lü et al., 2014). The accessory fenestra is shaped and positioned differently on the two sides of the skull: on the left it is more ovoid and closer to the dorsal edge of the maxilla ascending ramus, on the right it is more of a discrete rectangular window that is positioned more centrally on the ascending ramus. Finally, we note an additional pneumatic feature of the maxilla that was not mentioned by Lü et al. (2014): a small fossa positioned between the anterior margin of the antorbital fenestra and the posterior margin of the maxillary fenestra (Figs. 1, 2). This is not present in Alioramus altai (Brusatte et al., 2012). In the derived tyrannosaurines Tyrannosaurus (Brochu, 2003), Tarbosaurus (Hurum and Sabath, 2003), and Zhuchengtyrannus (Hone et al., 2011), there is an accessory fenestra which pierces the maxilla in this region (Brusatte and Carr, 2016:character 328). Qianzhousaurus does not have a fenestra, but the fossa may be homologous.

The maxillary tooth row (Fig. 3C, D) includes 15 alveoli, which are incrassate in shape, with a labiolingual width 60% or greater than mesiodistal length. This corresponds to the intermediate condition of a phylogenetic character (Brusatte and Carr 2016: character 201 and references therein), shared with tyrannosaurids such as Albertosaurus and Daspletosaurus, but differing from the much thinner alveoli of Alioramus altai and the much thicker alveoli of Tyrannosaurus and Tarbosaurus, whose width is nearly equal to length. Thus, the dentition of Qianzhousaurus is moderately incrassate. The alveoli in Qianzhousaurus possess strongly convex mesial and distal margins, and straight to slightly convex labial and lingual margins. This is unlike the almost rectangular alveoli in A. altai, which are pinched labiolingually (Brusatte et al., 2012).

Nasal—The left and right nasals (Figs. 1–4) are fused as in tyrannosauroids generally, such that the internarial suture (Fig. 4) is almost completely obliterated, and the conjoined nasals form a vaulted structure as in derived tyrannosauroids (Holtz, 2001; Snively et al., 2006). In concert with the overall lengthening of the snout, the nasals are anteroposteriorly elongate and shallow dorsoventrally, compared with the deeper and anteroposteriorly stouter nasals in large, deep-snouted tyrannosaurids such as Tyrannosaurus (Brochu, 2003) and Tarbosaurus (Maleev, 1974).

The nasals overlie nearly the entire length of the maxilla, and the long, nearly straight suture is clearly marked in lateral view (Fig. 2). The nasal-maxilla suture is tilted in an anteroposterior-posterodorsal direction, but the angle of incline is much lower than in large, deep-snouted tyrannosaurids like Tyrannosaurus (Brochu, 2003) and Tarbosaurus (Maleev, 1974; Hurum and Sabath, 2003). In lateral view (Fig. 4D), the dorsoventral depth of the nasal decreases steadily in a posterior direction, tapering into a thin plate of bone that contacts the lacrimals, prefrontals, and frontals. This contact is visible nearly exclusively in dorsal view (Fig. 3A, B), and is described below. As in Alioramus altai (Brusatte et al., 2009, 2012; Gold et al., 2013) and deep-snouted tyrannosaurids, there is no evidence of external pneumaticity on the lateral surface of the nasal, nor does the smooth antorbital fossa extend onto the nasal. In more basal tyrannosauroids, however, the nasals are internally pneumatized, via an external pneumatic foramen within a portion of the antorbital fossa that extends onto the nasal (e.g., Eotyrannus: Hutt et al., 2001; Dilong: Xu et al., 2004; Guanlong: Xu et al., 2006; Yutyrannus: Xu et al., 2012).

Anteriorly the nasals bifurcate to form the posterior corner of the external naris. The bifurcation is comprised of two projections. First, a large premaxillary process extends anteriorly to meet the dorsal ramus of the premaxilla, forming the dorsal border of the naris (Fig. 4). Second, a short subnarial process continues ventrally underneath the naris to overlap the maxilla, forming a small portion of the floor of the naris (Fig. 4). On the left side of the skull, the subnarial process does not appear to meet the premaxilla anteriorly, leaving the maxilla to form much of the narial floor, but this may be an artifact of breakage. On the right side, the subnarial process either meets or nearly meets the premaxilla. This contact is not evident in lateral view, where it is obscured by the lateral surface of the maxilla underneath the naris but is visible in dorsal view.

In dorsal view, the nasals are essentially a straight rod, whose mediolateral width remains nearly constant across the entire element (Fig. 4A). This is also the case in Alioramus remotus, Alioramus altai, albertosaurines and non-tyrannosaurid tyrannosauroids (Brusatte et al., 2010; Brusatte and Carr, 2016:character 41), differing from the condition in most Daspletosaurus, Tyrannosaurus, and Tarbosaurus specimens, where the conjoined nasals taper in width posteriorly. With that said, there is some variation within the ‘straight’ condition: the nasals are properly straight in Qianzhousaurus, with nearly linear lateral margins on left and right sides, whereas in both Alioramus species the lateral margins are moderately concave due to slight anterior and posterior expansions, giving the fused nasal element a stretched hourglass shape in dorsal view. This properly straight condition might be an autapomorphy of Qianzhousaurus; it was not recognized by Lü et al. (2014). We do note, however, that this may be a particularly variable feature; Gorgosaurus, for instance, is polymorphic: the standard hourglass shape is seen in some specimens (e.g., ROM 1247), whereas at least one specimen is straight (CMN 2120; Carr, 1999).

Posteriorly, the nasals contact the lacrimals, prefrontals, and frontals. In comparison to large, deep-snouted tyrannosaurids like Tyrannosaurus (Brochu, 2003) and Tarbosaurus (Maleev, 1974; Hurum and Sabath, 2003), the nasal-lacrimal contact in Q. sinensis is exceedingly dorsoventrally thin in lateral view. Thus, this suture is best seen in dorsal view, as a nearly straight line abutting the medial margin of the lacrimal (Fig. 3A, B). At their most posterior extent, the nasals splay into two short rami on the left and right sides of the conjoined element (Fig. 4A), separated by a medial region of jagged interlocking sutures. These features—visible only in dorsal view—comprise the contacts with the frontals (Fig. 3A, B) and small prefrontals (Fig. 4). The two rami in Q. sinensis are longer and project further posteriorly than those of A. altai (Brusatte et al., 2012:fig. 10), in which the rami are indistinguishable from the jagged central region between them; this, however, may be an artifact of breakage in the latter taxon.

The most salient feature of the nasals is the extensively rugose surface texture, particularly on the dorsal surface (Fig. 4). Among the overall roughened and blistered texture, a series of four pronounced, rugose mounds stand out from the dorsal surface of the midline of the fused nasal element. The sculptured nasal surface is seen in other tyrannosaurids, but is especially pronounced in alioramins, and the discrete midline rugosities are a synapomorphy of Alioramini, seen also in Alioramus remotus and Alioramus altai (Lü et al., 2014; Brusatte and Carr, 2016). The midline rugosities are larger, smoother, and more hillock-shaped compared with the slightly smaller, rugged mounds of the Alioramus altai specimen described by Brusatte et al. (2009, 2012), although this may be because the Qianzhousaurus holotype is more mature. Another nasal indicator of greater maturity in the Qianzhousaurus holotype compared with the Alioramus altai holotype is the more open suture between the left and right nasals in the latter, which continues farther anteriorly from the posterior end of the bone as a cleft (see the supplemental information of Lü et al., 2014).

Lacrimal—Both left and right lacrimals are present. The left lacrimal (Fig. 1) is nearly complete, but the right (Fig. 1) is broken along its ventral ramus, leaving only the dorsal portion of the bone, plus a small portion ventrally articulated with the maxilla and jugal. As in Alioramus altai and some other derived tyrannosaurids such as Tyrannosaurus (Molnar, 1991; Brochu, 2003) and Tarbosaurus (Hurum and Sabath, 2003) the lacrimal in Qianzhousaurus is shaped like the number 7, with elongate anterior (Fig. 5) and ventral rami (Fig. 5) that meet at an acute angle. These rami define the posterodorsal region of the antorbital fenestra, and the ventral ramus separates the fenestra from the orbit. There is also a small posterior ramus (Fig. 5) that projects backwards to form the anterodorsal corner of the orbit.

The ventral ramus (Fig. 5) extends anteroposteriorly when in articulation with the other skull bones, differing slightly from the more vertical orientation in Alioramus altai (Brusatte et al., 2009, 2012), Tyrannosaurus (Brochu, 2003), Tarbosaurus (Hurum and Sabath, 2003), and other large tyrannosaurids (e.g., Currie, 2003). This may be a potential autapomorphy of Qianzhousaurus, but it is difficult to rule out the effect of taphonomic distortion and ontogenetic variation (Tsuihiji et al., 2011; Carr, 2020).

The lacrimal is highly pneumatic, as is standard for tyrannosauroids (e.g., Gold et al., 2013). The most prominent pneumatic feature externally is a large lacrimal recess (Fig. 5), a funnel-like opening at the posterodorsal corner of the antorbital fossa, where the lacrimal anterior and ventral rami diverge. It is deep and well-defined, and partitioned internally by an arcuate ridge, as in Alioramus altai (Brusatte et al., 2012). There is also at least one accessory recess anterior to the lacrimal recess on the anterior ramus (Fig. 5), but poor preservation makes it difficult to determine if there were one or two accessory openings; this feature is highly variable phylogenetically and ontogenetically in tyrannosaurids (Brusatte et al., 2012).

Perched above the lacrimal recess, on the dorsal surface of the bone, is a pronounced cornual process, or ‘lacrimal horn’ (Figs. 5, 6). It projects markedly from the skull roof, as a large and rugose mound, whose peak is approximately dorsal to the level of the ventral ramus. It also is swollen laterally, such that it overhangs the orbit in dorsal view. There is great variability in the shape and position of this process in tyrannosauroids. It is expressed as a small, conical horn in Alioramus altai (Brusatte et al., 2009, 2012); compared with this condition, the cornual process of Qianzhousaurus is larger, more swollen, and more laterally overhanging, which is likely a sign of greater maturity of the Qianzhousaurus holotype vs. the Alioramus altai holotype (Lü et al., 2014). In adult Tyrannosaurus (Brochu, 2003) and Tarbosaurus (Hurum and Sabath, 2003), however, the cornual process is essentially absent, as it is so inflated that it has merged with the remainder of the dorsal lacrimal (Brusatte and Carr, 2016:character 316).

Jugal—Both left (Figs. 1, 3–5) and right (Figs. 1, 2) jugals are present but incomplete. The jugal is composed of three main processes: an anterior ramus (Fig. 5) that contacts the lacrimal and maxilla, a dorsal ramus (Fig. 5) that articulates with the postorbital to form the posterior border of the orbit, and a posterior ramus (Fig. 5) overlapped by the quadratojugal (Fig. 5), where the two bones form the floor of the lateral temporal fenestra (Fig. 3A, B).

There are two prominent features on the jugal. The first is the cornual process (Figs. 5, 6), located along the ventral margin of the bone underneath the orbit. This structure is present in tyrannosauroids generally, but is especially pronounced in tyrannosaurids (e.g., Brusatte et al., 2010; Brusatte and Carr, 2016: character 76; Carr and Williamson, 2010). In addition, Alioramus altai was described as having an autapomorphic condition in which there is also a laterally projecting ‘horn’ on the jugal, above the standard cornual process (Brusatte et al., 2009, 2012). Lü et al. (2014) described Qianzhousaurus as ‘lacking’ the autapomorphic horn of Alioramus, but this is misleading. In fact, Qianzhousaurus does have a laterally projecting rugosity (Figs. 5, 6), but compared with the condition in Alioramus, it is larger, more swollen, more ridge-like than conical, more expansive anteroposteriorly, and it is peaked at its midpoint, so that only this portion projects strongly laterally. Although the rugosity of Qianzhousaurus does not have the discrete horn-like shape of Alioramus, we consider it a homologous structure, and note that discrete lateral projections are absent in other tyrannosaurids. In Qianzhousaurus, the surface of the rugosity is scoured by dorsoventrally inclined lineations, and associated with several foramina, for neurovasculature (Fig. 6C).

The jugal of Qianzhousaurus is pneumatized, but external pneumatic fossae and foramina are difficult to describe due to breakage and crushing. On the better preserved right jugal, there is a pneumatic recess on the lateral surface of the anterior ramus (Fig. 1B), in the posteroventral corner of the antorbital fossa, as in Alioramus altai and other tyrannosauroids, but its shape and structure are unclear. The jugal is generally flat and not inflated laterally, a sign that the internal recess does not voluminously fill the bone; this is also the case in Alioramus altai and juveniles of Bistahieversor and Gorgosaurus, whereas juvenile and adult Tyrannosaurus, Tarbosaurus, and Daspletosaurus exhibit inflated jugals (Brusatte et al., 2012).

The jugal widely contributes to the ventral floor of the orbit, which is weakly concave and approximately level with the jugal-lacrimal contact. This is also the case in Alioramus altai and tyrannosauroids generally, whereas the largest deep-skulled species such as Tyrannosaurus and Tarbosaurus have a deeper, more U-shaped orbital floor that extends further ventrally than the jugal-lacrimal suture (Sereno et al., 2009; Brusatte et al., 2010; Brusatte and Carr, 2016:character 79). Posterior to the orbit, the dorsal ramus of the jugal meets the postorbital via an elongated, anteroventral suture (Fig. 5). The dorsal ramus (Fig. 5) of the jugal is particularly anteroposteriorly broad and mediolaterally flat, compared with the thin and tapering dorsal ramus of Alioramus altai (Brusatte et al., 2012). These taxa—or at least their holotypes—differ in another important way: in Alioramus altai there is a shallow notch for the postorbital on the jugal dorsal ramus, whereas in Qianzhousaurus there is a deep, interlocking notch for the postorbital (Brusatte and Carr, 2016:character 72). In sum, Qianzhousaurus has a thicker, broader postorbital wall—comprised of the jugal dorsal ramus and postorbital ventral ramus —than Alioramus altai, which is probably a sign of greater maturity of the holotype.

Postorbital—Left (Figs. 1–5) and right (Fig. 1A, B) postorbitals are present and well preserved (Fig. 4). They exhibit the characteristic T-shape of most theropods, consisting of anterior, posterior, and ventral processes. In general, the postorbital of Qianzhousaurus is a stouter bone with thicker processes compared with the more gracile postorbitals of the holotypes of Alioramus altai (Brusatte et al., 2009, 2012) and Alioramus remotus (Kurzanov, 1976), and more closely resembles the postorbitals of mature individuals of large, deep-skulled taxa such as Albertosaurus (Currie, 2003), Tarbosaurus (Hurum and Sabath, 2003), and Tyrannosaurus (Molnar, 1991; Brochu, 2003).

The most distinctive aspect of the postorbital is the enlarged cornual process (Figs. 5, 6), which is expressed as a large swelling in the region where the anterior and ventral processes meet, above the posterodorsal corner of the orbit. It projects above the skull roof as a convex boss, as is the case in mature specimens of all other tyrannosaurids, but differing from the much gentler cornual processes of the Alioramus altai and Alioramus remotus holotypes, which are restricted to a raised rim at the posterodorsal corner of the orbit (Sereno et al., 2009; Brusatte et al., 2010; Brusatte and Carr, 2016:character 81). The surface of the cornual process in Qianzhousaurus is rugose and scored with lineations, and on the right side appears to be abutted posteriorly by a subtle fossa, although this could be an artifact of damage.

The anterior process (Fig. 5) extends forward and sweeps medially to contact the frontal, above the orbit. In dorsal view (Fig. 3A, B), the contact between the postorbital and frontal is anteroposteriorly long, and the trace of the suture is a triangular medial edge of the postorbital slotting into a concavity on the frontal. Much of the dorsal surface of the postorbital anterior process is excavated by the smooth supraorbital fossa, which extends medially and posteriorly onto the dorsal surface of the frontal. Viewed dorsally, the postorbital approaches the lacrimal but does not contact it, allowing the frontal to form the middle of the dorsal orbital margin. The frontal contribution to the orbital border is a small notch, as in Alioramus altai, Teratophoneus, Albertosaurus, and Gorgosaurus. In contrast, basal tyrannosauroids have a much broader frontal contribution to the orbital margin, whereas in derived deep-skulled tyrannosaurines such as Daspletosaurus, Tarbosaurus, and Tyrannosaurus the postorbital and lacrimal contact to exclude the frontal from the orbital margin, sometimes associated with a separate palpebral ossification (Sereno et al., 2009; Carr and Williamson, 2010; Brusatte et al., 2010; Brusatte and Carr, 2016:character 120).

The posterior process of the postorbital (Fig. 5) projects nearly straight posteriorly, without the dorsal curvature of Alioramus altai (Brusatte et al., 2012), to contact the squamosal to form the dorsal bar of the lateral temporal fenestra. The postorbital contribution to this bar is limited: the posterior end of the posterior process does not reach the posterior margin of the lateral temporal fenestra, but terminates anterior to it, at about 70% of the anteroposterior length of the fenestra. In this sense, Qianzhousaurus is more like Albertosaurus and Gorgosaurus than Alioramus altai and other tyrannosaurids, in which the posterior processes reach the back end of the fenestra (Brusatte and Carr, 2016:character 85; note Qianzhousaurus is scored incorrectly). At the corner where the ventral process and posterior process meet, a small tab-like flange of bone (Fig. 5) with rugose surface texture protrudes into the lateral temporal fenestra. This is not present in Alioramus altai (Brusatte et al., 2012), but is seen in some mature tyrannosaurids (e.g., Albertosaurus: Currie, 2003:fig. 8).

The ventral process of the postorbital (Fig. 5) overlaps the dorsal ramus of the jugal to form the postorbital bar. The ventral process in Qianzhousaurus is intermediate in morphology between that of Alioramus altai and that of other derived tyrannosaurids such as Albertosaurus, Gorgosaurus (Currie, 2003), Tyrannosaurus (Molnar, 1991; Brochu, 2003), and Tarbosaurus (Hurum and Sabath, 2003). In Alioramus altai, this process is gracile, tongue-shaped, projects nearly straight ventrally, does not reach the floor of the orbit, and ends in a broadly rounded tip (Brusatte et al., 2012). In mature tyrannosaurids the ventral process is broad, oriented strongly anteroventrally, reaches or nearly reaches the orbital floor, and near its end gives rise to a suborbital (= subocular) process that projects into the orbit as a flange (e.g., Osborn, 1912; Brochu, 2003; Hurum and Sabath, 2003). In Qianzhousaurus, the ventral process does not reach the orbital floor, but it is broad and projects anteroventrally at a lower angle than in mature deep-skulled tyrannosaurids. The distal end of the ventral process has a suborbital process (Fig. 5), but it is incipient and takes the form of a small tab (Brusatte and Carr, 2016:char-acter 86). Furthermore, the suborbital process is positioned at the ventral end of the ventral process, as in Bistahieversor, Albertosaurus, Gorgosaurus, and Teratophoneus, rather than set off from the ventral tip by a notch as in the derived deep-skulled tyrannosaurines Daspletosaurus, Tarbosaurus, and Tyrannosaurus (Brusatte and Carr, 2016:character 87).

Squamosal—Both the left and right squamosals are present; the left bone is complete and well preserved (Fig. 5), but the right is broken ventrally (Fig. 1A, B). The squamosal consists of three main processes visible in lateral view: an anterior process (Fig. 5) that forks anteriorly to meet the postorbital, a ventral process (Fig. 5) that extends into the lateral temporal fenestra to join the quadratojugal, and a posterior process that forms the posterodorsal corner of the skull (Fig. 5). In dorsal view, there is also a medial process that extends medially to contact the parietal and potentially the quadrate, forming the posterolateral region of the supratemporal fenestra.

The anterior process projects forward, with a slight dorsal sweep, to clasp the postorbital posterior process above the lateral temporal fenestra. The clasp is defined by dorsal and ventral prongs on the squamosal anterior process, both of which project anteriorly so that they approach but do not extend further than the anterior edge of the lateral temporal fenestra. This is incorrectly reconstructed by Lü et al. (2014:fig. 1), who showed both prongs as short structures terminating near the midpoint of the lateral temporal fenestra. In Qianzhousaurus the ventral prong (Fig. 5) is widely visible in lateral view, unlike in Alioramus altai, where the prong is similarly anteroposteriorly extensive but mostly hidden behind the postorbital posterior process when viewed laterally (Brusatte et al., 2012). Compared with other tyrannosaurids, including Alioramus altai (Brusatte et al., 2012) and deep-snouted species (e.g., Maleev, 1974; Brochu, 2003; Currie, 2003; Hurum and Sabath, 2003), the dorsal prong (Fig. 5) in Qianzhousaurus is noticeably gracile and shallow dorsoventrally. The depth of this process in tyrannosaurids is caused by a flange of bone that extends dorsal to the main body of the dorsal prong (Sereno et al., 2009; Brusatte et al., 2010; Brusatte and Carr, 2016:character 97). This flange is present in Qianzhousaurus, but is much subtler than in other tyrannosaurids. This is a potential autapomorphy of Qianzhousaurus, not recognized by Lü et al. (2014).

When in articulation with the other skull bones, the ventral process of the squamosal (Fig. 5) has a long axis oriented nearly anteroposteriorly, such that it approximately parallels the trend of the anterior process. This is a potential autapomorphy of Qianzhousaurus, not recognized by Lü et al. (2014), as the angle between the anterior and ventral processes is a more widely diverging, acute angle in Alioramus altai (Brusatte et al., 2012) and deep-snouted tyrannosaurids (e.g., Brochu, 2003; Currie, 2003; Hurum and Sabath, 2003). In Qianzhousaurus, the ventral process meets the quadratojugal to protrude deeply anteriorly into the lateral temporal fenestra, bisecting it into dorsal and ventral partitions. However, the squamosal-quadratojugal protrusion does not make contact with, or nearly touch, the postorbital-jugal bar that defines the anterior margin of the fenestra. This is also the case in Alioramus altai (Brusatte et al., 2012), but conversely, in some specimens of deep-snouted tyrannosaurids such as Tyrannosaurus (Osborn, 1912; Brochu, 2003) and Gorgosaurus (Currie, 2003), such contact occurs, or nearly does. In both Qianzhousaurus and Alioramus altai (Brusatte et al., 2012), the anterior end of the ventral ramus is squared off, a condition shared with deep-snouted tyrannosaurines such as Tyrannosaurus (Brochu, 2003), but differing from the tapered point of albertosaurines (e.g., Currie, 2003) and the tyrannosaurid outgroup Bistahieversor (Carr and Williamson, 2010).

The posterior process of the squamosal (Fig. 5) projects posteriorly from the region where the anterior and ventral processes meet. The lateral surface of the squamosal here is smoothly and deeply excavated by the lateral temporal fossa, much more so than in Alioramus altai, where the fossa is a pinched crescent (Brusatte et al., 2012), but more similar to the condition in deep-skulled tyrannosaurids (e.g., Brochu, 2003; Currie, 2003; Hurum and Sabath, 2003). The posterior process of Qianzhousaurus is a large, prominent, heel-like structure, which is rounded at its distal end. In size and dorsoventral orientation it is more similar to the large processes of deep-skulled tyrannosaurids such as Daspletosaurus (Currie, 2003), Tarbosaurus (Hurum and Sabath, 2003), and Tyrannosaurus (Brochu, 2003), although in these taxa the process is more square-shaped than distally rounded. Alioramus altai exhibits a strikingly different morphology, in which the posterior process, although broken on both sides of the holotype, is clearly much smaller, thinner, and sheet-like (Brusatte et al., 2012).

The squamosal is pneumatic. In ventral view, there is a large, funnel-like pneumatic opening that leads into an interior recess, the squamosal sinus (Witmer, 1997; Gold et al., 2013). This is also the case in most other tyrannosaurids (Carr and Williamson, 2010; Loewen et al., 2013; Brusatte and Carr, 2016:char-acter 341). It is unclear, however, if the large pneumatic opening of Qianzhousaurus is associated with further, smaller pneumatic foramina. Tyrannosaurus has been described as having a single expansive opening (Brochu, 2003), but Alioramus altai has three smaller openings (Gold et al., 2013). The condition in Qianzhousaurus can probably only be determined via CT scanning, as in the aforementioned taxa. It is evident, however, that the internal squamosal sinus does not inflate the posterior process. This is also the case in Alioramus altai, Teratophoneus, and Gorgosaurus, whereas the sinus is more extensive and expands into the posterior process in Daspletosaurus, Tarbosaurus, and Tyrannosaurus (Brusatte and Carr, 2016:character 95).

Quadratojugal—The left quadratojugal is present and articulated within the skull (Figs. 1–5). As in other tyrannosaurids, this bone is L-shaped in lateral view, and is composed of two main processes: a dorsal process that expands dorsally to contact the ventral ramus of the squamosal to protrude into the lateral temporal fenestra, and an anterior process that is clasped by the jugal to form the ventral edge of the lateral temporal fenestra (Fig. 5).

The dorsal process (Fig. 5) is fan-shaped, and flares out dorsally to meet the squamosal, which it overlaps. The anterior margin of the lateral surface of the dorsal process is marked by a ridge, which is robust and extends to the dorsal margin of the bone, as in Alioramus altai (Brusatte et al., 2012), and most specimens of deep-snouted tyrannosaurines such as Tyrannosaurus, Tarbosaurus, and Daspletosaurus (Carr and Williamson, 2010; Brusatte et al., 2010; Brusatte and Carr, 2016:character 99). In contrast, Albertosaurus and Gorgosaurus have a fainter ridge that fades in strength dorsally. In Qianzhousaurus, the lateral surface of the dorsal process is deeply excavated by a smooth, concave fossa (Fig. 5), similar to Alioramus altai (Brusatte et al., 2012) and other tyrannosaurids (e.g., Hurum and Sabath, 2003).

The anterior process is slender (Fig. 5), as in Alioramus altai (Brusatte et al., 2012), and differing from the much stouter, dorsoventrally deeper processes of deep-snouted tyrannosaurids (e.g., Brochu, 2003; Hurum and Sabath, 2003), and the close tyrannosaurid outgroup Bistahieversor (Carr and Williamson, 2010). The anterior tip of the anterior process is broken, but the suture on the lateral surface of the jugal gives an indication of its shape and extent. It projected slightly anterior to the anterior edge of the lateral temporal fenestra, as is normal for tyrannosaurids (see Brusatte et al., 2012), and it ended in a rounded tip, as in Alioramus altai, Alioramus remotus, albertosaurines (Albertosaurus and Gorgosaurus), not a deeper spatulate or squared tip as in the deep-snouted tyrannosaurines Daspletosaurus, Tarbosaurus, Teratophoneus, and Tyrannosaurus (Sereno et al., 2009; Carr and Williamson, 2010; Brusatte et al., 2010; Brusatte and Carr, 2016:character 101).

In posterior view, the medial surface of the quadratojugal (Fig. 7) is seen to make a broad contact with the quadrate (Fig. 7), which is interrupted by a large, teardrop-shaped quadrate foramen (Fig. 7), whose borders are nearly equally formed by the quadratojugal and quadrate. There is a deep fossa on the quadratojugal bordering the foramen laterally, and there is a small posterior tab on the quadratojugal (Fig. 8) that wraps around the quadrate posteriorly. In these features Qianzhousaurus is identical to Alioramus altai, and similar to tyrannosaurids generally (Brusatte et al., 2012).

Quadrate—The left quadrate is complete and well preserved (Fig. 1C, D), whereas the right quadrate (Fig. 1C, D) is broken and missing the condylar region. Because the left quadrate is preserved in articulation with the surrounding skull bones, it is best seen in posterior (Fig. 7) and ventral (Fig. 8) views.

Ventrally, the quadrate expands into two condyles, between which is the saddle-shaped articulation for the lower jaw. These are indistinguishable from the condyles of Alioramus altai (Brusatte et al., 2012) in that the entire condylar region is rectangular in ventral view, the medial condyle (Fig. 8) is slightly thicker anteroposteriorly, and the medial condyle projects further ventrally. Dorsal to the condylar region is the quadrate foramen (Fig. 7), described above, above which the quadrate forms a shaft that terminates in a rounded head, which is mostly obscured in posterior view.

In ventral view, two salient features of the quadrate are visible. First, there is a thin flange that projects anteriorly from the condylar region and shaft (Fig. 8). Most of the medial surface of the flange is freely exposed before it braces against the pterygoid at its anterior tip, and this medial margin is strongly concave, where it may have sat against an air sac (Currie, 2003), or perhaps provided muscle attachment. Second, at the corner where the anterior flange diverges from the condyles there is a deep, funnel-like opening that leads into the interior of the quadrate (Fig. 8). This is the external foramen for the quadrate recess, a pneumatic structure that fills the bone internally. This recess is present in Alioramus altai (Brusatte et al., 2012; Gold et al., 2013), and tyrannosaurids generally (e.g., Brusatte and Carr, 2016:character 106). Without CT data, however, we are unable to confirm the expanse of the recess inside the bone, or whether there were partitioned internal chambers inside the funnel-shaped opening like in Alioramus altai. It is clear, though, that there is no pneumatic foramen on the posterior surface of the quadrate shaft associated with the recess. Such a foramen is present in the basal tyrannosauroid Dilong and other coelurosaurs, but is absent in tyrannosaurids (Gold et al., 2013; Brusatte and Carr, 2016:character 318).

Palatine—Portions of the left and right palatine are present but broken (Fig. 9). They are best visible in ventral view, where they are in articulation on the skull midline, medial to the posterior end of the maxillary tooth row. Both left and right palatines can also be seen laterally through the window of the antorbital fenestra, but they are highly damaged; the left is slightly better preserved in this view. From what is preserved, the palatine appears highly similar to that of Alioramus altai (Brusatte et al., 2012).

In lateral view, it can be seen that the palatine has a maxillary process extending anteriorly to contact the medial surface of the maxilla and a jugal process posteriorly (Fig. 9B, C). Where these processes come together is the deeply indented, smooth surface of the internal antorbital fossa (Fig. 9), in the same position as in Alioramus altai (Brusatte et al., 2012). In Alioramus altai, this fossa leads into the window-like opening of the palatine recess posteriorly, which in turn expands into a pneumatic chamber that excavates much of the bone internally (Brusatte et al., 2012; Gold et al., 2013). Such pneumaticity characterizes tyrannosauroids generally (e.g., Brusatte and Carr, 2016:character 136), and although the details of the palatine recess are obscured by breakage and might require CT data to properly describe, palatine pneumaticity can confidently be inferred in Qianzhousaurus. Finally, above the antorbital fossa pneumatic region, the palatine fans out into a dorsal vomeropterygoid process (Fig. 9B, C), to contact the midline vomer, pterygoid, and opposing palatine; the size and shape of this process are not clear.

Some additional features of the palatine are apparent in ventral view. First, there is a fourth, medial process, which extends posteromedially to articulate with the pterygoid. The divergence of this process with the jugal process forms the suborbital fenestra in ventral view. Thus, the palatine of Qianzhousaurus is tetraradiate, as in Alioramus altai and tyrannosauroids generally. Also, although this region is highly damaged, it can be inferred that the suborbital fenestra was an anteroposteriorly elongate oval as in Alioramus altai and most tyrannosauroids, not circular as in the largest deep-skulled species such as Tarbosaurus and Tyrannosaurus (Brusatte et al., 2010; Brusatte and Carr, 2016:character 146). Similarly, the internal choana anteriorly (Fig. 9), between the vomeropterygoid and maxillary processes, would have been oval-shaped as in most tyrannosauroids and not circular as in the derived deep-skulled taxa (Brusatte et al., 2010; Brusatte and Carr, 2016:character 145). Finally, in Qianzhousaurus, the maxillary process is evidently elongate and thin, as in Alioramus altai, indicating that the palatine was anteroposteriorly elongate in concert with the external bones of the snout (maxilla and nasal). In contrast, the palatines of deep-snouted tyrannosaurids are anteroposteriorly short (e.g., Currie, 2003; Hurum and Sabath, 2003).

Pterygoid and Ectopterygoid—Portions of both bones are preserved on the right side of the skull, visible medial to the lacrimal-jugal contact (Fig. 3C, D). This part of the skull is so damaged, however, that these bones cannot be scored for phylogenetic characters (Brusatte and Carr, 2016), and all that can be said confidently is that the ectopterygoid makes broad contact with the jugal laterally.

Vomer—The anterior extent of the vomer is present and visible in ventral view, where it fits between the opposing maxillae on the midline (Fig. 2C, D). The vomer is clearly anteroposteriorly elongate, in concert with the elongate palatine, maxilla, and nasal, and together these bones define the stretched snout region. The vomer is lanceolate in shape and expands ever so slightly laterally on both sides towards its anterior margin, as in tyrannosauroids generally, but it does not expand into the greatly enlarged anterior diamond of Tarbosaurus (Hurum and Sabath, 2003) and Tyrannosaurus (Brochu, 2003; Sereno et al., 2009; Brusatte et al., 2010; Carr and Williamson, 2010; Brusatte and Carr, 2016:character 128).

Mandible

The right mandible is preserved with the jaw bones in articulation, but part of its middle section is missing (Fig. 10). Thus, portions of the dentary, surangular, angular, prearticular, and splenial have been lost. The genuine anterior and posterior sections of the mandible were joined during conservation of the specimen, and the missing region is now filled in with artificial material in its current condition in the Ganzhou Museum. The shape of the resulting mandible is reasonably accurate, but slightly too deep dorsoventrally in the reconstructed missing region. Also note that this break obscures the details of the external mandibular fenestra.

Dentary—The dentary is almost complete and well preserved (Figs. 10, 11). Similar to the morphology of the snout, the dentary is also gracile, elongate, and shallow as in Alioramus altai (Brusatte et al., 2009, 2012). This differs from the deeper and stouter dentaries of large, deep-skulled tyrannosaurids (e.g., Brochu, 2003; Currie, 2003; Hurum and Sabath, 2003).

The general outline of the dentary is consistent with that of other tyrannosaurids, both long-snouted and deep-skulled forms, in that it is dorsoventrally deepest posteriorly where it meets the surangular, then shallows along the tooth row, before expanding somewhat at its anterior tip. With that said, Qianzhousaurus is different from all other tyrannosaurids, including the long-snouted Alioramus altai (Brusatte et al., 2009, 2012) and deep-snouted species such as Tyrannosaurus (Brochu, 2003) and Gorgosaurus (Currie, 2003), in one notable way: there is no projecting ‘chin’ at the anteroventral corner of the dentary (Brusatte et al., 2010; Brusatte and Carr, 2016:character 172). The ‘chin’ is present as a discrete bump that extends from the corner of the dentary in deep-snouted species, and expressed as more of a low convexity that braces the dentary symphysis in Alioramus altai. In Qianzhousaurus, however, the anterior edge of the dentary smoothly merges with the ventral margin, along a rounded curve. This is an autapomorphy of Qianzhousaurus among tyrannosaurids, noted in the description of Lü et al. (2014) but not listed as part of the diagnosis of Qianzhousaurus.

In lateral view, the dorsal margin of the dentary along the tooth row is concave, as in other tyrannosaurids. The ventral margin is straight to slightly convex anteriorly, and then posteriorly becomes more strongly convex as the dentary fans out in depth. This is similar to deep-snouted tyrannosaurids such as Tyrannosaurus (Brochu, 2003), Tarbosaurus (Hurum and Sabath, 2003), Lythronax (Loewen et al., 2013), and Gorgosaurus (Currie, 2003), but differs from Alioramus altai, wherein the ventral margin of the dentary is nearly straight across its entire length, only expanding ever so slightly at its posterior end (Brusatte et al., 2009, 2012). Finally, the anterior margin of the dentary in Qianzhousaurus is gently angled at ∼45° relative to the long axis of the bone, as in Alioramus altai (Brusatte et al., 2012) and albertosaurines (Currie, 2003), but unlike the much steeper margin of many tyrannosaurines, which is oriented at approximately a right angle to the dentary axis (Carr and Williamson, 2010; Loewen et al., 2013; Brusatte and Carr, 2016: character 358). This latter condition is present, for example, in Lythronax (Loewen et al., 2013), Zhuchengtyrannus (Hone et al., 2011), Tarbosaurus (Hurum and Sabath, 2003), and Tyrannosaurus (Brochu, 2003).

In dorsal view, the tooth row of Qianzhousaurus is wide, with a convex lateral surface. Posterior to the alveoli, the dentary noticeably thins into a plate-like region, before making contact with the surangular, which overlaps the dentary laterally in this area.

The lateral surface of the dentary is rugose as in other tyrannosaurids. Two rows of neurovascular foramina run largely parallel to each other (Fig. 10). The first is positioned ventral to the alveolar margin of the dentary tooth row and follows the concave trajectory of the dorsal dentary margin. The first row is closer to the dorsal margin of the dentary anteriorly, and then sweeps ventrally as it continues posteriorly. Individual foramina are more distinct as individual round openings, and are greater in number anteriorly, and they become sparser posteriorly, anterior to an open groove located at the back of the tooth row. The groove continues onto the lateral surface of the surangular. In these respects, the dorsal row of foramina is very similar to that in Alioramus altai (Brusatte et al., 2012). The second row of foramina in Qianzhousaurus is located immediately above the ventral margin of the dentary and follows the straight contour of the bone, before dissipating posteriorly where the dentary expands dorsoventrally. As in the dorsal row, these foramina are better defined anteriorly, and become sparser and more anteroposteriorly elongate posteriorly. Furthermore, as in other tyrannosaurids, there is a cluster of several randomly arrayed foramina at the anterior end of the dentary, between the dorsal and ventral foramina rows.

In medial view, the articulated supradentary/coronoid (Fig. 10) obscures the alveoli and interdental plates of the dentary. Ventrally, however, the medial surface of the dentary is exposed and well preserved. Anteriorly, the symphyseal surface for the opposing dentary is generally smooth (Fig. 11), with some rugosities ventrally that parallel the anterior margin of the bone. This is similar to the condition in Alioramus altai (Brusatte et al., 2012), albeit the symphysis is slightly more rugose in Qianzhousaurus. In contrast, derived deep-snouted tyrannosaurines such as Tarbosaurus (Hurum and Sabath, 2003), Tyrannosaurus (Brochu, 2003), Daspletosaurus (Currie, 2003), and Zhuchengtyrannus (Hone et al., 2011) have much more strongly rugose symphyseal surfaces, which are bevelled and scoured with interlocking ridges and convexities for articulation with the opposing symphysis (Brusatte et al., 2010; Brusatte and Carr, 2016:character 173).

The other salient feature of the medial surface is the Meckelian groove (Fig. 11), which follows the long axis of the dentary across approximately the mid-depth of the bone. It is similar in morphology to that of Alioramus altai (Brusatte et al., 2012), in that it is sharp anteriorly (dorsoventrally thin and deeply incised into the bone) but posteriorly funnels out into a shallower fossa (Fig. 12), which then becomes overlapped by the splenial (Fig. 12). At its anterior tip, the groove terminates in a small anterior Meckelian foramen (Fig. 11). The position of the foramen, in the vicinity of alveoli five and six, is similar to that in Alioramus altai and tyrannosaurids generally (Brusatte et al., 2012). Finally, as seen in medial view (Fig. 12), the articular surface for the splenial along the ventral region of the dentary ramus below the Meckelian groove (and then fossa) is dorsoventrally shallow and smooth, as in Alioramus altai and albertosaurines, not so deep that it is nearly as deep as the anterior region of the fossa itself, as is the case in derived deep-snouted tyrannosaurines such as Tarbosaurus, Tyrannosaurus, Daspletosaurus, and Zhuchengtyrannus (Brusatte et al., 2010; Brusatte and Carr, 2016:character 174).

There are 18 dentary alveoli in Qianzhousaurus. This is the same number as in the holotype of Alioramus remotus (Kurzanov, 1976), but less than the 20 in the holotype of Alioramus altai (Brusatte et al., 2009, 2012). All three of these long-snouted alioramins, however, exhibit a noticeably greater tooth count than deep-snouted tyrannosaurids: Tyrannosaurus possesses 12–17 dentary teeth depending on ontogenetic stage (tooth loss occurred through ontogeny: Carr, 1999, 2020; Carr and Williamson, 2004), and no non-alioramin tyrannosaurid specimen, to our knowledge, has more than 17 dentary teeth. Thus, all three alioramins are united by the derived condition of 18 or more dentary teeth, as encapsulated in the phylogenetic character of Lü et al. (2014:character 317) and Brusatte and Carr (2016:character 315).

In Qianzhousaurus, the first two alveoli are markedly smaller than those in the midpoint of the tooth row. This is also the case in Alioramus altai, Daspletosaurus, and albertosaurines, but contrasting with the condition in Tarbosaurus and Tyrannosaurus, in which only the first alveolus is abnormally small (Brusatte et al., 2010; Brusatte and Carr, 2016:character 175). Execpt for those at the very front and back of the tooth row, the alveoli of Qianzhousaurus are rectangular in shape, but with rounded corners as seen in dorsal view. This is somewhat similar to Alioramus altai, although its alveoli are more square-shaped, with sharper corners, and many are pinched in the middle, giving them more of a figure-8 profile (Brusatte et al., 2012). These, like the maxillary alveoli, imply that the dentary teeth were incrassate in shape. As with the maxilla, all erupted teeth from the dentary are missing.

Surangular—The anterior and posterior regions of the right surangular are present, but the bone is missing its middle portion (Fig. 13). In overall shape, the surangular of Qianzhousaurus is anteroposteriorly elongate and dorsoventrally shallow, as in Alioramus altai (Brusatte et al., 2012). Adult deep-snouted tyrannosaurids, on the other hand, have a shorter and deeper surangular (e.g., Hurum and Currie, 2000; Brochu, 2003; Currie, 2003; Hurum and Sabath, 2003; Loewen et al., 2013). Thus, the long and gracile skull shape of Qianzhousaurus (and Alioramus) manifests itself in the posterior part of the lower jaw, despite the fact that the posterior part of the upper skull (jugal, postorbital, quadratojugal, squamosal) is not elongated anteroposteriorly.

The surangular is thickened dorsally; but is a mediolaterally thin plate ventrally. Its lateral surface (Fig. 13A) is smooth where it overlaps the dentary. Most of this contact is preserved, although a portion of it is missing posteriorly. There is a large gap between the two bones at this articulation, filled with sediment, which suggests that the joint would have allowed a degree of motion in life, as in Alioramus altai.

Anteriorly a long, tapering process of the surangular is seen to slot into the dentary when the two bones are in articulation, as in Alioramus altai (Brusatte et al., 2012:fig. 4). This process is short in large deep-snouted tyrannosaurids such as Gorgosaurus (Currie, 2003:fig. 2), Tyrannosaurus (Hurum and Currie, 2003: fig. 2; Brochu, 2003:fig. 40), Daspletosaurus (Currie, 2003:fig. 36), Lythronax (Loewen et al., 2013:fig. 2), and Teratophoneus (Loewen et al., 2013:fig. 3). This is related to, but not identical with, a phylogenetic character encapsulating the elongate proportions of the surangular that unites Qianzhousaurus and Alioramus into Alioramini: the length of the anterior flange (the region of the surangular anterior to the anterior margin of the external mandibular fenestra) is greater than 30% of the overall length of the bone (Brusatte et al., 2010; Brusatte and Carr, 2016:character 187). Thus, the finger-like process may be a synapomorphy of alioramins, in addition to characters relating to the overall proportions of the surangular.

The dorsal edge of the surangular is a laterally projecting surangular shelf (Fig. 13). The shelf is prominent and dorsoventrally deep as in all tyrannosaurids (Sereno et al., 2009; Carr and Williamson, 2010; Brusatte et al., 2010; Brusatte and Carr, 2016:character 180), and also extends approximately straight anteroposteriorly in line with the long axis of the surangular as in other tyrannosaurids (Brusatte et al., 2010; Brusatte and Carr, 2016:character 182). Above the shelf is a deep attachment site for the jaw adductor muscles (Fig. 14). This surface faces almost equally dorsally and laterally, as in Alioramus altai, albertosaurines, and Teratophoneus (Brusatte et al., 2010; Brusatte and Carr, 2016:character 184). In contrast, the largest deep-snouted forms such as Lythronax, Tarbosaurus, Tyrannosaurus, and some specimens of Daspletosaurus have a muscle attachment that is so deep that it takes the form of a flange dorsal to the ridge, with a muscle attachment surface nearly entirely laterally facing. Unfortunately, the surface preservation of the surangular shelf and adjoining regions in Qianzhousaurus is too poor to determine whether it possessed the series of small fossae that are phylogenetically informative in tyrannosauroids generally (Brusatte et al., 2010, 2012; Brusatte and Carr, 2016:characters 185–186).

In Qianzhousaurus, there is an expansive surangular foramen ventral to a surangular shelf (Fig. 13). As in other tyrannosaurids, the foramen is so large that it takes the form of a fenestra, ∼30% of the depth of the entire posterior half of the surangular (Sereno et al., 2009; Carr and Williamson, 2010; Brusatte et al., 2010; Brusatte and Carr, 2016:character 179). The surangular foramen abuts the surangular shelf dorsally, and the shelf projects laterally here, as in Alioramus altai and albertosaurines. In contrast, the shelf is tilted ventrolaterally to overhang the foramen in large deep-snouted tyrannosaurines such as Tarbosaurus, Tyrannosaurus, Daspletosaurus, Teratophoneus, and Lythronax (Brusatte et al., 2010; Carr and Williamson, 2010; Brusatte and Carr, 2016:character 181). In Qianzhousaurus there is a deeply incised pneumatic pocket posterodorsal to the surangular foramen (Fig. 13), as in tyrannosaurids broadly (Brusatte et al., 2010; Gold et al., 2013; Brusatte and Carr, 2016:character 183). Further details of pneumaticity in this region, including whether there is a chamber inside the surangular, await confirmation from CT scanning, as Gold et al. (2013) did for Alioramus altai.

The surangular contacts several lower jaw bones in addition to the dentary. At the posteroventral corner of the jaw it articulates with the angular (Fig. 10). It appears as if the angular overlaps the surangular here in lateral view, as in Alioramus altai (Brusatte et al., 2012). The splenial, prearticular, and supradentary/coronoid (Fig. 10) all make contact with the medial surface of the anterodorsal region of the surangular, but again the articulated nature of the jaw makes describing the details challenging. The articular, likely along with a small portion of the prearticular, latches against the medial surface of the posterior part of the surangular, at the retroarticular process. In dorsal view, the anterolateral region of the glenoid fossa for articulation with the quadrate of the upper skull is on the surangular (Fig. 14). The posteromedial part of the glenoid is on the articular; the surangular and articular components each make up approximately half of the glenoid joint.

The retroarticular process of the surangular, which extends posteriorly beyond the glenoid as a flange (Figs. 13–14), is anteroposteriorly short as in tyrannosauroids (e.g., Holtz, 2001; Brusatte et al., 2014). It is triangular in shape, with a posteroventrally sloping posterior margin, similar to Alioramus altai (Brusatte et al., 2012) but differing from the more vertical posterior margin of deep-skulled tyrannosaurids, such as Gorgosaurus (Carr, 1999; Currie, 2003), Daspletosaurus (Carr, 1999), Tarbosaurus (Hurum and Sabath, 2003), Tyrannosaurus (Brochu, 2003), and Lythronax and Teratophoneus (Loewen et al., 2013). This distinction has not been encapsulated into a phylogenetic character in previous tyrannosauroid datasets, but the posteroventral condition may be a synapmorphy of alioramins. It is, however, also present in some juveniles of deep-skulled species like Gorgosaurus (Carr, 1999), so its phylogenetic utility may be problematic.

Angular—The right angular is mostly intact, but is missing its anterior end (Fig. 10). The bone consists of two processes: a shallow anterior process (some of which is missing) and a deeper, plate-like posterior process, that likely overlaps the surangular. The ventral margin of the bone is smoothly convex where these processes meet, as in Alioramus altai (Brusatte et al., 2012), albertosaurines (Currie, 2003), and close tyrannosaurid outgroups such as Bistahieversor (Carr and Williamson, 2010). In contrast, deep-skulled tyrannosaurines such as Tyrannosaurus (Brochu, 2003), Tarbosaurus (Hurum and Sabath, 2003), and Teratophoneus (Loewen et al., 2013) have a kinked angular, wherein the anterior region is flexed relative to the posterior region, such that there is a discrete step between them (Brusatte and Carr, 2016:character 188; Brusatte et al., 2010). Details of the contacts with the prearticular, splenial, and possibly the articular are obscured by the articulation between the bones.

Articular—The right articular (Fig. 14) is preserved in firm articulation with the surangular laterally and prearticular medially, and apparently were tightly joined in life permitting very little motion between them. The dorsal surface of the articular is visible, and several key details of the glenoid and retroarticular process are apparent. Given that the articular is not currently known and described for Alioramus (Brusatte et al., 2012), Qianzhousaurus therefore provides the first information on the details of this bone in a long-snouted alioramin.

In Qianzhousaurus, the articular forms the posteromedial half of the smooth glenoid joint for the quadrate with surangular comprising the anterolateral half. The retroarticular process of the articular is short, as in tyrannosauroids generally, and consistent with the adjacent short retroarticular process of the surangular, described above. The joint surface of the glenoid (Fig. 14) essentially abuts the concave jaw muscle attachment site on the dorsal surface of the retroarticular process, with only a thin ridge between them. This is similar to other tyrannosaurids, but differs from the condition in the most basal tyrannosauroids and other theropods, in which there is a wide, smooth, non-articular surface between the glenoid and muscle attachment fossa (Brusatte et al., 2010; Rauhut et al., 2010; Brusatte and Carr, 2016:character 190). Furthermore, the mediolateral width of the muscle attachment is greater than the width of the glenoid, another feature shared with other tyrannosaurids but contrasting with the mediolaterally narrow muscle fossa in the most basal tyrannosauroids and other theropods (Brusatte et al., 2010; Rauhut et al., 2010; Brusatte and Carr, 2016:character 189).

The jaw muscle attachment fossa on the retroarticular process faces posterodorsally in Qianzhousaurus. This is consistent with the posteroventrally sloping posterior margin of the retroarticular process of the surangular in lateral view. This suite of characters, as noted above, may characterize alioramins, as deep-skulled tyrannosaurids have more vertical posterior edges of the retroarticular process, and thus more strictly posteriorly facing jaw muscle attachment sites on the articular. At present, however, it is unclear if this variation relates to ontogeny or phylogeny.

Prearticular—Much of the right prearticular is present on the articulated lower jaw, although the middle portion is missing (Fig. 10). As in other tyrannosaurids, the prearticular is a mediolaterally thin and arched bar of bone that spans across most of the medial surface of the surangular, and is widely visible when the jaw is observed in medial view. Anteriorly the prearticular expands into a spade-shaped anterior process, and posteriorly into a medially convex cup that embraces the articular. In between these two processes is a bar of bone; that is the shallowest part of the prearticular. The overall shape of the prearticular is somewhat intermediate between the elongate and shallowly curved profile of Alioramus altai (Brusatte et al., 2012) and the much more deeply arched shape of adult deep-snouted tyrannosaurids such as Tyrannosaurus (Brochu, 2003) and Tarbosaurus (Hurum and Sabath, 2003). Curvature of the bone is known to increase during tyrannosaurid ontogeny (Carr, 1999), in concert with the deepening of the jaw, so the phylogenetic utility of this feature is uncertain.

The medial surface of the prearticular is generally smooth. There appears to be an ovoid foramen (Fig. 10), filled with sediment, towards the anterior end of the anterior process. However, we are unable to confirm whether this is genuine (in which case it would be an autapomorphy of Qianzhousaurus among tyrannosaurids), or an artifact of taphonomic or preparation damage.

The anterior process is strap like, and tapers anteriorly at a low angle. This is similar to the condition in Alioramus altai (Brusatte et al., 2012), but contrasts with the deeper, more wedge-shaped anterior processes of adult deep-snouted tyrannosaurids.

The anterior process has complex articulations with surrounding lower jaw bones, as it contacts the surangular, splenial, and supradentary/coronoid, but apparently not the dentary. In medial view, the anterior margin of the anterior process appears to abut the splenial, but in fact the splenial medially overlaps a small portion of the prearticular here, as in Alioramus altai (Brusatte et al., 2012) and deep-snouted tyrannosaurids such as Tarbosaurus and Tyrannosaurus (Hurum and Currie, 2000; Brochu, 2003; Hurum and Sabath, 2003). Posterior to this contact, the splenial and prearticular are separated from each other by a small foramen/fenestra, but contact once more near the ventral margin of the mandible, at a simple abutting suture where the ventral edge of the prearticular sits atop the dorsal edge of the splenial. No contact between these bones is shown in this position in the reconstructed articulated lower jaws of Tarbosaurus and Tyrannosaurus (Hurum and Currie, 2000; Brochu, 2003; Hurum and Sabath, 2003); instead, the gap between the prearticular and splenial is wider, and extends to the ventral margin of the dentary, exposing the angular in medial view here. Poor preservation and breakage in most specimens make interpreting these discrete features difficult, however. The condition in Alioramus altai is not immediately clear, as the bones are not preserved in articulation, yet it has a convex notch on the ventral margin of its prearticular anterior process in the same position where a convexity on the prearticular makes contact with the splenial in Qianzhousaurus, suggesting that it shared the same morphology (Brusatte et al., 2012). Because articulated lower jaws are rare among tyrannosauroids, it is not yet clear whether alioramins share a derived morphology, or whether deep-skulled tyrannosaurids share a derived morphology, or whether this is random variation.

Further posteriorly, the prearticular makes contact with several other bones of the mandible. The bar-shaped central part of the prearticular contacts the angular laterally. Details of this joint surface are obscured by the articulated condition of the jaw bones, and breakage in this region. It is clear, however, that the ventral margin of the central bar (and the adjacent posterior process) is smooth, and lacks the rugosities of Tyrannosaurus (Brochu, 2003). The posterior process expands into a triangular shape in medial view. It clearly contacts both the articular and surangular, and almost certainly cupped the articular across most of its depth, with slightly made contact with the surangular dorsally and ventrally (based on comparisons to Alioramus altai; Brusatte et al., 2012). The degree of fusion between the bones of the prearticular and the bones of the posterior mandible is difficult to decipher, given the articulation of the jaw and the remaining matrix. Parts of the prearticular, angular, articular, and surangular are fused here in large individuals of Tarbosaurus (Hurum and Currie, 2000; Hurum and Sabath, 2003), but there is no fusion in Alioramus altai (Brusatte et al., 2012).

Articulation between the prearticular and articular occurs on the posterior face of the posterior process of the prearticular at a cup-like contact that grips the articular on its ventral face. Although the lateral surface of the prearticular is obscured, there appears to be at least some fusion between the posterior process of the prearticular with the articular and surangular, unlike in A. altai which sees no fusion with the surangular and only light articulation with the articular. The condition in Qianzhousaurus is closer in similarity with that of Tyrannosaurus and Tarbosaurus, which exhibit a similarly fused condition.

Splenial—The right splenial (Figs. 10, 11) is preserved in articulation with the surrounding jaw bones, and it is nearly complete, and only the posterior tip of the posterior process has been lost. As in all tyrannosaurids, the splenial covers much of the medial surface of the posterior dentary in medial view, and is roughly triangular and composed of three main processes: an anterior process that extends nearly straight anteriorly following the long axis of the jaw, a posteroventrally descending posterior process, and a dorsal flange (Fig. 12). The posterior process and dorsal flange diverge from each other at an acute angle in medial view, as in Alioramus altai (Brusatte et al., 2012). This angle is more obtuse, and thus the two processes separate more widely from each other, in large deep-snouted tyrannosaurids such as Tyrannosaurus (Brochu, 2003) and Tarbosaurus (Hurum and Sabath, 2003), in concert with the mediolaterally deeper jaws of these species. The shallower divergence of the processes in Alioramus altai, which manifests itself as a dorsal flange with a more horizontal long axis compared with the more anteroventral axis of deep-snouted species, was considered a possible autapomorphy of Alioramus altai by Brusatte et al. (2012). Given its occurrence in Qianzhousaurus, this suggests the character is an alioramin synapomorphy.

The anterior process is long and triangular, and is separated from the dorsal flange by a small notch, as in Alioramus altai (Brusatte et al., 2012), and other tyrannosaurids (e.g., Brochu, 2003). The anterior process covers the medial surface of the Meckelian groove (Fig. 11) and fossa (Fig. 12) on the dentary. The dorsal flange has a long, nearly straight, gently sloping anterior margin and a shorter, more steeply sloping posterior edge. It contacts both the dentary and the supradentary/coronoid laterally. The dorsal flange slightly overlaps the prearticular medially, as in Alioramus altai (Brusatte et al., 2012), whereas it is the prearticular that overlaps the splenial here in Daspletosaurus, Tarbosaurus, and Tyrannosaurus (Brusatte et al., 2010; Brusatte and Carr, 2016:character 192). This latter condition is well illustrated in Tyrannosaurus by Brochu (2003:fig. 40) and Tarbosaurus (Hurum and Sabath, 2003:fig. 20). The posterior process tapers posteriorly. It is separated from the dorsal flange by a notch, which is much larger and more open than the notch between the anterior process and dorsal flange. Compared with Alioramus altai (Brusatte et al., 2012), the notch is slightly more widely open, and more rounded. With the lower jaw bones in articulation, this notch forms the open space between the splenial and the anterior process of the prearticular.

The anterior mylohyoid foramen pierces the anterior process of the splenial. It is expressed as an anteroposteriorly elongate oval, similar to that of Alioramus altai (Brusatte et al., 2012), although slightly deeper dorsoventrally. The anteroposteriorly elongate foramen was previously considered autapomorphic of Alioramus altai (Brusatte et al., 2009, 2012), but is now known to be present in Qianzhousaurus, and is thus an alioramin synapomorphy. In contrast, deep-skulled tyrannosaurids such as Daspletosaurus (Currie, 2003), Tyrannosaurus (Brochu, 2003) and Tarbosaurus (Hurum and Sabath, 2003) have much larger, more circular foramina that are nearly as deep dorsoventrally as the entire anterior process (Brusatte et al., 2010; Carr and Williamson, 2010; Brusatte and Carr, 2016:character 191).

In Qianzhousaurus, the anterior mylohyoid foramen is entirely enclosed within the splenial (Fig. 12) as in Alioramus altai (Brusatte et al., 2012) and most tyrannosaurid specimens, but it does breach the ventral margin of the bone in most deep-snouted tyrannosauroids. The foramen is positioned higher from the ventral margin of the splenial in the Qianzhousaurus holotype, compared with the Alioramus altai holotype, supporting the hypothesis that the foramen became more enclosed within the splenial during ontogeny (Currie and Zhiming, 2001). Slightly anterior and posterior to the anterior mylohyoid foramen in Qianzhousaurus is an accessory fossa depression (Fig. 10), as also described in Alioramus altai (Brusatte et al., 2012).

Supradentary/Coronoid—The right supradentary/coronoid (Fig. 10) is present and is well preserved, in articulation with the dentary, splenial, and prearticular. Although the details of these contacts are not easily observable due to the articulated nature of the specimen. The anterior supradentary and posterior coronoid portions are firmly fused into one element. These two portions are confluent as in Alioramus altai (Brusatte et al., 2012) and Daspletosaurus (Currie, 2003), unlike Tyrannosaurus (Brochu, 2003) and Tarbosaurus (Hurum and Sabath, 2003), in which they are offset by a concave notch (Brusatte et al., 2010; Brusatte and Carr, 2016:character 195).

The supradentary region follows the contours of the tooth row, covering it (and the interdental plates of the dentary) medially. The supradentary is an elongate, shallow strip like in Alioramus altai (Brusatte et al., 2012) and most tyrannosauroids, but not a deep, crescentic bone as in Tyrannosaurus (Brochu, 2003) and Tarbosaurus (Hurum and Sabath, 2003) (Brusatte et al., 2010; Brusatte and Carr, 2016:character 194). The dorsal margin of the supradentary is much more concave in Qianzhousaurus than in Alioramus altai (Brusatte et al., 2012) and deep-snouted Daspletosaurus (Currie, 2003), Tyrannosaurus (Brochu, 2003), and Tarbosaurus (Hurum and Sabath, 2003); this is potentially an autapomorphy of Qianzhousaurus among tyrannosaurids.

The coronoid region of Qianzhousaurus is much smaller than the supradentary, and triangular in shape. Only portions of this region are visible in the articulated jaw, with many cracks in this region making it difficult to distinguish sutures from breakage. Very little of the coronoid is observable on the specimen in medial view, compared with the condition in Tyrannosaurus (Brochu, 2003) and Tarbosaurus (Hurum and Sabath, 2003), where it is visible along the dorsal margin of the jaw posterior to the prearticular. If genuine, the limited medial exposure of the coronoid may be autapomorphic among tyrannosaurids, with the caveat that the condition in Alioramus altai is uncertain because the jaw bones were found loose and not in articulation (Brusatte et al., 2012). Alternatively, it may actually be that a large portion of the coronoid is visible here, taking the form of an anteroposteriorly elongate triangular shape (Fig. 10), which projects far posterior to the prearticular. If this is the case, then the elongate coronoid, which is widely visible in medial view when the jaws are articulated, would be autapomorphic of Qianzhousaurus.