D. horneri sp. nov. Etymology Horneri, Latinized form of Horner, in honor of Jack Horner, in recognition of his successful field program in the Two Medicine Formation that has recovered many new species of dinosaurs that are critical for our understanding of the palaeobiology of dinosaurs in Laramidia, support in the preparation and curation of these specimens, and to acknowledge that his mentoring efforts have launched many professional scientific careers. Holotype MOR (Museum of the Rockies, Bozeman) 590, a complete skull, partial pectoral limb, and nearly complete hindlimb (Fig.1, see Supplementary Tables S1–S3). Paratypes MOR 1130, an incomplete skull, partial axial series, and partial pelvic girdle and hindlimb. MOR 553S/7.19.0.97, a nearly complete dentary of a small juvenile, based on four features shared with the holotype and the other paratype: 17 dental alveoli, first three alveoli form a rostromedially extending arcade, laterally bowed dentary, rostroventral corner of bone is below the level of the septum between alveoli three and four. Referred material AMNH FARB (American Museum of Natural History, Fossil Amphibians, Reptiles, and Birds, New York) 5477, a maxilla, partial postorbital, and parietal; MOR 3068, a partial mandibular ramus; MOR 553D.9.19.91, left ectopterygoid; MOR 553E.7.6.91.196, right ectopterygoid. Horizon and localities Glacier County (Co.), Lewis and Clark Co., and Teton Co., Montana, USA; Upper Cretaceous upper Two Medicine Formation. Stratigraphic Distribution The type and paratype specimens all come from the upper portion of the Two Medicine Formation 32, 33 (TMF); MOR 590 occurs 65 m below a dated bentonite horizon (TM-4) 7, and the MOR 553 specimens sit at least 10 m above this same bentonite. TM-4 occurs 480m above the base of the ~545 m–thick TMF 32, and recalibration of legacy standards on 40 Ar/ 39 Ar ages from the TMF 32 indicates that the TM-4 tuff is older than previously considered, yielding a recalibrated age of 75.03 Ma +/− 0.72 Ma 34. The fourth skeleton (MOR 1130) comes from within the disturbed belt on the eastern flank of the Rocky Mountains where folding and faulting make determining exact stratigraphic positions difficult. However, the MOR 1130 specimen sits 5.9 m above a newly dated bentonite, reported here to be 74.38 +/− 0.72 Ma (U-Pb zircon weighted mean age (1 σ; MSWD = 0.55); see Supplementary Fig.S1, Discussion S1, Table S4; for bentonite locality information see ref. 35), indicating that it is slightly younger than the other specimens. Diagnosis Can be distinguished from all other derived tyrannosauroids, including Daspletosaurus torosus, by the presence of: a wide dental arcade at the front of the snout, where the maxillary and dentary tooth rows extend distinctly rostromedially and the first interdental plate of the maxilla is narrow, which resembles those of the premaxilla where the tooth row is mediolaterally oriented; dentary distinctly bowed (convex) laterally; promaxillary sinus stopping between alveoli 3 and 4, as observed in medial view of the completely prepared pneumatic chamber; rostral end of the choana on the maxilla above alveolus 7; inflated dorsal surface of the lacrimal not reaching the medial edge of the bone; medial pneumatic recess of the lacrimal tall and narrow slot; concave upper half of orbital margin of the lacrimal; entire circumference of the pneumatic recess of the squamosal is undercut and clearly defined; sinuous rostral edge in dorsal view of the dorsotemporal fossa on the frontal; joint surface for the squamosal on the parietal covers the base of the caudolateral process; and the tympanic ridge extends onto the prootic. Several autapomorphies were obtained by the cladistic analysis; autapomorphies were not included in the data matrix, but several characters were optimized on the terminal branch of D. horneri. These include a pneumatic foramen penetrating the lateral surface of the quadratojugal, shallow notch between the basal tubera, short epipophyses of the anterior cervicals, and the humerus is ~34% the length of the femur (for further comparisons see Supplementary Discussion S2 and Supplementary Fig.S2). Previous work The so-called Two Medicine tyrannosaurine has been included in several phylogenetic analyses 2 – 4, 8, 11, 36, with scorings based primarily on MOR 590. It has universally been recovered as a derived tyrannosaurine, but its specific relationships are unresolved, where it has been recovered as the sister species of D. torosus 2, 3, 8, 36 or closer to T. rex 4, 36. Description of the taxon has been limited to brief summaries of salient features (e.g., form of lacrimal horn) that distinguish it from D. torosus and to differentiate Daspletosaurus from other genera of tyrannosaurs 1, 37, but an extensive description of the taxon has not been made. Description and comparisons The D. horneri holoype is estimated to be ~9.0 m in total length and 2.2 m tall at the acetabulum (for additional measurements, see Supplementary Tables S1, S2 and S3). D. horneri differs from its sister species in that D. torosus has a rostral ramus of the lacrimal that is longer than the ventral ramus, indicating that D. horneri has a taller skull. The antorbital fossa of the lacrimal is separated by a deep concavity from the ventral ramus of the bone in D. torosus, whereas these surfaces are confluent in D. horneri. In D. torosus, the coronoid region of the surangular faces equally dorsally and laterally, whereas it faces more laterally than dorsally in D. horneri. Daspletosaurus is one of the best-supported tyrannosaurid clades, distinguished by a suite of unique, mostly cranial features (Fig.2) that suggests a relatively long evolutionary history. In particular, the cornual processes (‘horns’) of Daspletosaurus are enhanced: a new, secondary horn extends from the side of the large, triangular primary lacrimal horn that is expressed by most tyrannosaurids; and the postorbital horn, which nearly spans the rostrocaudal width of the postorbital bar, is the largest seen among tyrannosaurids. The primary cornual process of the lacrimal differs between the two Daspletosaurus species, being taller in D. torosus (ratio height of process to maximum height of rostral ramus: 69%) than in D. horneri (ratio: 53%). Results Phylogeny. The phylogenetic analysis (see Supplementary phylogenetic character list, Tables S5 and S6) recovered 18 most parsimonious trees, each having a tree length of 802 steps, a CI of 0.56, an HI of 0.44, and an RC of 0.45. The topology conforms to that of earlier works 3, 11, but the strict consensus tree and 50% majority rule tree shows that Aviatyrannis and Proceratosauridae and the lineage leading to more derived tyrannosauroids form an unresolved trichotomy (Fig.2 A; see Supplementary Fig.S3). Also, the tyrannosaurines Lythronax, Teratophoneus, and Nanuqsaurus form an unresolved trichotomy. Alioramus and Qianzhousaurus are recovered as sister species, and given that relationship we regard the genus Qianzhousaurus as a junior synonym of Alioramus (Fig.2 A). We follow the taxonomic practice in regarding sister species as congeneric, so this renaming does not constitute a phylogenetic rearrangement of alioramins. This convention is followed elsewhere in the tree (e.g., Albertosaurus, Daspletosaurus, Tyrannosaurus). Our results recover Timurlengia as more derived than Xiongguanlong and as the sister species of a new taxon from the Iren Dabasu Formation. The latter taxon (under study by TDC) is based on a partial individual (AMNH FARB 6556) that includes several teeth (premaxillary, lateral) and skull bones (lacrimal, jugal, pterygoid, ectopterygoid, quadratojugal) that was collected in 1923 under the auspices of the AMNH. The presence of D-shaped premaxillary teeth and the presence of a jugal pneumatic recess with a secondary fossa identifies it as a derived tyrannosauroid. The absence of hindlimb bones prevents comparison with the lectotype of Alectrosaurus olseni, and so it was treated as a separate taxon in our analysis (for a complete list of unambiguously resolved synapomorphies see Supplementary synapomorphy list). Importantly, Daspletosaurus horneri is recovered as the sister species of D. torosus; this relationship is supported by 11 unambiguously optimized synapomorphies (Fig.2 B–G), including the presence of a coarse subcutaneous surface of the maxilla, an accessory cornual process on the lacrimal, a partly concealed maxillary process of the lacrimal, a cornual process of the postorbital that closely approaches the laterotemporal fenestra, a rostral tip of the squamosal that stops caudal to the rostral margin of the laterotemporal fenestra, a ridge along the nasal process of the frontal, a deep ventral keel on the vomer, a caudal pneumatic recess of the palatine that is positioned caudal to the rostral margin of the dorsal process, a distinct mediolaterally oriented ridge on the dorsum of the laterosphenoid that extends toward the medial edge of the dorsotemporal fenestra, a ‘chin’ of the dentary (where the rostral and ventral margins of the dentary join) that is positioned below the third dentary tooth, and there are more than 13 maxillary teeth. Historical biogeography. The pattern of historical biogeography indicated by our parsimony analysis is consistent with the hypothesis of Brusatte and Carr 11, where North American taxa are not divided into northern and southern subclades 4. Our topology shows an Asian diversification intermediate-grade tyrannosauroids (Xiongguanlong baimoensis, Timurlengia euotica, Iren Dabasu taxon) during the mid- and early Late Cretaceous that preceded a dispersal event to North America, which was followed by subsequent and frequent exchange between the landmasses throughout the Late Cretaceous. Ontogeny. The growth series of D. horneri includes juveniles, subadults, and an adult (Fig.3). The smallest specimen, a dentary, has a tooth row that is 221.5 mm long, in contrast to the 423.0 mm tooth row of the adult, which has a 947.0 mm long skull. The length of the dentary tooth row has been shown by Currie 38 that in comparison with skull length, it shows a weak negative allometry. Therefore, the original skull length of the small specimen was greater than 496.0mm, slightly more than half the length of the adult skull. The parsimony analysis of morphological features (see Supplementary ontogenetic character list, Data S1) resulted in a single most parsimonious growth series of 168 steps, with a CI of 0.93, an HI of 0.07, an RI of 0.93, and an RC of 0.91 (Fig.3; see Supplementary Fig.S4). The transition from a gracile juvenile to a robust adult in D. horneri is similar to that seen in other tyrannosauroids 8, 26, 28. Soft tissues. The excellent quality of preservation of these specimens permits us to assess the type of soft tissue that covered the face (premaxilla, maxilla, nasal, lacrimal, jugal, postorbital, squamosal, dentary). In D. horneri, and in all derived tyrannosauroids, the subcutaneous texture is coarse and shows a hierarchy of textures (see Supplementary Discussion S3). In order to identify the soft tissue that produced this complex surface, we compared the condition of tyrannosaurids with that of crocodylians (Alligator mississippiensis) and birds (Struthio camelus, Anser sp., Anas sp., Cygnus sp., Meleagris gallopavo), and we followed several studies 29, 31, 39, 40 to identify osteological correlates imprinted on the cortical surface of facial bones to deduce their causal soft tissues. Although tyrannosaurids, crocodylians, and birds share neurovascular foramina that are densely clustered at the front of the jaws and form rows along the oral margin, the smooth texture of the snout in birds is sharply different from the hummocky texture of the facial bones of crocodylians and tyrannosaurids. The coarse surface of crocodylians is covered by many flat scales 29, whereas in birds the rhamphotheca (beak) covered the smooth surface of the snout and jaws. The coarse texture shared between tyrannosaurids and crocodylians indicates a primary covering of flat scales on the face of the nonavian dinosaurs (Fig.4). In archosaurs, a scaly integument appears to be correlated with the multiple rows of foramina seen in crocodylians and tyrannosaurids, whereas in birds the foramina deep to the beak are limited to the jaw tips, the caudal end of the maxilla, and in a row along the side of the lower jaw. A similar localization of foramina is seen in ornithischian dinosaurs, whose jaws and oral margins were sheathed by beaks ahead of the tooth row 39. Unlike crocodylians, the alveolar row of foramina in the dentary of derived tyrannosauroids and birds occurs in a common groove that extends for much of the length of the bone. This groove is not seen in crocodylians, although caudodorsally extending sulci extend from the foramina. In birds the groove is covered by the rhamphotheca, and the ventral branches of the rictal vessels and the external branch of the mandibular nerve lie in the groove 40. The groove in tyrannosauroids is shallowly inset, unlike the sharply inset condition that is seen in birds. The presence of the groove in tyrannosauroids indicates that the ventral branches of the rictal vessels evolved prior to, and might have been prerequisites for, the later appearance of the beak (in terms of the stepwise vascular changes leading from nonbeak to beak), at least on the lower jaw. The caudal end of the groove fades out in birds, a condition that is also seen in mature tyrannosauroids 41. Variation in the coarse zone of the face in tyrannosaurids indicates a variety of epidermal types, in addition to flat scales. The rostral surface of the premaxilla, dorsum of the nasals, dorsolateral surface of the lacrimal, cornual process of the jugal, and the rostroventrolatral surface of the dentary, bear small bony papillae, which indicate regions of armor-like skin 29 (Fig.4, see Supplementary Discussion S3). Finally, the coarse and rim-like edges of the postorbital horn, and its smooth central region, indicate a cornified sheath-like covering on its surface and part of the postorbital bar (Fig.4). Discussion Anagenesis. A hypothesis of anagenesis amongst tyrannosaurids (D. torosus -> D. horneri -> T. rex) is defensible if: (1) the taxa are sister species or a phylogenetically successive series of species, (2) the species are stratigraphically sequential, (3) the phylogenetic relationships do not conflict with their stratigraphic sequence, and the taxa (4) are from the same landmass or adjacent landmasses that had connections that do not conflict with their chronological sequence. Our new phylogeny (Fig.2 A) shows that D. torosus and D. horneri are sister species, whereas T. rex is nested in a separate clade. This topology is different from the phylogenetic arrangement of previous workers 1, where D. torosus is the sister species of a clade composed of D. horneri and T. rex. In our phylogeny, T. rex is separated from the Daspletosaurus clade by two phylogenetically- and almost certainly stratigraphically-successive 4 sister species, Zhuchengtyrannus magnus and T. bataar. Chronostratigraphic and lithostratigraphic correlation between the TMF and Dinosaur Park Formation (DPF) suggests that the DPF correlates with the upper 220 m of the TMF 42 – 45. In-progress high-precision U-Pb dating of the four ash beds spanning the top of the Oldman Formation, the DPF and the lower part of the Bearpaw Formation (BPF) indicates an age range between 76.69 Ma–74.26 Ma (internal error D. torosus specimens are restricted to the lower two-thirds of the DPF (~76.7–75.2 Ma), and D. horneri specimens are limited to the uppermost part of the TMF 45 (~75.1–74.4 Ma) there appears to be little or no stratigraphic overlap between Daspletosaurus specimens from these two field areas. New high-precision CA-TIMS U-Pb dating is underway and promises to better resolve stratigraphic uncertainties. Therefore, D. torosus and D. horneri meet the primary criteria for anagenesis: They are sister species, stratigraphically successive, and are from the Northern Rocky Mountain Region. In contrast, the divergence between the Daspletosaurus clade on the one hand, and the Z. magnus + Tyrannosaurus clade on the other, was the result of a cladogenetic (lineage-splitting) event, since they do not form a continuous series of stratigraphically sequential taxa. Therefore, T. rex was not a continuation of the anagenetic Daspletosaurus lineage 1. The Z. magnus + Tyrannosaurus clade consists of a pair of sister species (T. bataar + T. rex) and a sister taxon that extends from the preceding node (Z. magnus). This topology is consistent with a hypothesis of anagenesis: Z. magnus lived no less than 73.5 Ma 47, and T. rex lived between 66.0 and ~67.2–67.4 Ma 48. The age of the Nemegt Formation, from which T. bataar has been collected, is less than 75.0 Ma based on radiometric dating of the underlying Barun Goyot Formation 49. Therefore, if T. bataar is intermediate in age between Z. magnus and T. rex, then the chronological sequence of these tyrannosaurs is also consistent with anagenesis. The time gap (~6.1– 6.3 Myrs) that separates Z. magnus and T. rex greatly exceeds that between D. torosus and D. horneri; therefore, the hypothesis of anagenesis from Z. magnus to T. rex will be tested as new tyrannosaur specimens are collected from that interval in Asia and included in new phylogenetic analyses. This is also the case for the two sister species of Albertosaurus, which are stratigraphically successive and from the same geographic region, in northern Laramidia 37, 50. If the Asian sister species Alioramus remotus and A. sinensis are shown to be stratigraphically successive (the taxa are widely separated geographically, and neither is constrained by radiometric dates 5, 10), then they meet the criteria as well, but in the absence of those data the decision between anagenesis and cladogenesis is equivocal. Based on this approach, the evidence for anagenesis might be widespread among other dinosaur lineages with high-quality fossil records. If so, then anagenesis was an important contributor towards the generation of species diversity, in addition to cladogenes