The dinosaur tracks of Tyrants Aisle: An Upper Cretaceous ichnofauna from Unit 4 of the Wapiti Formation (upper Campanian), Alberta, Canada

The Wapiti Formation of northwest Alberta and northeast British Columbia, Canada, preserves an Upper Cretaceous terrestrial vertebrate fauna that is latitudinally situated between those documented further north in Alaska and those from southern Alberta and the contiguous U.S.A. Therefore, the Wapiti Formation is important for identifying broad patterns in vertebrate ecology, diversity, and distribution across Laramidia during the latest Cretaceous. Tracksites are especially useful as they provide a range of palaeoecological, palaeoenvironmental, and behavioural data that are complementary to the skeletal record. Here, we describe the Tyrants Aisle locality, the largest in-situ tracksite known from the Wapiti Formation. The site occurs in the lower part of Unit 4 of the formation (~72.5 Ma, upper Campanian), exposed along the southern bank of the Redwillow River. More than 100 tracks are documented across at least three distinct track-bearing layers, which were deposited on an alluvial floodplain. Hadrosaurid tracks are most abundant, and are referable to Hadrosauropodus based on track width exceeding track length, broad digits, and rounded or bilobed heel margins. We suggest the hadrosaurid trackmaker was Edmontosaurus regalis based on stratigraphic context. Tyrannosaurids, probable troodontids, possible ornithomimids, and possible azhdarchid pterosaurs represent minor but notable elements of the ichnofauna, as the latter is unknown from skeletal remains within the Wapiti Formation, and all others are poorly represented. Possible social behaviour is inferred for some of the hadrosaurid and small theropod-like trackmakers based on trackway alignment, suitable spacing and consistent preservation. On a broad taxonomic level (i.e., family or above), ichnofaunal compositions indicate that hadrosaurids were palaeoecologically dominant across Laramidia during the late Campanian within both high-and low-latitude deposits, although the role of depositional environment requires further testing.


Introduction
The site was hand mapped using a baseline and a portable 1x1 m grid square. The handdrawn maps were digitised and combined into a single comprehensive site map created with Inkscape v. 0.92 . In total, the mapped area measured~1, 400 m 2 . Measurements of the best-preserved and most distinct tracks of each observable morphotype were collected on-site using a tape measure, and additional measurements were collected digitally in Inkscape from either a scaled photograph, map, or digital elevation model.
Measurements largely followed Therrien et al. [43] and Salisbury et al. [44] and, for individual footprints, included: compass bearing, track length, track width, basal digit lengths, heel to digit tip lengths, heel to hypex lengths, divarication angles, length of digit III extension beyond the tips of II and IV, widths of digit bases, and widths of digits at their mid-points (Fig 5A and  5B). For trackway sequences-defined here as two or more sequential tracks produced by the same individual-additional measurements included (where possible) pace and stride lengths, pace angulation, relative foot rotation, and compass bearing of the trackway midline (Fig 5C  and 5D).
Oriented rock samples from each track layer were also collected for thin sectioning and petrographic observation. Samples were ground down and set on top of glass slides at the University of New England in Armidale, Australia. Viewing and imaging of each slide was performed at Macquarie University in Sydney, Australia with a Nikon Eclipse 50iPOL petrographic microscope, equipped with a Nikon DS-Fi1 digital camera, and using Nikon NIS Elements-D software to manipulate image settings.
Throughout this work, individual tracks or trackways are documented using a numbering system that allows specific specimens to be referred to in-text. When referring to an isolated track, or a trackway sequence as a whole, the system uses a three-part code, which first broadly identifies the trackmaker type (H = hadrosaurid; Ty = tyrannosaurid; Di = didactyl theropod; Th = indeterminate theropod-like trackmaker; Tri = indeterminate tridactyl dinosaur; In = indeterminate trace), followed by an indication of whether the code refers to an isolated track or a trackway (I = isolated; Tw = trackway) and finally the grid location of the specimen within the overall tracksite (e.g. 8B or 32-C, the hyphen in the latter indicating that the grid is on the negative side of the baseline). For tracks that were identified outside the main study area, the location is designated simply as OG (outside gridded area). When a single track was spread across multiple grids, the grid that contained the largest portion of the track was taken as its location. For trackways, which almost always span multiple grids, the grid locations of the first and last footfalls were given as a range. Trackways were also numbered in order of their proximity to the beginning of the baseline, to distinguish between different trackways produced by the same type of trackmaker. For example, the code Th.Tw1.9B-6B pertains to the first theropod-like trackway (i.e., closest to the beginning of the baseline), whose first and final prints are located in grids 9B and 6B, respectively, while print H.I.28-B is an isolated hadrosaurid track located in grid 28-B. When referring to specific tracks within a trackway sequence, a four-part code is used, identifying the trackmaker first, followed by the trackway number, the number of the print within the sequence, and finally the grid location of the print. For example, Th.Tw3.4.24H is the fourth footprint in the third theropod-like trackway, and is located in grid 24H. The utility of these codes is that they allow any track referred to in-text to be located on the corresponding tracksite maps (Figs 1B and 2-4).

Photogrammetry and interpretive outlines
Most tracks were photographed using a Nikon D810 SLR camera and 24 mm lens. For the best-preserved examples, these images were used to reconstruct photogrammetric 3D digital elevation models (DEMs), from which final interpretive track outlines were produced. All digital models used within this study-and the photographs used to create them-are available for download within the supplementary information, in accordance with suggested practice [45,46]. To build the DEMs, images were first imported into VisualSFM v.0.5.26 [47] to produce sparse point cloud reconstructions. Dense point cloud reconstructions were created using CMVS/PMVS [48,49] and then trimmed in MeshLab version 2016.12 [50,51], following which a Poisson surface reconstruction was performed. DEMs of the Poisson meshes were produced using Cloud Compare v.2.9.1 [52] and final colour edits made in Paraview v.5.5.2 [53]. Outline interpretations of tracks were then drawn in Inkscape v.0.92 using the DEMs. These outlines, together with the DEMs, silicone moulds, on-site photographs, and track measurements, formed the basis for our descriptions of the track morphotypes.

Identification of trackmaker morphotypes
All tracks were identified to the lowest possible taxonomic level, using their overall morphological characters and contextual (i.e., chronostratigraphic) relevance. Based on skeletal material recovered within Late Cretaceous terrestrial strata of both the Wapiti Formation and more broadly in western Canada [8,12,15,16,54,55], tridactyl prints from Tyrants Aisle can be confidently regarded as belonging to either hadrosaurid, theropod or, possibly, thescelosaurid dinosaurs; tracks that possess broad, rounded digits with blunt terminations and relatively broad heels were treated as hadrosaurid tracks, while prints with relatively slender digits, sharp claw marks, and more narrow heels were regarded as theropod-like tracks [7,56]. The term "theropod-like" is used herein to accommodate the theoretical possibility that some of these tracks may pertain to thescelosaurids, which arguably produced similar track morphologies to those of theropods, making them difficult to distinguish. While thescelosaurids are presently unknown within Unit 4 of the Wapiti Formation, possible fragmentary remains of these dinosaurs have been recovered from Unit 3 [12].
Tridactyl theropod-like tracks were sorted into tyrannosaurid or indeterminate theropodlike morphotypes, primarily according to size. As tyrannosauroids are the only known largebodied functionally-tridactyl theropods within post-Cenomanian strata from North America [57,58], and given the presence of indeterminate tyrannosaurid body fossils and tracks within the Wapiti Formation [10,12,14,17], it was assumed that any tridactyl theropod-like track �45 cm in length-that does not show evidence of being a deep undertrack [59]-must belong to a tyrannosaurid maker. Deep undertracks may be considerably larger than the corresponding surface track and shallow undertracks, but can be recognised by their poorly defined, flattened track margins, and by the absence of such features as sharp claw marks, digital pad impressions, and skin impressions [60][61][62][63][64]. Our threshold value of 45 cm is based on the largest plausible foot length for non-tyrannosaurid tridactyl theropods within post-Cenomanian

PLOS ONE
The Tyrants Aisle dinosaur tracksite strata from western Canada. This value was found by Enriquez et al. [35] to be approximately 40 cm based on the largest known, fragmentary ornithomimid pedal unguals recovered from the mid-late Campanian Dinosaur Park Formation in southern Alberta [65], and scaling these with reference to complete ornithomimid pedes [66,67]. Thus, below 45 cm, we acknowledge that there is potentially size overlap between multiple major theropod clades and Thescelosauridae, and refer most of these tracks to indeterminate theropod-like trackmakers. However, some examples are tentatively identified as possible ornithomimid tracks (see 'taxonomic affinities'). Non-tyrannosaurid theropod-like tracks are here divided arbitrarily into small (track length < 25 cm), medium (25-35 cm), and large (>35-45 cm) indeterminate size classes, although there is some morphological overlap. Thus, each size class does not necessarily pertain to a distinct trackmaker. In addition to the tridactyl footprints, a collection of small, didactyl tracks from Tyrants Aisle are probably those of troodontid theropods [37].

Trackmaker hip height and speed estimation
Estimation of hip height for non-tyrannosaurid theropod-like trackmakers and facultatively bipedal hadrosaurids followed the general rule first applied by Alexander [5] and supported by Henderson [68], namely that hip height is approximately four times the length of the print.
For Where V is the velocity in metres per second, g is the acceleration due to gravity (approximately equal to 9.8 m/s 2 ), λ is the stride length in metres, and h is the hip height, also in metres.

Depositional environment, tracksite stratigraphy and lithology
Sediments from Unit 4 of the Wapiti Formation, including those at Tyrants Aisle, accumulated on floodplains situated between the newly forming Rocky Mountains to the west and the receding Western Interior Seaway further east [33,40]. While Tyrants Aisle is currently situated at 55.064˚N, its palaeolatitude within Laramidia during the late Campanian is estimated to have been~63.5˚N [69,70]. Therefore, the site records a relatively high latitude dinosaur assemblage.
At least three successive in-situ track-bearing horizons are present at Tyrants Aisle (Fig 6). The main track-bearing layers are numbered in stratigraphic order, 1 being the lowest and 3  the highest. Each has a strike that is approximately parallel with the flow of the modern Redwillow River (i.e., 235˚), with an average sub-horizontal dip of 7˚east.
The three layers are petrographically similar in thin section: grains are well sorted and typically (>90%) below 120 μm in diameter (very fine sand to silt), but infrequently range up tõ 250 μm (fine sand) (Figs 6, 7A and 7B). Clast mineralogy is dominated by quartz and feldspars (~50%, combined), which are embedded within a coarse calcitic cement (~40% of composition). Secondary iron oxides are also present (~5% of composition). Trace minerals (<1% of composition) include detrital zircons, tourmalines, chlorites, and muscovites. Clasts display low sphericity and range from angular to subrounded, the majority being subangular. These shape characteristics, in addition to the presence of elongate detrital muscovite fragments and crisp feldspars, suggest that the rocks are immature (i.e., grains had not travelled far from their source). Some grains of feldspar show partial replacement by calcite or chlorite, suggesting exposure to low heat, probably during shallow burial.
As fine-grained sediments are typically deposited under low energy, each track layer was likely laid down by slow-moving water. While Unit 4 of the Wapiti Formation is collectively a coal-rich unit [33,40], no coal seams are present within the localised stratigraphic section at Tyrants Aisle, and the track layers contain relatively low organic content. Only one carbonised log fragment was observed on track layer 3 within grids 16E-17E ( Fig 1B).

Track layer 1
Layer 1 is exposed adjacent to the Redwillow River as a narrow band of outcrop~0.2 m in vertical thickness and ranging from <1 m to 4.5 m in width across the length of the site. Tracks on this layer are restricted to metres 25-54 along the baseline (Figs 1B and 3). Only hadrosaurid tracks occur on this layer, and they are nearly all oriented at a bearing of~120-150˚and preserved as natural moulds.

"Track layer 2"
Layer 2 exhibits variable geometry, extensive rock fracturing, and vertical displacement, resulting in a series of discontinuous ledges. The precise margins and lateral equivalences of these ledges are often difficult to determine. For simplicity, and due to the rarity of tracks on these surfaces, each ledge has been lumped together as a single area of outcrop situated between layers 1 and 3. It is probable that there is more than one distinct stratigraphic level contained within this interval. Downcutting of layer 2 caused its vertical thickness to become variable across the site, as the layer is shallowest towards the start of the mapping baseline (~0.1 m at grid number 8) and thickest near the far end of the site (~0.4 m at grid number 89). Two

PLOS ONE
The Tyrants Aisle dinosaur tracksite distinct track preservation styles were observed on layer 2 horizons: natural moulds were predominant, but there was at least one convex infilling (H.I.82C).

Track layer 3
Layer 3 is the most extensively exposed stratigraphic layer at Tyrants Aisle. It is present across almost the entire length of the site and varies in exposed width from~7 m to~12.5 m. Vertical thickness of track layer 3 varies from~0.1 to~0.4 m across the site. Layer 3 is the main trackbearing horizon, with both the greatest quantity and diversity of dinosaur tracks, almost all of which are preserved as natural moulds (n�169). Many of these natural moulds have been eroded into subcircular potholes, with few morphological details remaining. At least two insitu tracks on layer 3, Th.Tw3.4.24H and H.I.74C, are preserved as convex infillings.
In addition to the tracks preserved on the three in-situ layers, at least three hadrosaurid tracks (H.I.3I; H.I.78C; H.I.81-D) are preserved as natural casts on ex-situ blocks that were evidently transported some distance by the Redwillow River. It is unclear which horizon these natural casts originate from.
Invertebrate ichnites are limited to traces of Rhizocorallium, identified based on their Ushaped burrow morphology and oblique orientation with respect to the bedding planes ( Fig  7C and 7D) [72]. These ichnites were possibly produced by crustaceans, annelid worms or mayfly larvae [72], and occur sporadically as grouped clusters of~2-6 on each of the three track layers (Figs 2 and 3). Not enough are exposed to determine if they are especially abundant within any particular layer (~15 were identified across all track layers). The burrows generally consist of a U-shaped marginal tube with interior spreiten (i.e., sedimentary laminae produced by the tracemaker while foraging and/or excavating), which join the two parallel 'limbs' of the marginal tube ( Fig 7C and 7D), although some of the burrows lack visible spreiten. These invertebrate traces are most often observed in shallow marine facies, only rarely occurring in terrestrial strata [72]. No faecal pellets are visible within the burrow casts, which have been in-filled by the same sediments that form the rest of the track layers. Rhizocorallium are useful indicators of substrate consistency, as burrows containing actively-filled spreiten that differ in composition or texture to the marginal tube indicate relatively soft substrates. Conversely, burrows containing passively-filled spreiten that are identical to the in-filling of the marginal tube indicate relatively firm substrates [72]. As spreiten compositions within the Rhizocorallium at Tyrants Aisle are generally homogenous with respect to that of their marginal tubes (Fig 7C), passive in-filling is suggested, and the original substrates were likely relatively firm at the time of trace formation. However, as some dinosaur tracks on track layer 3 show evidence for suction effects and are relatively deep (see 'indeterminate morphotype B'), a range of sediment consistencies from soft through to firm were apparently present at Tyrants Aisle, possibly reflecting variation in the timing of track formation.
Current directions can also be inferred from Rhizocorallium, as the parallel 'limbs' of the marginal tube are generally aligned (but oblique in the third dimension) with the prevailing current orientation [72][73][74]. Rhizocorallium burrows on track layer 1 are mostly aligned southeast-northwest, while those on track layer 3 are usually aligned north-south (Figs 2 and 3), consistent with prior observations that indicate a predominantly northward current flow within Unit 4 of the Wapiti Formation [33,40]. Other palaeoflow indicators from Tyrants Aisle include ripple marks on track layer 3, although these are restricted to only a small area at the western end of the site (Fig 1B). The orientation of these ripples indicates a north-easterly direction of flow in this particular area. Given that the original rivers depositing sediments on the ancient Wapiti floodplains likely meandered to some degree, variability in flow direction is expected. Overall, an average northward flow is supported by both Rhizocorallium and ripple marks.

Vertebrate ichnological diversity
Measurements of relatively well-preserved tracks and trackways at Tyrants Aisle are collated in Tables 1 and 2. To avoid repeating similar descriptions for multiple tracks, only the best-preserved examples with the most distinct and representative morphology are individually described for each trackmaker morphotype.

Hadrosauropodus isp.
H.I.28-B (exemplar print)-H.I.28-B is an isolated natural mould (Fig 8A-8C) (track length �56 cm; track width = 61 cm). The presence of an adjacent manus impression, which we assume is associated, anterolateral to digit III suggests that the track likely represents a left footprint [75]. The tip of digit III is not fully impressed. Digits II and IV gently taper to blunt points, with free lengths of 22 cm and 24 cm, and basal widths of 22 cm and 21 cm, respectively. Digit III is rounded, widest at its midpoint, and moderately constricted at the base. This basal constriction is also demonstrated in an adjacent isolated track (H.I.28-C; Fig 8A-8C), and in the holotype of Hadrosauropodus [75]. The heel impression is broad, with a weakly bilobed posterior margin. Other exemplars (e.g., H.Tw3.1.30-C and H.Tw4.2.31-B) are more prominently bilobed, and taper posteriorly into two small, blunt, typically obtuse points. Furthermore, some examples have entirely rounded heel impressions (e.g., H.Tw6.2.53A and H. Tw7.1.54A). These differences in heel shape are natural variants within Hadrosauropodus [75] and likely the result of variable substrate properties, kinematics, and preservation. In H.I.28-B, the divarication between digits II and III is 17˚, while that between III and IV is 10˚. The associated manus impression is a subrounded shallow depression (length = 11 cm; width = 18.5 cm), and overlies the tip of digit III in the adjacent track H.I.28-C. An additional subrounded manus impression is located immediately behind the heel base of H.I.28-B (Fig 8A-8C). It is unclear if this second manus imprint was created by the individual responsible for producing H.I.28-B, although this seems likely given its close proximity.
Less well-preserved Hadrosauropodus tracks are sometimes part of short trackway sequences (see 'possible trackmaker sociality'). The poor preservation of these trackways precludes a detailed description of the individual tracks, although they remain useful for inferring the locomotory characteristics of the Hadrosauropodus trackmakers. The clearest examples on track layers 1 and 3 span only a single pace or stride length, and show predominantly inward foot rotation with respect to the trackway midline (Figs 1B, 2 and 3; Table 2). As the pace lengths are typically short (average of 1.35 m [n = 7]; Table 2), it is assumed that these Hadrosauropodus trackmakers were walking. Longer possible hadrosaurid trackways were identified on track layer 3 (Figs 3 and 4; Table 2), e.g., H.Tw5.38B-31H, although tracks within this sequence are ambiguous, comprising eroded depressions of variable size that can only tentatively be attributed to the same individual.
Remarks-Tanke [8] originally observed roughly a dozen ornithopod tracks at Tyrants Aisle in 2003, and referred these to Amblydactylus, an ichnogenus of probable hadrosauroid affinity containing two ichnospecies, Amblydactylus gethingi and Amblydactylus kortmeyeri (= Caririchnium kortmeyeri sensu [76]). Both were originally erected based on tracks from the Gething Formation (Aptian) of northeast British Columbia [77,78]. This referral is not followed herein, given that tracks of Amblydactylus gethingi are longer than they are broad, while the heel impressions of the type series of Amblydactylus kortmeyeri are not prominently bilobed, and are generally convex in outline posterior to digit III [78]. In contrast, hadrosaurid tracks at Tyrants Aisle are almost always broader than they are long, and often have bilobed heel impressions, with a heel margin that is concave posterior to digit III (Fig 8A-8C).
Hadrosauropodus langstoni, the type ichnospecies of Hadrosauropodus, was originally erected based on TMP 1987.076.0006, a natural pes cast from the Maastrichtian horizons of the St. Mary River Formation in southern Alberta [60,75]. Pes tracks generically referable to Hadrosauropodus are diagnosed as tridactyl, equally wide or wider than they are long, with rounded or bilobed heel margins wider than the proximal part of digit III, and broad digits with blunt distal terminations [75,76,79]. Within trackways, Hadrosauropodus pes tracks are generally rotated inward and have short pace lengths (approx. double track length). Smaller manus impressions are also sometimes associated with pes tracks [60,75]. Based on the occurrence of the aforementioned characters in the majority of the hadrosaurid tracks at Tyrants Aisle (including H.I.28-B), as well as their appropriate geological age, we refer these tracks to Hadrosauropodus [75,76,79].

Hadrosaurid morphotype A
H.I.3I (exemplar print)-H.I.3I is a large hadrosaurid pes natural cast located on an ex-situ block ( Fig 8D-8F) (track length = 64 cm; track width = 64 cm). The heel is V-shaped, tapering to a blunt apex positioned in line with the axis of digit III. The free-length of each digit is short relative to the track length (left digit = 11 cm, digit III = 20 cm, right digit = 12 cm), pushing the hypices to a more anterior position than in typical examples of Hadrosauropodus [60,75], and also greatly increasing the relative heel area of the print. As preserved, digit III is wider at its base than digits II and IV and tapers to a blunt, rounded termination. Divarication between digit III and the left digit is 24˚, and that between digit III and the right digit is 45˚.
Remarks-This hadrosaurid track morphotype is less abundant at Tyrants Aisle (n �2) than Hadrosauropodus, and is characterised by a digit III that widens towards the base (rather than being basally constricted as in H.I.28-B), a heel that is V-shaped rather than bilobed, and a proportionally larger heel area, features that create an overall more robust track morphology (Fig 8D-8F). Despite these morphological differences, it is assumed based on the provenance of the tracks (see 'taxonomic affinities') that the same trackmaker species was responsible for producing both these prints and those referred to Hadrosauropodus. Interestingly, the two best examples of this morphotype (H.I.3I and H.I.78C) are preserved as ex-situ natural casts, which suggests that preservation style may correlate with the occurrence of this morphotype.

Tyrannosauridae
Ty.I.OG (exemplar print)-The largest theropod track at Tyrants Aisle is an incomplete, isolated natural cast or convex infilling (track length�62 cm; track width >51 cm) (Fig 9G-9H; Table 1) situated~40 metres upstream from the mapped area. Although the specimen may be in-situ, loss of outcrop in the area between it and the main tracksite prevents Ty.I.OG from being assigned stratigraphically to any of the recognised track layers. It is difficult to determine which foot produced Ty.I.OG given that it is isolated, incomplete, and its morphology was substantially distorted by kinematics or poor preservation. Digit III is sinuous, unusually broad at the base (24 cm wide) and tapers to a blunt apex that is truncated by a break. The most dorsal surface of digit III forms a thin ridge that is most prominent towards the distal end. To the left of this ridge, the upper surface of digit III slopes downward at a shallow angle, whereas the right side slopes more steeply. The left digit diverges from digit III at an angle of 52˚, and the free digit length of the former is less than half that of the latter (free length of left digit = 14.7 cm; free length of digit III = 38 cm). The right digit is truncated by a break, so that its length cannot be measured, although it appears to extend farther anteriorly than the left digit. Based on the remaining portion of the right digit, a divarication of~35˚can be estimated from digit III. Although incomplete, Ty.I.OG is strongly mesaxonic, with digit III projecting      34.5 cm beyond the left digit. The heel base is broad and tapers to a V-shaped apex, which is offset towards the right side of the track. Collectively, the offset heel and extreme length, width, sinuous form and asymmetrically sloped edges of digit III suggest that some sliding or other kinematic influence altered the shape of this track (see [64,80] and references therein for further discussions of kinematic influence on dinosaur track morphology). Remarks-Despite possibly being exaggerated in length, presumably due to trackmaker kinematics, Ty.I.OG is referable to a tyrannosaurid maker based on its large size (track length >45 cm; see 'identification of trackmaker morphotypes') and convex, high relief preservation that is inconsistent with an enlarged undertrack [62,63]. As preserved, Ty.I.OG is equal in length to the holotype of the tyrannosaurid ichnospecies Bellatoripes fredlundi (i.e., print 2 of PRPRC 2011.11.001) [17]. The type material for B. fredlundi also occurs within lower Unit 4 of the Wapiti Formation, but was found further upstream along the Redwillow River, in the Tumbler Ridge area of British Columbia [17,35]. Bellatoripes fredlundi tracks are diagnosed as being longer than wide, and having wide digit impressions that lack defined digital pads and are thick proximally but taper strongly distally [17]. In addition, the free length of digit III in B. fredlundi is relatively short, and the heel margin is relatively wide [17]. As the morphology of Ty.I.OG is distorted as a result of kinematic factors, this track cannot be convincingly referred to B. fredlundi. In particular, the relatively long free length of digit III in Ty.I.OG is inconsistent with the diagnosis of B. fredlundi.
An in-situ, isolated right theropod track preserved as a natural mould on track layer 3, Ty. I.8C, is also large enough to be confidently regarded as that of a tyrannosaurid (track length = 49 cm; track width = 51 cm) (Fig 9A, 9B and 9H). Ty.I.8C preserves low but welldefined track walls, several faint digital pad margins and sharp claw marks, indicating that it is not a deep undertrack with significantly exaggerated dimensions. Therefore, the length of this track can be taken at face value, and suggests it was produced by a tyrannosaurid~1.95 m tall at the hip. Ty.I.8C lacks the features diagnostic of Bellatoripes fredlundi, including greater track length than track width, and digits that are wide proximally and taper strongly distally. Thus, Ty.I.8C is also not referable to B. fredlundi. For further description of Ty.I.8C see Enriquez et al. [35].
Regardless of ichnotaxonomic affinities, given their spatial and stratigraphic proximity, the same tyrannosaurid species possibly produced tracks Ty.I.OG, Ty.I.8C, and those of B. fredlundi at their type locality in British Columbia [17]. Morphological differences between these tracks may be partly explained by ontogeny [35].  Table 2). A silicone mould of this trackway is accessioned as UALVP 59920. Th.Tw5.3.72-A is longer than wide (track length = 18 cm; track width = 17 cm; length/width ratio = 1.06), and is preserved as a natural mould on track layer 3 (Fig 10A-10C). Digit III is widest at the base (greatest width = 4 cm) and tapers steadily until the most distal third, at which point the digit tapers more gradually. This track was produced by a right foot, given its placement relative to preceding tracks in the sequence (Fig 10D and 10E). Divarication between digits II and III is 36˚, while that between III and IV is 43˚. The heel impression tapers to a blunt V-shape that forms a~90å ngle, although the preceding tracks within the same sequence show more gently rounded heels, which probably reflects variation in substrate properties, kinematics or preservation.

Indeterminate small theropod-like tracks
Trackway parameters for Th.Tw5 H.Tw5.38B-31H -2˚�: (4) Th.Tw2.9B-9C Indet. small theropodlike  [6]. Similar track size, pace and stride length observed in an adjacent and parallel small theropod-like trackway, Th.Tw4.71A-70-A, suggests that the two trackmakers may have been walking together at a similar speed (Fig 10D and 10E; Table 2) (see 'possible trackmaker sociality'). Remarks-At least 11 tridactyl theropod-like tracks occur within the smallest indeterminate size class, and these tracks are morphologically similar to one another. The smallest example, Th.I.10C, measures 12.5 cm in length and-along with the probable deinonychosaur tracks Di. Tw1.1.27D and Di.I.34E-is among the smallest non-avian dinosaur tracks documented from the Wapiti Formation (Figs 2 and 3; Table 1). At least two small tridactyl theropod-like tracks are isolated, while the rest occur within three (or possibly four) short trackway sequences (Figs 1B, 2 and 10).

Indeterminate medium theropod-like tracks
Th.Tw3.27D-24H (exemplar trackway)-Th.Tw3.27D-24H is a sequence of three (or possibly four) tracks, which are highly variable in morphology and preservation (Fig 11). The first track in the sequence, Th.Tw3.1.27D (track length = 30 cm; track width = 40.5 cm), is relatively clear (1-2) -17˚�: and possesses well-defined, short track walls (Fig 11A and 11B). Based on its position relative to subsequent footfalls, Th.Tw3.1.27D is a right print. The shape of Th.Tw3.1.27D is highly irregular: digit II is short and rounded, while digits III and IV are longer and more pointed. In addition, the heel impression is short, with a posterior margin that is almost straight and perpendicular to the digit III axis, making the track considerably wider than it is long (length: width ratio = 0.74). The second track, Th.Tw3.2.26E, occurs 1.7 m ESE from Th.Tw3.1.27D and is a poorly defined left print, comprising three faint, very shallow and highly eroded depressions that correspond to the expected positions of the digits (Fig 11C)

PLOS ONE
The Tyrants Aisle dinosaur tracksite preserved as a convex infilling (track length = 28 cm) (Fig 11D)  length:hip height ratio of 3.06 suggests a running gait [6]. However, this interpretation is tentative as the track size and, by extension, the hip height may be an underestimate, given that the lengths of Th.Tw3.1.27D and Th.Tw3.4.24H both appear to have been reduced by poor preservation and erosion.
Remarks-In addition to Th.Tw3.27D-24H, five isolated tracks on track layer 3 fall into the medium indeterminate theropod-like size class: Th.I.8C, Th.I.62H, Th.I.65G, Th.I.74J, and Th. I.89-A. The long, slender digit proportions of some of these tracks, particularly Th.I.8C, are comparable to those seen in larger specimens such as Th.Tw1.4.6B and Th.I.98-B (Fig 9), which may pertain to distinct size classes of the same trackmaker.

PLOS ONE
The Tyrants Aisle dinosaur tracksite

Indeterminate large theropod-like tracks
Th.I.98-B (exemplar print)-Located on track layer 2, Th.I.98-B is an isolated large theropod track (track length�40.5 cm; track width = 40 cm; length:width ratio�1.01) preserved as a shallow, eroded natural mould (Fig 9E, 9F and 9H). Digit impressions are long, slender, and gently tapering. A shallow, curving trace at the distal end of digit III may be a toe drag or, alternatively, is unrelated (Fig 9E, 9F and 9H). If the curving trace records the distal curvature of digit III, the track is likely a right footprint, as the tip of digit III in theropod tracks is usually in-turned medially [81][82][83]. Divarication between digits II-III and III-IV is 42˚. The metatarsophalangeal area is small relative to the track length, although the posterior heel apex is ambiguous and difficult to identify due to erosion. No interphalangeal pads are visible.
Remarks-Tridactyl theropod-like footprints in this upper size class almost certainly pertain to theropod trackmakers rather than thescelosaurids. This conclusion is based on expected body size; the largest known thescelosaurid species-Thescelosaurus garbanii from the upper Maastrichtian Hell Creek Formation of Montana-possesses a pes approaching 30 cm in length (based on LACM 33542), while other members of the clade are considerably smaller [84]. The two most exemplary of these tracks, Th.Tw1.4.6B and Th.I.98-B, both possess relatively long, slender digit impressions and are almost identical in size (Fig 9) [85][86][87]. Although tracks assigned to Ornithomimipus are generally treated as the footprints of ornithomimosaurs [85][86][87], the trackmaker of Th.Tw1.9B-6B was possibly a juvenile tyrannosaurid [35]. Th.I.98-B demonstrates similarly slender digit proportions to Th.Tw1.4.6B, but we refrain from also  (Table 1). Digit III is broad and elliptical in both tracks, whereas digit IV is relatively short (IV:III length ratio = 0.6-0.68) and rounded in Di. Tw1.1.27D (Fig 11A, 11B and 11E (Figs 3, 11A, 11B and 11E). However, suboptimal preservation, differences in digit counts, and abnormalities in relative spatial positioning make this collective trackway designation tentative [37]. The variation between didactyly and monodactyly within Di.Tw1.27D-29D is likely related to differential levels of track erosion, while missing footprints or the presence of additional trackmakers may explain their spatial abnormalities [37].

Indeterminate morphotype A
In.Tw1.76-A-77-A (exemplar trackway)-Three elongate traces follow one another in close succession on track layer 3 (Fig 12A-12C), the first of which (In.Tw1.1.76-A) is the largest and most distinct. In.Tw1.1.76-A measures 28.5 cm and 8.7 cm in greatest length and width, respectively. The longest portion of the trace consists of a narrow, trench-like impression that is both deepest and widest at the presumed anterior end, and comparatively shallow and narrower at the posterior end (Fig 12A-12C). A shorter lateral impression diverges at~90˚from near the midpoint of the main trace, the tip of which is curved anteriorly. Raised topography near the middle of the main trace (immediately adjacent to where the shorter impression diverges) probably represents minor mud collapse or partial infill. The second trace in the sequence, In.Tw1.2.76-A, is poorly preserved and appears only as a narrow, shallow furrow that is eroded and more distinctly visible in the digital elevation model (Fig 12B) than the corresponding photo (Fig 12A). There is no evidence of a shorter, laterally divergent impression like that present in In.Tw1.1.76-A. The third and final trace, In.Tw1.3.77-A, is also heavily eroded and best viewed from the digital elevation model (Fig 12B). As preserved, only part of the trace is recognisable, which is nearly identical to the anterior portion of In.Tw1.1.76-A. The anterior end of In.Tw1.3.77-A is the most deeply impressed part of the trace, and there is evidence of a shorter, laterally diverging impression that curves anteriorly and terminates in a rounded tip. The striking similarity between In.Tw1.1.76-A and In.Tw1.3.77-A, as well as their close association, indicates that the observed morphology is real and not a product of track surface erosion. No other traces with a similar morphology were observed on any of the track layers.
Remarks-Tracks assigned to indeterminate morphotype A are not morphologically consistent with those that any of the previously identified dinosaurian trackmakers would be expected to produce while engaging in 'normal' terrestrial locomotion. Swimming dinosaurs may produce tracks with atypical morphologies, including parallel sets of sinuous scratches, the presence of kick-off scours, and raised mounds of sediment at the posterior end of the track [90,91]. As these features are lacking in indeterminate morphotype A tracks at Tyrants Aisle, and indeed are absent among tracks across the whole locality, the probability that tracks of this morphotype are swim traces is low. Alternatively, the long, narrow, posterior trace within In.Tw1.1.76-A could represent a metatarsal impression of a dinosaurian trackmaker that sank in soft sediment, or was crouching as it walked [92]. However, dinosaur tracks preserving metatarsal impressions generally also retain impressions of their weight-bearing pedal digits at the anterior end of the trace [92]. Only a single impression occurs at the anterior ends of In.Tw1.1.76-A and In.Tw1.3.77-A, which is inconsistent with a tridactyl dinosaurian trackmaker.
In.Tw1.1.76-A and In.Tw1.3.77-A resemble some pterosaur manus tracks [93][94][95][96], both in their elongate, narrow form and in the presence of a shorter impression that diverges from the longest furrow axis (digit II in most pterosaur manus tracks). If In.Tw1.1.76-A was produced by a pterosaur, its large size suggests an azhdarchid trackmaker, which is consistent with its late Campanian age. Skeletal remains of azhdarchids are rare in western Canada, and presently limited to the Oldman and Dinosaur Park formations (mid-late Campanian) of southern Alberta [97][98][99][100][101][102], and possibly also the Northumberland Formation (Campanian) of Hornby Island, British Columbia [103]. Nevertheless, another possible large pterosaur manus track (TMP 1987.55.39) was previously described from lower Unit 4 of the Wapiti Formation by Bell et al. [9]. TMP 1987.55.39 is preserved as a natural cast on an isolated boulder and was collected from Pinto Creek,~30km from Tyrants Aisle [9]. TMP 1987.55.39 measures 25.5 cm in length, slightly smaller than In.Tw1.1.76-A. Both tracks consist of a relatively long, narrow trace that is nearly perpendicular to a shorter trace, which diverges from the former close to its midpoint. However, unlike in TMP 1987.55.39, the shorter traces of In.Tw1.1.76-A and In. Tw1.3.77-A are curved, presumably in the anterior direction. Furthermore, in TMP 1987.55.39, the longer trace (i.e., the aligned impressions of digits I and III) is curved towards the shorter, perpendicular trace (i.e., the impression of digit II). By contrast, the longer trace in In.Tw1.1.76-A curves away from the presumed digit II impression (Fig 12A-12C). These differences do not necessarily suggest distinct trackmakers, as in some other azhdarchid manus prints (e.g., the holotype manus track of Haenamichnus uhangriensis, CNUPH.P2 [93]), the axis of the longer trace has a similar direction of curvature to that seen in In.Tw1.1.76-A.
Thus, an azhdarchid pterosaur is tentatively suggested as the trackmaker of indeterminate morphotype A. This occurrence is notable as the trackway In.Tw1.76-A-77-A and the isolated, previously described track TMP 1987.55.39 presently constitute the only possible evidence of azhdarchids within the Wapiti Formation.

Indeterminate morphotype B
Tri.I.4B (exemplar print)-Tri.I.4B occurs on track layer 3 and measures 32 cm in length and 37.5 cm in width (Fig 12D-12F; Table 1). Its three digit impressions are deeply impressed into the substrate, whereas the rounded heel margin is shallower and more extensively eroded. Digit III curves to the right in plan view, suggesting that Tri.I.4B may be a left footprint. Divarication between digits III and IV is 57˚, and 44˚between digits II and III (Table 1). Digits II  Fig 12. Traces of uncertain affinity. A, photograph, B, digital elevation model, and C, outline drawing of the indeterminate morphotype A trackway In.Tw1.76-A-77-A, tentatively assigned to an azhdarchid pterosaur. D, photograph, E, digital elevation model, and F, outline drawing of indeterminate morphotype B track Tri.I.4B. and IV are relatively short, broad at the base, and the free portion of these digits are difficult to distinguish from the rest of the track.
Remarks-Three isolated specimens are assigned to this morphotype (Tri.I.4B, Tri.I.6A and Tri.I.10C), which were probably made by a hadrosaurid or theropod. Indeterminate morphotype B is characterised by tridactyl prints with semi-circular, rounded heel margins and relatively short, broad digits II and IV that taper strongly and are nearly 'closed' (i.e., lacking a defined free length) (Fig 12D-12I). Such a pattern of digit morphology suggests that the creation of this morphotype involved track wall collapse or suction during withdrawal of the foot from the substrate. Similar tracks have also been observed elsewhere (e.g., Fig 6A of Razzolini et al. [104]), and their occurrence at Tyrants Aisle indicates that at least some of the tracks at the site were formed when the substrate was soft and relatively water saturated.

Taxonomic affinities
Large ornithopod tracks from Tyrants Aisle are attributed to hadrosaurids, whose skeletal remains and tracks are ubiquitous throughout the Wapiti Formation [8, 10-12, 15-17, 55, 105]. Within Unit 4, a 'mummified' specimen of the saurolophine Edmontosaurus regalis (UALVP 53722) was found near Red Willow Falls,~16km upstream from Tyrants Aisle [15]. This specimen occurred two metres below a bentonite layer dated to 72.58±0.09 Ma [15]. As Tyrants Aisle is at roughly the same stratigraphic level within Unit 4 (or slightly stratigraphically higher, see 'geographic and geological setting'), and is a relatively short distance away (Fig 1A and 1C), it seems likely that the hadrosaurid trackmakers at Tyrants Aisle were also individuals of Edmontosaurus regalis. This is further supported by the temporal correlation between Unit 4 of the Wapiti Formation and the lower members of the Horseshoe Canyon Formation in southern Alberta [34,39,42], in which E. regalis is the sole hadrosaurid [106].
Tridactyl theropod-like footprints at Tyrants Aisle greater than 45 cm in length can be attributed with confidence to tyrannosaurids (see 'identification of trackmaker morphotypes'). The presence of tyrannosaurid tracks is significant, as their body fossil record within the Wapiti Formation is presently restricted to teeth, an isolated vertebra (TMP 1989.062.0004), and a metatarsal (TMP 2005.066.0047) [10,12,14,54,105]. Neither TMP 1989.062.0004, TMP 2005.066.0047, nor any of the isolated teeth are generically identifiable. However, the most spatiotemporally proximate species is Albertosaurus sarcophagus, recovered from the Danek Bonebed (~71.  in the Horsethief Member of the Horseshoe Canyon Formation, near Edmonton,~450 km SE of Tyrants Aisle [24-25, 34, 42, 107]. Probable Albertosaurus teeth within the underlying Drumheller Member extend the occurrence of this taxon to~73 Ma [24,42]. Tyrants Aisle is temporally correlated with the Drumheller Member, and it therefore seems probable that tyrannosaurid tracks from Tyrants Aisle (and elsewhere within Unit 4 of the Wapiti Formation) were produced by a species of Albertosaurus [35].
Theropod-like tracks below 45 cm in length are more challenging to identify, as a range of small-to-medium sized theropods with overlapping foot lengths are known from the Campanian-Maastrichtian of Laramidia, including members of Ornithomimidae, Dromaeosauridae, Troodontidae, Caenagnathidae, Alvarezsauridae, Therizinosauridae, and juvenile Tyrannosauridae [31, 54,57,105]. Furthermore, members of Thescelosauridae may theoretically produce theropod-like tracks up to approximately 30 cm in length [84]. Distinguishing these G, photograph, H, digital elevation model, and I, outline drawing of indeterminate morphotype B track Tri.I.10C. All digital models-and the photographs used to create them-are available for download within the supplementary information.
https://doi.org/10.1371/journal.pone.0262824.g012 trackmaker types, where possible, relies predominantly on differing functional digit counts. For instance, tridactyl theropod-like tracks are unlikely to pertain to therizinosaurids or caenagnathids, which probably produced tetradactyl footprints [31,108]. Di.Tw1.1.27D and Di. I.34E are attributable to deinonychosaurs on the basis of didactyly and, more specifically, to Troodontidae based on the relative shortness of digit IV in comparison to digit III [37]. Although we refrain from assigning most tridactyl theropod-like footprints below 45 cm in length to any particular clade, some of the larger and more distinct examples (e.g., Th.I.98-B, Th.I.8C, Th.I.89-A, and those within Th.Tw1.9B-6B) were possibly produced by ornithomimids based on their long, slender digits, occasional tendency for the impression of digit II to be separated from the rest of the print (present in Th.Tw1.2.8B), and constriction at the base of digit III (present in Th.I.89-A and Th.Tw1.4.6B) [85][86][87]. Rare, isolated ornithomimid bones are known from Unit 3 of the Wapiti Formation, and possibly Unit 4, but are generically indeterminate [8,12,14,16]. Alternatively, some large tridactyl theropod tracks (such as Th. Tw1.4.6B) could pertain to juvenile tyrannosaurids, which are likely to have exhibited a similarly gracile foot morphology to that seen in ornithomimids [35].

Possible trackmaker sociality
Aligned, unidirectional trackways may suggest the presence of multiple individuals moving side-by-side, providing evidence of gregarious behaviour [4,7,17]. However, such sequences can also be produced by a succession of solitary animals, at different times, which were 'funnelled' through the same space due to physical barriers, or were following the orientation of palaeoshorelines [109]. Therefore, it is important to consider additional evidence when postulating gregarious behaviour, such as trackway and print spacing, relative depth, and degree of preservational similarities [110].
The lowest track-bearing horizon (layer 1) at Tyrants Aisle preserves at least 15 hadrosaurid pes prints, some in single-step sequences (i.e., sets of exactly two successive tracks), and all concentrated within 15 linear metres of outcrop (Figs 3 and 13). No additional tracks have been observed in places where this same layer is exposed further west (Fig 1B). Nearly all of these hadrosaurid tracks, except H.I.28-B, were made by trackmakers travelling approximately SE (average = 143˚). Due to the shortness and close proximity of the trackway sequences, it is difficult to confirm the precise number of individuals represented on this track layer. Nevertheless, between grid rows 28 and 33 on track layer 1, we estimate that five or more individuals were possibly walking side-by-side, based on the lateral spacing (~0.5-1.0 m apart) and nonoverlapping arrangement of their tracks (Fig 13). By contrast, H.I.28-B is in opposition to the main trackway grouping and oriented in a westerly direction (Fig 3). H.I.28-B is also less eroded and more distinctly preserved than the surrounding tracks, and has an associated manus impression (Fig 8A-8C) that overprints the tip of digit III in H.I.28-C, which is part of the main grouping of aligned footprints. These attributes suggest H.I.28-B formed later than the aligned tracks. Thus, based on these trackway data, the concentration of prints within this relatively small area, and their relative degree of preservation, we conclude that the hadrosaurid tracks on layer 1 were produced during at least two temporally separated events: 1) a possible herd of at least 5 individuals walked across track layer 1 in a generally SE direction (Figs 3 and 13), and 2) at least one individual (the trackmaker of H.I.28-B) traversed the same area sometime later, travelling in a westerly direction (Figs 3 and 8A-8C). Evidence of possible hadrosaurid sociality is also present on track layer 3. In particular, two individuals may have been travelling NW together at a similar pace, separated by a lateral distance of about 1 m, between grid rows 52 and 54 (trackways identified as H.Tw6.53B-53A and H.Tw7.54A-54-A; Fig 1B; Table 2). However, it must be clarified that alternative scenarios, including the possibility that several lone individuals were 'funnelled' through the same area in succession, cannot be ruled out.
On both track layers 1 and 3, sets of aligned, regularly spaced hadrosaurid tracks are generally characterised by fairly uniform track lengths. If such sets of tracks document genuine social behaviour, the socialising individuals must have primarily been similar in size and, presumably, age (Figs 1B, 3 and 13; Table 1). Most other hadrosaurid tracks preserved across track layers 1 and 3 vary more widely in size and are largely isolated, providing no evidence of social behaviour. However, in one area of track layer 3 (Fig 10D and 10E), juvenile-sized and adult-sized tracks occur together within an area of 2 m 2 ; H.I.71A is the smallest confidently identified hadrosaurid track at Tyrants Aisle (track length = 23 cm), and occurs less than one metre from H.I.72A (track length = 33cm), both of which are similarly oriented. Two larger hadrosaurid footprints (track length�53 cm) are preserved close to H.I.71A and H.I.72A ( Fig  10D and 10E), but are only roughly comparable to H.I.71A and H.I.72A in orientation. Therefore, it is uncertain whether the hadrosaurids at Tyrants Aisle were consistently or only occasionally age segregated, if herding was indeed taking place.
The occasional presence of manus impressions alongside pes prints (e.g., in H.I.-2B, H. Tw1.2.2A, H.I.28-B, and H.I.72A) indicates that some of the hadrosaurids that left traces on track layers 1 and 3 were walking quadrupedally. By contrast, the majority of the hadrosaurid pes tracks lack associated manus impressions. As most of the animal's body weight was supported by the pedes-indicated by pronounced heteropody and a centre of mass inferred to have been positioned dorsal to the pedes [124][125][126]-smaller manus impressions may have failed to register on the substrate or were shallowly impressed, and thus more easily lost by erosion, which is prevalent at Tyrants Aisle. Despite this, we acknowledge the possibility that in some cases the absence of manus impressions may reflect a genuinely bipedal mode of walking.
Probable theropods-The parallel alignment of two particular trackways on track layer 3 (Th.Tw4.71A-70-A and Th.Tw5.71-A-72-A) may suggest social behaviour between theropods (hip height�0.72 m) at Tyrants Aisle (Fig 10). Both animals were walking in a roughly NW direction,~1 m apart, with similar pace and stride lengths ( Table 2). A possible trackway from a third individual was identified, and is oriented in the same direction, between the two main trackways (Fig 10). However, this possible sequence is heavily eroded and difficult to interpret with confidence. Quality of track preservation varies between and within the two main trackways. Print Th.Tw5.3.72-A is relatively clear, whereas the preceding two footprints in the same trackway are more poorly defined. In the adjacent trackway, Th.Tw4.4.70-A retains only the impressions of the distal ends of the digits, while Th.Tw4.1.71A is represented by a single curved trace of digit III. Furthermore, track Th.Tw4.2.70A is entirely mud collapsed, exhibiting only thin creases pertaining to each digit (Fig 10D and 10E). By contrast, the three tracks within Th.Tw5.71-A-72-A, although variable in morphology and preservational clarity, are not similarly mud collapsed. These differences may suggest the two individuals traversed track layer 3 at different times. However, given that preservation is also highly variable within each trackway, we suggest that the differences in preservation between Th.Tw4.71A-70-A and Th. Tw5.71-A-72-A are instead due to differential sediment properties and erosion. Based on the

Palaeoecological insights
Based on the footprint orientation data, the majority of trackmakers on layer 1 were walking in a SE direction (Fig 14A), whereas those on layer 3 show a more bi-directional pattern aligned along the SE-NW axis (Fig 14B and 14C). Although the average direction of current flow during sediment deposition was northward (see 'depositional environment, tracksite stratigraphy and lithology'), some Rhizocorallium on both track layers 1 and 3 are similarly oriented SE-NW. Therefore, some of the trackmakers were likely following the orientation of a meandering river margin.
Dinosaurian ichnodiversity at Tyrants Aisle is a subset of that captured within the entire track and body fossil record of Unit 4 of the Wapiti Formation (Fig 15). The largest and most abundant tracks in the Tyrants Aisle assemblage are those of hadrosaurids, the only ornithischian trackmakers identified at the site. Hadrosaurid prints occur on both of the main track layers 1 (number of definitive pes tracks �15; manus tracks �2) and 3 (number of definitive pes tracks �25; manus tracks �4). On track layer 1, only hadrosaurid tracks are present. On track layer 3, an abundance ratio of 25:1:23 is obtained for all confidently identifiable hadrosaurid, tyrannosaurid, and non-tyrannosaurid theropod-like pes tracks, respectively. When trackways are treated as a single occurrence, the faunal ratio for track layer 3 becomes 17:1:14. However, these ratios do not include the high amount (�90) of heavily eroded and non-diagnostic tracks on layer 3, which could not be reliably counted. Many indeterminate tracks on layer 3 appear as subrounded, eroded potholes, whose large size indicates that they likely represent former hadrosaurid tracks. Therefore, the true proportion of hadrosaurid trackmakers is probably even higher than these ratios imply. Furthermore, as tyrannosaurid tracks below 45 cm in length were not readily distinguishable from other tridactyl theropod-like track types, the number of tyrannosaurid tracks recognised here is likely also an underestimate. Nevertheless, the available data across all track layers collectively support an overall palaeoecological dominance of hadrosaurids at Tyrants Aisle, corroborating their skeletal abundance both within Unit 4 of the Wapiti Formation [8,15,16], and broadly across many Late Cretaceous terrestrial ecosystems within Laramidia [24,54,57,[137][138][139]. Similarly, a relative scarcity of tracks attributable to probable troodontids [37] and possible ornithomimids is congruent with their sparse body fossil record in the Wapiti Formation. Except for Th.I.98-B and Ty.I.OG, all of the theropod-like tracks at Tyrants Aisle occur on track layer 3, although this is likely due to the much greater exposure of this particular layer (Fig 1B).
Confidently measurable hadrosaurid tracks at Tyrants Aisle range from~23-65 cm in length, with most tracks measuring between 45 and 65 cm long ( Table 1), suggesting that the majority of individuals were between~1.8 m and~2.6 m tall at the hip. By comparison, ROM 801-one of the heaviest known individuals of Edmontosaurus regalis-stood approximately 3 metres tall at the hip and has a reconstructed pes~66.5 cm in length [140,141]. Among hadrosaurids, Horner et al. [142] and Evans [143] consider juveniles to be individuals measuring 50% or less than the maximum known adult body length. Thus, as most hadrosaurid footprints at Tyrants Aisle approach the predicted pes length of ROM 801, the majority of hadrosaurid trackmakers at Tyrants Aisle were likely adults. Track assemblages such as this, in addition to relative skeletal rarity within the body fossil record, suggest that large Edmontosaurus individuals exceeding 3.0 metres tall at the hip constituted only a small minority of a given population [144].
Ceratopsids-whose tracks usually possess five manual and four pedal digits [4]-are conspicuously undocumented from the Tyrants Aisle track fauna, but are known from body fossils within Unit 4 of the Wapiti Formation [14]. Specifically, the centrosaurine Pachyrhinosaurus is abundant within two distinct monodominant bonebeds, which are situated on Pipestone Creek near the Unit 3-Unit 4 boundary [13,14], and on the Wapiti River near the middle of Unit 4 [14]. Thus, Tyrants Aisle is stratigraphically positioned between these two bonebeds and falls within the stratigraphic range of Pachyrhinosaurus within the Wapiti Formation. We attribute the lack of identifiable ceratopsid tracks to inadequate sampling, as many footprints at Tyrants Aisle are heavily eroded and could not be identified. Alternatively, their absence may simply reflect a chance mismatch between Pachyrhinosaurus herd movements and track layer deposition. more faithful palaeocommunity reconstructions (Fig 17). Additional studies should further assess the role of lithology and palaeoenvironment as a determinant of observed dinosaur ichnodiversity-across a broad sample of tracksites-as this may further improve understanding of Laramidian dinosaur palaeoecology by clarifying the influence of depositional setting on the faunal signals identified herein.