Non-technical Summary
In 1855 Ferdinand Hayden collected a single tooth from the Judith River badlands of central Montana. Joseph Leidy in Philadelphia named this specimen the following year as Troodon formosus, “beautiful wounding tooth.” So began the somewhat troubled history of the dinosaur Troodon. Over the next 130 years, Troodon would be considered a lizard, possibly a pachycephalosaur (a thick-headed dinosaur), or simply too incomplete to understand. Resolution of Troodon’s true nature did not come until Philip Currie in the 1980s described more material, including jaws with teeth, and correctly identified Troodon formosus as a species of small, carnivorous dinosaur close to the ancestry of birds. We know Troodon now as possibly one of the smartest dinosaurs, or at least one of the brainiest. But because of the fragmentary nature of the original discovery, some scientists think that the name Troodon formosus should be thrown out.
We present previously undescribed material from Montana that helps to clarify Troodon’s true nature, much as Currie envisioned. We propose that these new specimens should be used to formally ground the definition of the species. Based upon these new specimens, as well as a reading of the International Code of Zoological Nomenclature, we argue that Troodon formosus remains valid as the senior synonym of Stenonychosaurus inequalis and captures the species concept first envisioned by Leidy 165 years ago for an unknown animal with unusual teeth in the Cretaceous of Montana. Preservation of the name means that Troodon formosus remains among the earliest dinosaurs named from North America.
Introduction
In 1871, Ferdinand Hayden led a geological survey team through the Yellowstone region documenting the geography and geology of the area and eventually leading to its recognition in 1872 as the first National Park of the United States of America. But even earlier, in 1854, Hayden traveled largely alone through the Cretaceous badlands of what was then part of the recently named Nebraska Territory (Brown, Reference Brown1971). The Smithsonian Institution had hired Hayden for an 18-month-long collecting expedition to the American West to explore the geology and gather rocks, fossils, and modern vertebrates (Brown, Reference Brown1971). Hayden traveled by steamboat, keelboat, and overland through the Upper Missouri River area (today central Montana). He amassed six tons of geologic specimens, including a small but valuable collection of vertebrate fossils near the Judith River. He shipped these back to the Smithsonian Institution, where some were then sent on to Joseph Leidy at the University of Pennsylvania. Spencer Baird at the Smithsonian accompanied the shipment to Leidy with the following message, “You will find quite a variety of saurian teeth, some of which will make your eyes water” (Cassidy, Reference Cassidy2000, p. 72).
In 1856, Leidy described and named one of these teeth (ANSP 9259) as a new “lacertian,” Troodon formosus, meaning “beautiful wounding tooth.” Thus began the taxonomic saga of Troodon formosus. Whereas other species named by Leidy on teeth collected by Hayden have long fallen into disuse, Troodon formosus has persisted until very recently when a few publications (e.g., Zanno et al., Reference Zanno, Varricchio, O’Connor, Titus and Knell2011; Evans et al., Reference Evans, Cullen, Larson and Rego2017; van der Reest and Currie, Reference van der Reest and Currie2017) have suggested that the name T. formosus be considered a nomen dubium. Here we briefly review the taxonomic history of T. formosus, provide descriptions of relevant troodontid material from the Cretaceous Two Medicine Formation of Montana (Fig. 1), and offer a taxonomic solution that retains nomenclatural stability, a prime directive of the International Code of Zoological Nomenclature (ICZN), which is more consistent with taxonomic procedures as outlined in the ICZN.
Figure 1. Geography and stratigraphy of Troodon formosus-bearing formations in Alberta, Canada and Montana, USA—Two Medicine Formation (TMF, yellow); Dinosaur Park Formation (DPF), Oldman Formation (OF) and Foremost Formation (FF) (orange), and Judith River Formation (JRF, red). (1) Geographic ranges of TMF (Childs, Reference Childs1985), DPF (including Oldman Formation; Rogers et al., Reference Rogers, Horner, Ramezani, Roberts and Varricchio2025), and JRF (Childs, Reference Childs1985; Rogers et al., Reference Rogers, Horner, Ramezani, Roberts and Varricchio2025); visualized using Macrostrat (macrostrat.org; Peters et al., Reference Peters, Husson and Czaplewski2018). The DFP and OF were mostly undifferentiated until the early 1990s but are now distinguished by a regional disconformity (Eberth and Hamblin, Reference Eberth and Hamblin1993). The TMF, DPF, and JRF are closely related, and the divisions are exaggerated by political boundaries. (2) Stratigraphic ranges for these same units. Full color segments indicate the fossiliferous ranges for the Alberta and Judith River formations and the upper Flag Butte Member (FBM) of the Two Medicine Formation (Rogers et al., Reference Rogers, Horner, Ramezani, Roberts and Varricchio2025). Lighter color shows further extent of formation (TMF descends beyond scale to ca. 82.4 Ma in its entirety) (Rogers et al., Reference Rogers, Horner, Ramezani, Roberts and Varricchio2025). Geographic (1) and stratigraphic (2) locations of specimens are shown with solid square (MOR 246, 430), solid stars (MOR 553, 748, 11775), and open star (MOR 563).
Leidy (Reference Leidy1860) followed up his initial description a few years later by figuring the holotype tooth and briefly discussing its possible function. He suggested that T. formosus might be related to monitor lizards. Ten years passed and Cope (Reference Cope1870) included T. formosus in the Dinosauria among his Goniopoda group of theropods without comment. Whereas both Zittel (Reference Zittel1890) and Nopcsa (Reference Nopcsa1901) would maintain this theropod association, Lambe (Reference Lambe1902), reporting on the first additional specimens, briefly described two teeth from the Belly River series of Alberta, Canada, as both T. formosus and squamate. But complications arose in 1924 when Gilmore (Reference Gilmore1924) described the skull of a new dome-headed ornithopod dinosaur. He considered the teeth of this complete specimen as sufficiently similar to T. formosus to warrant congeneric status. He named his new dinosaur Troodon validus, “the true Troodon,” and established the Troodontidae for thick-skulled ornithischians. This usage continued for over 20 years, until Sternberg (Reference Sternberg1945) correctly identified T. formosus, and hence the Troodontidae, as distinct from the Pachycephalosauridae and likely representing theropod dinosaurs. Subsequent discoveries and re-descriptions of isolated dentaries finally confirmed the theropod affinities of these taxa (Russell, Reference Russell1948; Sternberg, Reference Sternberg1951) nearly 100 years after the initial description by Leidy (Reference Leidy1856).
Russell (Reference Russell1969) provided greater details on the nature of troodontid theropods in describing a series of isolated elements as well as associated material (CMN 12340) from the Upper Cretaceous Oldman Formation of Alberta. He organized these under the name Stenonychosaurus inequalis Sternberg, Reference Sternberg1932, a species first established on a foot and a few additional elements. Russell (Reference Russell1969) documented the close relationship of Stenonychosaurus inequalis to the Asian Saurornithoides mongoliensis Osborn, Reference Osborn1924, and suggested that one or both might be junior synonyms to T. formosus. Importantly, Russell (Reference Russell1969) further clarified the Troodontidae as a clade of predominantly Asian and North American small theropods with slender metatarsi, large brains, and numerous small teeth—a usage that has persisted through today (e.g., Osmólska and Barsbold, Reference Osmólska, Barsbold, Weishampel, Dodson and Osmólska1990; Makovicky and Norell, Reference Makovicky, Norell, Weishampel, Dodson and Osmólska2004; Zanno et al., Reference Zanno, Varricchio, O’Connor, Titus and Knell2011; Evans et al., Reference Evans, Cullen, Larson and Rego2017; van der Reest and Currie, Reference van der Reest and Currie2017). Carpenter (Reference Carpenter1982) curiously and incorrectly, given the relative dates of publication, synonymized the earlier named Troodon formosus into Stenonychosaurus inequalis, which he considered congeneric with Saurornithoides. Following the nomenclature of Russell (Reference Russell1969), Currie (Reference Currie1985, p. 1643) described additional cranial material from the “Judith River (Oldman) Formation” of Alberta and importantly provided the first formal diagnosis for S. inequalis involving eight cranial characters.
Currie (Reference Currie1987) later described a series of isolated dentaries and teeth from the Judith River and Horseshoe Canyon formations of Alberta, Canada. He concluded that “the type specimen of Troodon formosus is diagnostic” (Currie, Reference Currie1987, p. 73) and that “three isolated dentaries (ROM 1445, TMP 67.14.39, TMP 82.19.151) from the Cretaceous of North America referred to ‘Stenonychosaurus inequalis’ have germ teeth that can be identified as those of Troodon formosus” (Currie, Reference Currie1987, p. 80). He considered the name Troodon formosus as “clearly the senior synonym” and that “there was no justification to suppress it in favor of ‘Stenonychosaurus’” because the name had been used consistently over the years (Currie, Reference Currie1987, p. 80). Additional cranial elements would be included in the species from both past and subsequent descriptions (Currie, Reference Currie1985; Currie and Zhao, Reference Currie and Zhao1993). Varricchio et al. (Reference Varricchio, Horner and Jackson2002) identified in ovo embryonic material that included both cranial and post-cranial elements from the Campanian Two Medicine Formation of Montana, USA, as belonging to Troodon formosus following the diagnosis of Currie (Reference Currie1985).
Currie (Reference Currie, Currie and Koppelhus2005) changed his view slightly on the relationship between Troodon formosus and Stenonychosaurus inequalis and presented a new combination, Troodon inequalis. Four specimens were presented as “reference specimens”: the type of S. inequalis, two earlier described cranial pieces included in S. inequalis by Currie (Reference Currie1985), and one new additional partial braincase. Currie (Reference Currie, Currie and Koppelhus2005, p. 375) stated that the T. formosus type “was distinctive enough to eventually lead to the identification of the rest of the skeleton” and that, “although Currie (Reference Currie1987) advocated considering the holotype and material from the [Dinosaur Provincial] Park as conspecific, it is more conservative to retain Sternberg’s Reference Sternberg1932 specific name.” However, neither a species list nor a diagnosis was provided for the genus Troodon nor is an explicit, morphological justification for the new combination provided.
More recently, van der Reest and Currie (Reference van der Reest and Currie2017) reexamined the troodontid material from the Dinosaur Park Formation, including the material that was recognized first as Troodon formosus (Currie, Reference Currie1987) and then later as T. inequalis (Currie, Reference Currie, Currie and Koppelhus2005). Van der Reest and Currie (Reference van der Reest and Currie2017) argued that two taxa are represented in this assemblage, and established Latenivenatrix mcmasterae van der Reest and Currie, Reference van der Reest and Currie2017, as a troodontid of the upper Dinosaur Park Formation (MAZ-2) with a frontal having a more right-triangular outline. The second taxon appears restricted to the lower portion of the formation (MAZ-1) and has a more L-shaped frontal. Van der Reest and Currie (Reference van der Reest and Currie2017) recognized this species as Stenonychosaurus inequalis, using the original type specimen (CMN 8539) of Sternberg (Reference Sternberg1932). Similarities between the Dinosaur Park Formation material and that of the Two Medicine Formation of Montana suggest this species was also present in the latter formation. They (van der Reest and Currie, Reference van der Reest and Currie2017) opted to use S. inequalis rather than Troodon formosus because they deemed the holotype, and troodontid teeth from the Dinosaur Park Formation in general, as undiagnostic at the species level (Larson and Currie, Reference Larson and Currie2013). However, they also noted that Talos sampsoni Zanno et al., Reference Zanno, Varricchio, O’Connor, Titus and Knell2011, from the Cretaceous of Utah also possesses the only feature, a flat to convex anterior surface on metataral III, in the S. inequalis diagnosis visible in its type specimen (van der Reest and Currie, Reference van der Reest and Currie2017). Cullen et al. (Reference Cullen, Zanno, Larson, Todd, Currie and Evans2021) tested the validity of two stratigraphically separated troodontids in the Dinosaur Park Formation by undertaking a morphometric analysis of frontals. They found considerable morphologic and stratigraphic overlap between S. inequalis and Latenivenatrix mcmasterae and considered the latter as a junior subjective synonym of the former. The recognition of L. mcmasterae further suffers from the fragmentary nature of specimens. Most consist of only a few elements, non-overlapping with those of other specimens. Consequently, we follow the taxonomy of Cullen et al. (Reference Cullen, Zanno, Larson, Todd, Currie and Evans2021) and use “L. mcmasterae” to refer to specimens or morphologies that van der Reest and Currie (Reference van der Reest and Currie2017) included in this taxon. Nevertheless, the validity of L. mcmasterae could be demonstrated in the future if better, associated material is discovered.
Here we describe a small collection of troodontid material from the Campanian Two Medicine Formation of Montana that largely correlates to the Dinosaur Park Formation of Alberta (Fig. 1). This material ranges from embryonic material to a clutch-associated adult and includes most elements of the skeleton. We concur with van der Reest and Currie (Reference van der Reest and Currie2017) that this Two Medicine troodontid material is synonymous with their taxon from the lower Dinosaur Park Formation, however we show that Troodon formosus remains the more appropriate name. It has been applied to the Montana material previously and has priority over S. inequalis. While issues remain with the diagnostic nature of the holotype, the name Troodon formosus satisfies three principles of the International Code of Zoological Nomenclature as discussed below: universality, priority, and stability. Further, the type of S. inequalis has the same diagnostic issues as that of T. formosus. We include here only a few elements pertinent to the diagnosis because a more complete description is to follow. We also propose specimens to form a neotype for T. formosus (a petition for which will be sent to the International Commission on Zoological Nomenclature).
Phylogenetic methods
To evaluate the phylogenetic position of Troodon formosus among theropod dinosaurs, we ran parsimony and Bayesian analyses using the character matrix from van der Reest and Currie (Reference van der Reest and Currie2017). The matrix, which consisted of 93 taxa and 366 characters, was originally adapted from the matrix of Gao et al. (Reference Gao, Morschhauser, Varricchio, Liu and Zhao2012) with scoring for Gobivenator and Urbacodon incorporated from Tsuihiji et al. (Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014) and Averianov and Sues (Reference Averianov and Sues2007), respectively. We revised and added new data for T. formosus based on information from additional specimens (see the Supplementary Materials for revised character scores). The updated scoring consisted of 16 revised characters and new scorings for 118 characters that were previously missing. The revised character scores are explained by the improved preservation quality of the newly described material, particularly MOR 553. These corrections were made directly to van der Reest and Currie’s (Reference van der Reest and Currie2017) original character matrix with no additional character modifications. We did not include Albertavenator curriei Evans et al., Reference Evans, Cullen, Larson and Rego2017, described on an isolated frontal, in our analysis due to the lack of data. Given the recent synonymization of “Latenivenatrix mcmasterae” with “Stenonychosaurus inequalis” (Cullen et al., Reference Cullen, Zanno, Larson, Todd, Currie and Evans2021), we also carried out the same parsimony and Bayesian analyses excluding both “L. mcmasterae” and three related characters established by van der Reest and Currie (Reference van der Reest and Currie2017) in their study. An alternative hypothesis is that “L. mcmasterae,” S. inequalis, and T. formosus are distinct taxa. Therefore, we additionally included the “S. inequalis” type specimen (CMN 8539) in our phylogenetic analyses, along with “L. mcmasterae” and our revised coding of T. formosus. Out of 27 characters that could be scored for the S. inequalis type specimen (CMN 8539), none differs from those in the Two Medicine Formation sample (Supplementary Materials, Zenodo https://doi.org/10.5281/zenodo.13305737).
For the parsimony analysis, we followed the protocol of van der Reest and Currie (Reference van der Reest and Currie2017) using TNT (Tree analysis using New Technology) v1.6 (Goloboff et al., Reference Goloboff, Farris and Nixon2008; Goloboff and Morales, Reference Goloboff and Morales2023). This consisted of using the TBR (Tree Bisection Reconnection) swapping algorithm, completing 99999 random addition sequences with 1000 trees per replicate. The search including “L. mcmasterae” and CMN 8539 retained 12 most parsimonious trees (parsimony score = 1385 steps, consistency index = 0.323, retention index = 0.739), which were used to produce strict and 50% majority rules consensus trees. Bootstrap support values were then calculated from 5000 replicates of those retained trees. We also ran a Bremer analysis to calculate Bremer node support values. This involved a subsequent series of tree searches using TBR while retaining suboptimal trees with one to eight additional steps (i.e., one to eight steps longer than the six most parsimonious trees). We retained 8000 suboptimal trees for calculating the Bremer support values.
For our Bayesian phylogenetic analysis, we used MrBayes v3.2.6 (Ronquist et al., Reference Ronquist, Teslenko, van der Mark, Ayres, Darling, Höhna, Larget, Liu, Suchard and Huelsenbeck2012). We ran four Markov-chain Monte Carlo (MCMC) replicates for 10 million generations, each with four chains, sampling every 1000 generations, and a burn-in of 25%. Rates of character evolution were allowed to vary across states under a gamma distribution and vary between binary and multistate character partitions. We described the character scoring as informative and treated Allosaurus as the outgroup. The average standard deviation of split frequencies was less than 0.01, indicating convergence of the MCMC chain after burn-in. A 50% majority rules consensus tree was produced from the resulting posterior distribution of trees.
To ensure that our results were comparable with previous studies, we successfully replicated the results of van der Reest and Currie (Reference van der Reest and Currie2017) using the parsimony and Bayesian protocols described above (Supplementary Materials, Figs. S1–S5). All data and code used in this study are available on Zenodo (https://doi.org/10.5281/zenodo.13305737).
Materials
Materials described in this paper represent portions of the following specimens from the Upper Cretaceous Two Medicine Formation: Museum of the Rockies (MOR) 246, embryonic remains in eggs; MOR 430, partial small juvenile skeleton; MOR 563, large juvenile skeleton; MOR 553, collection of bonebed material representing multiple individuals from large juveniles through adults that includes most elements excepting nasal, quadratojugal, jugal, costal series, ilium, some manual and pedal phalanges; MOR 748, partial adult skeleton including pelvis, tail, and hindlimbs, associated with egg clutch; and MOR 11775, an isolated frontal. All these specimens were collected between 1983 to 1993 by Jack Horner and crews. Jack Horner found the embryonic material on September 14, 1983, amidst a snow squall and collected it later that fall in better weather. MOR 430 and MOR 563 come from 1986 and 1988, respectively. The site for MOR 553 was discovered June, 15, 1988, with excavations continuing through 1993 during which the nearby specimens, MOR 748 and MOR 11775 were collected.
Repositories and institutional abbreviations
Examined and referenced specimens used in this study are housed in the following institutions: Academy of Natural Sciences of Drexel University (ANSP), Philadelphia, USA; Canadian Museum of Nature (CMN), Ottawa, Canada; and the Museum of the Rockies (MOR), Bozeman, USA; Royal Ontario Museum (ROM), Toronto, Canada; Royal Tyrrell Museum (TMP), Drumhelller, Canada; University of Alberta Laboratory of Vertebrate Palaeontology (UALVP), Edmonton, Canada.
Systematic paleontology
Theropoda Marsh, Reference Marsh1881
Maniraptora Gauthier, Reference Gauthier and Padian1986
Troodontidae Gilmore, Reference Gilmore1924, sensu Turner et al., Reference Turner, Makovicky and Norell2012
Troodontinae Gilmore, Reference Gilmore1924, sensu van der Reest and Currie, Reference van der Reest and Currie2017
Troodon Leidy, Reference Leidy1856
Type species
Troodon formosus Leidy, Reference Leidy1856, by original designation from the Judith River Formation of central Montana, U.S.A. (Leidy, Reference Leidy1856, Reference Leidy1860, pl. 9, figs. 53–55).
Troodon formosus Leidy, Reference Leidy1856
Figure 2. Adult, juvenile, and embryonic maxillae to scale for Troodon formosus. Scale bar for (1–9) = 2 cm. (1–4) Adult left maxilla (MOR 553S-8-11-92-205) in dorsal (1), lateral (2), ventral (3), and medial (4) views. Asterisks mark breakage and breakage with bone displacement. (5–8) Juvenile right maxilla (MOR 553S-8-3-9-375) shown reversed in dorsal (5), lateral (6), ventral (7), and medial (8) views. (9, 10) MOR 246-11, embryonic maxilla in lateral view at scale with other maxillae (9) and at twice its size (10).
Figure 3. MOR 11775, left Troodon formosus frontal in lateral (1), dorsal (2), ventral (3), medial (4), posterior (5), and anterior (6) view. Note distinctive L-shape in dorsal view and slot-like contact for laterosphenoid. Scale bar = 2 cm; dpr = depression anteromedial to postorbital contact; lsc = laterosphenoid contact; str = ridge marking anterior border of supratemporal fossa.
Figure 4. Frontal comparisons using the parameters and data of Cullen et al. (Reference Cullen, Zanno, Larson, Todd, Currie and Evans2021). Two Medicine frontal specimens (MOR 553S-8-4-92-150 and 11775) are denoted by black triangles. Open diamonds and open circles represent triangle- (“Latenivenatrix”) and L-shaped (“Stenonychosaurus”) frontals, respectively, from Alberta and corresponding regression lines (dashed and solid) (Cullen et al. Reference Cullen, Zanno, Larson, Todd, Currie and Evans2021, fig. 4B). The Two Medicine specimens fall with the L-shaped for two of the three ratios, highlighting the morphologic overlap between these morphs (Cullen et al., Reference Cullen, Zanno, Larson, Todd, Currie and Evans2021).
Figure 5. (1–6) Troodon formosus right pubis, MOR 553S-8-3-9-387, in anterior (1), medial (2), posterior (3), lateral (4), proximal (5), and distal (6) views. Anterior is to the left in (5, 6). Scale bar = 2 cm. Asterisks mark clear evidence of long-axis compression that telescoped the shaft with a portion of shaft extending beyond the pubic boot. (7) MOR 246-11, T. formosus embryo showing ilium, pubis, and femur in articulation. Note straight pubic shaft as in adult specimen. Scale bar = 1 cm. (8, 9) Partial right ilium from small juvenile T. formosus, MOR 430, in lateral (8) and medial (9) views. Scale bar = 2 cm; ac = acetabulum; ap = ambiens process; at = antitrochanter; ilpp = iliac peduncle of pubis; ipi = ischiadic peduncle of ilium; ipp = ischiadic peduncle of pubis; ppi = pubic peduncle of ilium; ps = pubic shaft.
Figure 6. Metatarsals of T. formosus. (1, 2) Right metatarsus, MOR 748, in anterior (1) and posterior (2) views. Proximally, metatarsal III is only visible in posterior view. But much of the distal half of the shaft is hidden posteriorly. (3–7) MOR 553S-7-29-92-113, left juvenile metatarsal III in distal (3), anterior (4), lateral (5), posterior (6), and medial (7) views. Note weakly convex anterior face at widest point and triangular extensor fossa. The latter is not clearly expressed in MOR 748, because this feature appears to change with ontogeny. All scale bars = 2 cm; caf = convex anterior face; ef = extensor fossa; lc = lateral condyle; mc = medial condyle, MT II–IV = metatarsals II–IV.
Reference Leidy1856 Troodon formosus Leidy, p. 72.
Reference Sternberg1932 Stenonychosaurus inequalis Sternberg, p. 102, fig. 2.
Reference Gilmore1932 Polyodontosaurus grandis Gilmore, p. 117.
Reference Carpenter1982 Saurornithoides inequalis; Carpenter, p. 127, fig. 2.
Reference Currie, Currie and Koppelhus2005 Troodon inequalis; Currie, p. 375.
Synonymy
See Table 1 for expanded synonymy.
Table 1. History and current status of Troodon formosus synonymies and revisions
* Generically synonymized but hesitant at the species level.
† Provisionally synonymized by Sternberg, Reference Sternberg1951, and Russell, Reference Russell1969, but not unequivocally until Carpenter, Reference Carpenter1982; later, Currie, Reference Currie1987, gave the most formal and detailed synonymy.
‡ Fragmentary material with significantly differing stratigraphy; associations unclear.
§ Synonymy contested; more analysis required.
Holotype
Tooth (ANSP 9259) from the Judith River Formation, Montana, U.S.A. by original designation (Leidy, Reference Leidy1856, Reference Leidy1860, pl. 9, figs. 53–55).
Diagnosis
A troodontid within Troodontinae that is distinguished from other troodontids by possessing a maxilla with anteriorly, a larger, more broadly rounded maxillary fenestra, a low-angled nasal process with a stepped anterior portion, bearing 23 teeth, and having a large palatal shelf extending posteriorly along the midline to the posterior limit of the maxillary fenestra; more pronounced basioccipital tubera (Currie, Reference Currie1985); an L-shaped to triangular frontal with a flat, shallowly anteroposteriorly rippled nasofrontal contact (van der Reest and Currie, Reference van der Reest and Currie2017); and a relatively short metatarsus (~0.66 metatarsal III/femur length in adult) with a flat to convex anterior surface and triangular to oval-shaped extensor fossa on metatarsal III. (Note: Currie, Reference Currie1987, listed several additional cranial features as diagnostic for T. formosus. However, these and other features needed to be evaluated in a thorough review of cranial material from both Montana and Alberta before their inclusion here.)
Occurrence
Descriptions are based on six Museum of the Rockies (MOR) specimens from the Campanian Two Medicine Formation of Montana: MOR 246, a clutch of eggs with embryos (Horner and Weishampel, Reference Horner and Weishampel1988; Varricchio et al., Reference Varricchio, Horner and Jackson2002); MOR 430, an associated skeleton of a small juvenile; MOR 563, an isolated large juvenile skeleton; MOR 553, a collection of elements representing multiple individuals of differing ontogenetic stages from a large dinosaur bonebed, Jack’s Birthday Site (Varricchio, Reference Varricchio1995); MOR 748, a partial adult skeleton associated with eggs (Varricchio et al., Reference Varricchio, Jackson, Borkowski and Horner1997); and MOR 11775, an isolated frontal. These specimens come from either Teton (MOR 246, 430) or Glacier (MOR 558, 563, 748, and 11775) County, Montana.
Specimens of Teton County are positioned in the Two Medicine Formation (Fig. 1) at a time of maximum regression (Horner et al., Reference Horner, Schmitt, Jackson and Hanna2001a; Rogers et al., Reference Rogers, Kidwell, Deino, Mitchell, Nelson and Thole2016, Reference Rogers, Horner, Ramezani, Roberts and Varricchio2025). MOR 246 was collected at Egg Island (MOR locality TM-024) and MOR 430 from the nearly stratigraphically equivalent Egg Mountain site (TM-006), approximately 1.4 km away. Both sit near the top of a 10- to 15-meter sequence dominated by lacustrine deposition (Shelton, Reference Shelton2007). Stratigraphic placement of this lacustrine carbonate interval (LCI) by Rogers et al. (Reference Rogers, Kidwell, Deino, Mitchell, Nelson and Thole2016) and Ramezani et al. (Reference Ramezani, Beveridgem, Rogers, Eberth and Roberts2022, fig. 6) suggests a date of approximately 77–76.5 Ma (Rogers et al., Reference Rogers, Horner, Ramezani, Roberts and Varricchio2025, fig. 15). Three specimens, MOR 553, 748, and 11775, occur both close geographically and stratigraphically (TM-068). The isolated MOR 11775 occurs close to a thick bentonite (Varricchio, Reference Varricchio1995) thought to correlate to 90TMT-590-U60 (Ramezani et al., Reference Ramezani, Beveridgem, Rogers, Eberth and Roberts2022; Rogers et al., Reference Rogers, Horner, Ramezani, Roberts and Varricchio2025) and listed as TM-4 in Rogers et al. (Reference Rogers, Swisher and Horner1993). MOR 748 and the nearby MOR 553 sit 12.5 and 14 meters, respectively, above this same bentonite. These specimens are all from relatively high in the formation, within 100 m below the contact with the overlying Bearpaw Shale in a thicker sequence of the Two Medicine Formation (Varricchio, Reference Varricchio1995; Rogers et al., Reference Rogers, Horner, Ramezani, Roberts and Varricchio2025). MOR 563 comes from an isolated patch of outcrops and is difficult to place stratigraphically (locality TM-071). However, the associated lithologies are more consistent with a position above MOR 748 and 553 (J. Horner, pers. comm., 2024; Rogers et al., Reference Rogers, Horner, Ramezani, Roberts and Varricchio2025).
Although there was some debate about which radiometric dates should be recognized (see Fowler, Reference Fowler2017, for discussion), the lithologic relationship of the Two Medicine Formation with both the Judith River Formation and Dinosaur Park Formation is consistently recognized, each lying beneath the Bearpaw Shale (Fig. 1). All the specimens here come from the upper Two Medicine Formation, recently recognized as the Flag Butte Member (Rogers et al., Reference Rogers, Horner, Ramezani, Roberts and Varricchio2025). This unit correlates to and encompasses the entirety of both the upper, fossil-producing portion of the Judith River Formation, the Dinosaur Park Formation, and likely at least portions of the Oldman Formation (Horner et al., Reference Horner, Schmitt, Jackson and Hanna2001a; Roberts et al., Reference Roberts, Deino and Chan2005; Rogers et al., Reference Rogers, Kidwell, Deino, Mitchell, Nelson and Thole2016, Reference Rogers, Horner, Ramezani, Roberts and Varricchio2025; Fowler, Reference Fowler2017; Beveridge et al., Reference Beveridge, Roberts and Titus2020; Ramezani et al., Reference Ramezani, Beveridgem, Rogers, Eberth and Roberts2022). Recent radiometric dates reflect these relationships, with the Flag Butte Member dating 76.99–74.78 Ma (Rogers et al., Reference Rogers, Horner, Ramezani, Roberts and Varricchio2025, fig. 15). Whereas embryonic and small juvenile specimens (MOR 546, 430) occur early in this range, the large juvenile and adult material (MOR 553, 563, 748, 11775) represent a much smaller interval closer to 75 Ma. In total, the specimens represent approximately 1.5 million years of the Flag Butte Member. The larger, ontogenetically older Two Medicine specimens span the most likely source of the type specimen from the Judith River and the ranges of “Stenonychosaurus inequalis” and “L. mcmasterae” as reported by van der Reest and Currie (Reference van der Reest and Currie2017).
Isolated troodontid teeth from Jack’s Birthday Site (MOR 553) have been used in several studies on comparative morphometrics of teeth from various Upper Cretaceous Formations of Montana and Alberta (Larson and Currie, Reference Larson and Currie2013; Torices et al., Reference Torices, Funston, Kraichy and Currie2014; Larson et al., Reference Larson, Brown and Evans2016; Evans et al., Reference Evans, Cullen, Larson and Rego2017). In each of these studies, these teeth are erroneously identified as representing the Judith River Formation, rather than the Two Medicine Formation. Importantly, these teeth represent the same assemblage of MOR 553 troodontid elements described here.
Description
Five elements from the Two Medicine Formation are here described. These include the maxilla, frontal, ilium, pubis, and metatarsal III.
Maxilla
Available maxillae from the Two Medicine Formation include an embryonic specimen (MOR 246-11), portions of both maxillae in the juvenile specimen (MOR 563), and four isolated specimens from Jack’s Birthday Site (MOR 553) (Fig. 2). The latter includes an individual equal in size to MOR 563 (MOR 553S-8-3-9-375) and three significantly larger specimens (MOR 553S-7-21-92-47, MOR 553S-7-28-91-246, and MOR 553S-8-11-92-205) (Table 2).
Table 2. Measurements (mm) of Troodon maxillae from the Two Medicine Formation; GL = greatest length; DV = dorsoventral; ML = mediolateral
*1st alveous damaged; + incomplete
The maxilla (Fig. 2) has a triangular profile in lateral view, consisting of a stout ventral ramus that curves anteromedially to contact the premaxilla and floor the narial fossa, a more slender nasal process rising posterodorsally at an angle of approximately 16° (range 14–20°), and a concave posterior margin representing the anterior boundary of the antorbital fenestra. These three also define the maxillary contribution to the antorbital fossa containing the antorbital and maxillary fenestrae separated by the interfenestral bar and a small promaxillary recess.
The ventral ramus is slightly convex below the narial fossa but becomes increasingly more concave posteriorly. This is particularly so below the ventral rim of the antorbital fossa which becomes ridge- then rod-like posteriorly. However, no specimen fully preserves this latter feature; therefore, its ultimate termination remains unknown.
An irregular line of neurovascular foramina runs parallel and dorsal to the alveolar margin. Foramina spacing increases posteriorly from 2–3 mm to occasionally over 10 mm for the most posterior in the larger specimens (MOR 553S-7-28-91-246 and -8-11-92-205). Each foramen opens ventrally to ventrolaterally and gives rise to a single or bifurcating groove extending ventrally, presumably accommodating blood vessels and nerves. Additional foramina occur closely interspersed among and around the line anteriorly, on the broad expanse anterior to the antorbital fossa, and as a line of some four to six running parallel to the anterior ventral border of the antorbital fossa. Some of these additional foramina have associated grooves extending ventrally, dorsally, or both.
The maxilla tapers drastically beginning at the level of the 18th or 19th tooth socket. Consequently, the last alveolus becomes strikingly shallow, with a maximum depth of 8.5 mm in MOR 553S-8-11-92-205, the largest specimen. An irregularly floored groove, ~3 mm wide in this specimen, angles posteroventrally the rod-like ventral rim of the antorbital fossa to just posterior to the last tooth socket. This likely represents the contact for the jugal. The angular ventroposterior termination of the maxilla consists of the confluence of the narrowing ventral ramus and a medial shelf of bone representing the posterior extension of the bony palate.
Anteriorly, the ventral ramus consistently extends the length of three alveoli beyond the junction of the nasal process. As in the premaxilla, a distinct change in both slope and bone texture demarcates the ventral border of the narial fossa. A dorsal-opening foramen sits at the posteroventral corner of the narial fossa at the base of the nasal process. A large triangular slot for the premaxilla marks the floor of the narial fossa and points posteriorly. This extends dorsomedially above and slightly posterior to this foramen and nearly or possibly reaching the recess for the subnarial process of the nasal. In MOR 553S-8-11-92-205 this potential gap measures less than 2 mm.
The maxilla sends forward a robust, anteromedial process that would dorsomedially cover the premaxilla. This process floors the remainder of the narial fossa and forms the anterior contribution of the maxilla to the secondary bony palate. This premaxillary process has a triangular cross-section with a subhorizontal ventral face dorsally overlying the palatal contribution of the premaxilla to the palate. The process extends as far forward as the posterior base of the nasal process of the premaxilla but fails to extend over the medial palatine fissure formed by the premaxillae. The apex of the triangular cross-section becomes increasingly more pronounced and ridge-like anteriorly. This ridge, with that of the complementary maxilla, defines a narrow medial passage above the palate, and between nares and the respective floors of the narial fossa, and connecting to the oral cavity through the single palatine fissure. The medial-facing foramina in the premaxillae, just dorsal to the fissure, would open into this space.
The alveolar suture for the premaxilla matches the corresponding face of that element, and consists of a porous sheet of bone topped by a neurovascular groove arching dorsomedially to lateroventrally to the subnarial foramen.
The contribution of the nasal process to the lateral, external aspect of the skull is broadest at its base anterior to the antorbital fossa. As the process arches posterodorsally, the lateral face quickly tapers, terminating as a posteriorly projecting point at or slightly posterior to the interfenestral bar. The external aspect of the process becomes increasingly more rugose and sculptured through ontogeny. Small specimens (e.g., MOR 553S-8-3-9-375) bear a relatively smooth texture (Fig. 2.6), but fine, irregular bumps and larger grooves mark the lateral aspect of large individuals (e.g., MOR 553S-7-28-91-246, MOR 553S-8-11-92-205). In these, sculpturing continues across the entire length of the external face of the nasal process.
The dorsal surface of the nasal process consists of a longitudinal groove for the nasal (Fig. 2.1). Above the maxillary fenestra, the groove is broad and relatively shallow. Moving anteriorly it narrows and deepens before eventually shallowing near its forward limit. The recess becomes increasingly shallow as it passes posteriorly above the interfenestral bar. Here, a triangular slot for the lacrimal sits lateroventrally. The medial aspect of the nasal process is broadly concave throughout its length representing the lateral margin of the dorsal nasal meatus.
Across ontogeny and among individuals the antorbital fossa retains a consistent form bearing a distinct perimeter throughout and a broadly rounded anterior margin (Fig. 2). In contrast, the contained fenestrae, interfenestral bar, and interconnecting sinus show fairly extensive individual variation. In most specimens the anterior margin of the antorbital fenestra is more broadly rounded than that of the fossa. A broad, hourglass-shaped interfenestral bar, in-set within the fossa, separates the antorbital fenestra from the ellipse-shaped maxillary fenestra. In MOR 553S-8-3-9-375 and MOR 553S-8-11-92-205, the interfenestral bar stands vertically and the maxillary fenestra is symmetrical or reaches its maximum height more posteriorly.
In contrast, the interfenestral bar of MOR 553S-7-28-91-246 angles anterodorsally, defining a more asymmetric maxillary fenestra with its greatest height anterior to its midlength. A still more radical condition occurs in MOR 553S-7-21-92-47, which completely lacks an interfenestral bar. In all other aspects the morphology of this maxilla conforms with other Two Medicine Formation specimens, such that the variation in sinus features is insufficient evidence to recognize this as a form distinct from Troodon formosus.
The maxillary fenestra opens medially into a chamber completely encapsulated by exceedingly thin bone. Only MOR 553S-8-3-9-375 preserves this maxillary fossa largely intact. In all other specimens, the thin walls, only 0.3–0.35 mm thick in the largest maxilla, have largely been broken away. The fossa has the form of an oblate ellipsoid with the longest axis running anteroposteriorly and the shortest dorsoventrally. The fossa extends medially above the secondary palate approaching the midline; anteriorly, medial to the base of the nasal process; and posteriorly as both a pocket into the anteromedial aspect of the interfenestral bar and as a more medial expansion posterior to the bar.
A fairly large recess extends anteriorly from the antorbital fossa into the base of the nasal process lateral to the maxillary fossa. This space may be homologous to the accessory maxillary or promaxillary fenestra of other troodontids and theropods. This recess connects through two large openings to the anterior end of the maxillary fossa in MOR 553S-7-28-91-246 or may be completely isolated, as in MOR 553S-8-11-92-205. This feature was described previously as the maxillary sinus in NMC 12392 from the Dinosaur Park Formation (Currie, Reference Currie1985), with connections anterior to the external narial opening and dorsally to the nasal passages. The Two Medicine specimens lack such openings, however NMC 12392 is fragmentary, so these may represent breaks or reflect variability in sinus morphology.
In posterior view, the interfenestral bar broadens ventrally creating a posteriorly directed concavity. At its base, a broad but low opening passes forward and opens just anterior to the interfenestral bar in the floor of the maxillary fossa. This passage is present even in MOR 553S-7-21-92-47, the maxilla lacking an interfenestral bar. The anterior opening of this passage varies from a single round opening (MOR 553S-7-21-92-47), a very large oval opening (MOR 553S-7-28-91-246), to two round openings (MOR 553S-8-11-92-205). Presence of these openings suggests a sinus space exists lateral to the ventral nasal meatus, anterior to the interfenestral bar and dorsal to the tooth row, and ventral/beneath the floor of the anterior portion of the antorbital fossa.
The maxilla bears a broad medial shelf (Fig. 2.7). This extends to the midline from the premaxillary process posterior to the level of the interfenestral bar. The shelf narrows posterior to this point, angling to the posterolateral limit of the bone. The broad anterior portion of the shelf arches to create a broad, ventrally concave palatal roof. One or two foramen may pierce the palatal shelf ventrally. Two features mark the palatal shelf dorsally. First, a broad groove, ~5 mm wide in MOR 553S-8-11-92-205, wraps around the anterior extension of the maxillary fossa and ends blindly between the fossa and the articular recess for the nasal. The second feature consists of a longitudinal depression running along the midline of the premaxillary process to the level of the anterior margin of the maxillary fenestra. This represents the posterior continuation of the midline passage connecting to the palatine fissure. A single maxilla contributes only to the ventral and lateral margin of this passage. Additionally, a stout, bumpy ridge likely accommodating tooth roots runs from near the posterior limit of the medial shelf longitudinally disappearing beneath the interfenestral bar.
The median edge of the palatal shelf remains relatively thin and irregular over its anterior third. It then thickens and develops a vertical contact with anteroventrally oriented striations. The shelf reaches a maximum thickness as the shelf angles away from the midline and just posteromedial to the arched roof of the palate (Fig. 2.4). This would accommodate the anteroposteriorly shortened vomer. The next section, ~15 mm of edge in MOR 553S-8-11-92-205, is thin, smooth, and rounded, suggesting an absence of a bony contact. This corresponds to the anterolateral border of the choanae. The final posteriormost section again bears a slightly roughened edge. Ventrally in this region the shelf exhibits faint irregular longitudinal lineations and a lightly depressed margin. Both features represent the attachment for the palatine.
On the medial side, the bone surface ventral to the palatal shelf down to the dental foramina remains generally smooth. Two exceptions include the area for the palatine attachment discussed above and a patch of anteroventral striations just below the shelf and some distance dorsal to tooth sockets 6–10. Ventral to the dental foramina sits the scalloped medial wall of the alveoli, bearing numerous thin foramina and channels. Interdental plates are present through the length of the maxilla. The three largest maxillae (MOR 553S-7-21-92-47, MOR 553S-7-28-91-246, MOR 553S-8-11-92-205), which represent three individuals, each possess 23 alveoli. The embryonic specimen, MOR 246-11, appears to have at least 20 teeth. Consequently, there appears to be little or no increase in tooth count through ontogeny. The quadrangular tooth sockets are relatively small anteriorly, slowly increase through positions 4–10, reach their maximum size generally in positions 11–19, then rapidly decrease posteriorly (Table 3). Alveoli 23 and 22 followed by 1 and 2 represent the smallest tooth positions in the maxilla. Teeth in all maxillae are either of embryonic form (Varricchio et al., Reference Varricchio, Horner and Jackson2002), poorly preserved, or completely absent. Description of shed and rooted teeth from MOR 553 will await a more comprehensive study on the entire set of cranial elements.
Table 3. Alveolar lengths (mm) in Troodon maxillae
*shortened by crushing; T = fore-aft basal length of teeth
The maxilla largely defines the shape of the nasal cavity and several anterior sinus spaces. With the premaxillae, vomer, and nasal, the maxillae form a nasal cavity that would have an hourglass cross section posterior to the nares. The medial extensions of the maxillary fossae nearly separate a dorsal nasal meatus directed toward the olfactory lobes from a ventral nasal meatus connecting to the internal choanae. A narrow midline passage extends anteriorly from the ventral meatus to the palatine fissure.
A complex sinus invades the maxilla lateral to the nasal cavity. This extends forward from the antorbital fenestra below the interfenestral bar to beneath the floor of the anterior portion of the antorbital fossa. The sinus connects up though the floor into the antorbital fossa, continuing forward within the promaxillary recess and medially through the maxillary fenestra into the thin-walled maxillary fossa. These last two spaces may share a connection within the base of the nasal process. A sinus may also extend into the premaxillary process. In several specimens, this structure is partially collapsed, suggesting the presence of an open air-filled or marrow-filled space within.
The maxilla appears to deepen from the embryo to the large juvenile state (Fig. 2). Over the size range from the juvenile MOR 563 and MOR 553S-7-21-92-47 to their largest, there are no discernible changes in the overall proportions of the maxilla, but some increase in the lateral, surficial rugosity.
Maxilla comparisons
As in Saurornithoides mongoliensis and Zanabazar junior (Barsbold, Reference Barsbold1974), the Troodon maxilla contributes to a significant portion of the lateral aspect of the snout. The maxillae of these three taxa share similar proportions and differ from the reduced, elongate element of Mei long, and the deep, blunt maxilla of Sinovenator changii, Xu et al., Reference Xu, Norell, Wang, Makovicky and Wu2002. The three larger taxa also share deep exposure of the maxilla beneath the naris. As a result, the dorsal border of the facial contribution of the maxilla consists initially of a step up from the tooth row before following the angle of the nasal process. This contrasts markedly with Byronosaurus jaffei, Norell et al., Reference Norell, Makovicky and Clark2000 where the naris extends virtually to the tooth row and the dorsal border of the facial contribution is an uninterrupted straight line.
The three maxilla morphologies of troodontids result from differences in the angle and form of the nasal process as well as the overall snout profile. Short-snouted species such as Sinovenator, Mei long, and Jianianhualong have steeply angled processes (Xu et al., Reference Xu, Norell, Wang, Makovicky and Wu2002, Reference Xu, Currie, Pittman, Xing, Meng, Lü, Hu and Yu2017; Xu and Norell, Reference Xu and Norell2004). Others (Gobivenator, Byronosaurus, Almas) have nasal processes that form a line running nearly uninterrupted down to the tooth row, giving a sharper triangular form to the snout (Makovicky et al., Reference Makovicky, Norell, Clark and Rowe2003; Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014; Pei et al., Reference Pei, Norell, Barta, Bever, Pittman and Xu2017). In contrast, Saurornithiodes, Zanabazar (Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009), and Troodon have longer snouts with a lower angle nasal process with a distinct step anteriorly.
Tooth count in Troodon is lower than that of Mei long (24) but higher than Zanabazar junior (20); Saurornithoides mongoliensis, Gobivenator mongoliensis Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014, and Sinusonasus (~19); and Almas ukhaa Pei et al., Reference Pei, Norell, Barta, Bever, Pittman and Xu2017 (17) (Xu and Norell, Reference Xu and Norell2004; Xu and Wang, Reference Xu and Wang2004; Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009; Tsuihiji et al. Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014; Pei et al., Reference Pei, Norell, Barta, Bever, Pittman and Xu2017). The tooth row extends well posterior of the interfenestral bar.
Both the premaxilla and maxilla contribute to the ventral border of the naris as in most other troodontids, such as Sinornithoides, S. mongoliensis, Zanabazar, and Byronosaurus. In Troodon a triangular process of the premaxilla fits into a corresponding slot in the maxilla. Consequently, the maxilla borders this process both ventrolaterally and mesodorsally in the floor of the narial chamber. This complex articulation may be unique to Troodon or may be unobservable or unclear in articulated specimens. The premaxillary suture in the tooth row is located at the level of the naris, not anterior to it as in Byronosaurus. As in Byronosaurus and Saurornithoides mongoliensis, Troodon bears a foramen in the posteroventral corner of the naris.
The maxilla contributes to an extensive secondary palate similar to that of Byronosaurus, Sinornithoides, Saurornithoides mongoliensis, and Gobivenator mongoliensis, extending from the premaxillae to choanae medial to the antorbital fenestra (Makovicky and Norell, Reference Makovicky, Norell, Weishampel, Dodson and Osmólska2004; Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014). The maxillary contribution of T. formosus would have been even more extensive than that of Gobivenator (Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014), because it only accommodates the foreshortened vomer at its posterior margin.
In general, troodontids lack a promaxillary fenestra (Makovicky and Norell, Reference Makovicky, Norell, Weishampel, Dodson and Osmólska2004). In Troodon, a recess extends from the antorbital fossa anteriorly into the base of the nasal process. This space variably connects to the maxillary fossa. This feature differs from the laterally visible promaxillary fenestra of Sinusonasus and Sinovenator (Xu et al. Reference Xu, Norell, Wang, Makovicky and Wu2002; Xu and Wang, Reference Xu and Wang2004). However, it approaches the condition in Sinornithoides, which possesses a small foramen within an anterior recess (Currie and Dong, Reference Currie and Dong2001). The large, elliptical maxillary fenestra of Troodon compares to those of other troodontids like Byronosaurus, Sinornithoides, Saurornithoides, and Zanabazar. It appears to begin posterior to the naris and terminates anterior to the antorbital fenestra, as in Zanabazar, but unlike Byronosaurus and Sinovenator (Xu et al. Reference Xu, Norell, Wang, Makovicky and Wu2002; Makovicky et al., Reference Makovicky, Norell, Clark and Rowe2003; Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009). As in at least some troodontids (e.g., Byronosaurus and Zanabazar) (Makovicky et al., Reference Makovicky, Norell, Clark and Rowe2003; Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009), the fenestra opens into a space fully closed off by bone within the maxilla.
The interfenestral bar stands inset within the antorbital fossa, as in Sinornithoides, Saurornithoides mongoliensis, and Zanabazar, contrasting with the condition in Mei long and Byronosaurus (Currie and Dong, Reference Currie and Dong2001; Makovicky et al., Reference Makovicky, Norell, Clark and Rowe2003; Xu and Norell, Reference Xu and Norell2004; Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009). Although a ventral passage occurs between the antorbital and maxillary fenestrae, as in Byronosaurus, S. mongoliensis, and Zanabazar, that of Troodon differs in two ways. First, the connection lies largely ventral to the entire bar. Secondly, the passage first connects to a completely enclosed lateral sinus before continuing dorsally to the maxillary fenestra. A portion of the nasal passage does not pass through the interfenestral bar of Troodon, as in Byronosaurus and Zanabazar (Makovicky et al., Reference Makovicky, Norell, Clark and Rowe2003; Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009). Instead, the bar forms the lateral margin of the dorsal nasal meatus. Also, the interfenestral bars do not appear to meet dorsally, as described for other troodontids (Makovicky and Norell, Reference Makovicky, Norell, Weishampel, Dodson and Osmólska2004). The maxilla of Troodon thins posteriorly below the antorbital fenestra, as in Byronosaurus and Zanabazar (Makovicky et al., Reference Makovicky, Norell, Clark and Rowe2003; Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009). Here it flattens for the articulation of the palatine, as in Zanabazar. The maxillae of both taxa are laterally concave posterior and ventral to the maxillary fenestra (Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009).
Frontal
In dorsal view, the frontals from the Two Medicine Formation have an approximately L-shaped outline with a long medial interfrontal suture and a shorter postorbital ramus projecting laterally (Fig. 3). A broad parasagittal trough runs from the nasal contact posteriorly to a low rise corresponding to the underlying cerebral fossa. Anteriorly, the trough accentuates the raised lateral orbital rim, which begins from a shallow lateral indentation for the lacrimal and extends posteriorly with only a slight lateral deviation back to the level of the anterior margin of the cerebral fossa. More posteriorly, the rim curves sharply posterolaterally producing an L-shaped outline, rather than the more triangular morph of “Latenivenatrix mcmasterae.” A distinct, primary supraciliary foramen, as in Albertavenator curriei, was not recognized in these specimens. The orbital rim comprises the anterior aspect of the postorbital ramus, whereas a prominent posterolaterally facing rugose concavity, the postorbital contact, marks the posterior aspect. A distinct but thin supratemporal ridge runs from this contact medially to terminate at the contact for the parietal. A shallow depression sits anterior to this ridge and lateral to the dorsal bulge of the cerebral fossa and faces posterolaterally. The anterior margin of the temporal fossa is steeply angled. The contact for the parietal consists of a small rectangular patch adjacent to the midline marked by vertical ridges facing posteriorly. Lateral to this and angling ventrolaterally is a lappet that completes the parietal contact.
The nasofrontal suture is not fully preserved in any Two Medicine specimen. However, it appears to gradually angle from near the midline to the orbit and to have accommodated a thin overlying extension of the nasal. A distinct dorsolateral abutment, as in “L. mcmasterae” specimens, is lacking from the lateral margin of the suture in the Two Medicine specimens.
The ventral aspect of the frontal bears an abutment for the lacrimal near the anterior end of the orbital rim. This feature sits on the lateral margin of the crista cranii that outlines the elongate olfactory bulbs, and more posteriorly, the long olfactory tracts and enlarged cerebral fossa. A narrow (< 1 mm), distinct groove runs along the lateral margin of the olfactory tract at the medial base of the crista cranii from the cerebral to olfactory fossa. It remains unclear whether this feature is present in “L. mcmasterae” because only a very small length may be present in TMP 1979.008.0001.
The laterosphenoid contact extends from the margin of the large cerebral fossa laterally and slightly anteriorly. It consists of a shallow slot bordered anteriorly by a thin wall, and posteriorly by a shorter but more robust one. This feature differs from that in TMP 1979.008.0001 where the contact lacks a distinct posterior wall producing a flat, notch-like rather than slot-like contact.
Frontal comparisons
The frontals from the Two Medicine Formation specimens clearly represent those of troodontids because they possess several features typical of the group. The frontals exhibit a raised supraorbital rim with a medial trough as well as slightly bulbous expansions dorsal to the cerebral hemispheres, as in Almas, Gobivenator, Mei, Jianianhualong, Linhevenator tani, Xu et al., Reference Xu, Tan, Sullivan, Han and Xiao2011, Sinovenator, Xixiasaurus, and Zanabazar (Xu et al., Reference Xu, Norell, Wang, Makovicky and Wu2002, Reference Xu, Tan, Sullivan, Han and Xiao2011, Reference Xu, Currie, Pittman, Xing, Meng, Lü, Hu and Yu2017; Xu and Norell, Reference Xu and Norell2004; Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009; Lü et al., Reference Lü, Xu, Liu, Zhang, Jia and Ji2010; Gao et al., Reference Gao, Morschhauser, Varricchio, Liu and Zhao2012; Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014; Pei et al., Reference Pei, Norell, Barta, Bever, Pittman and Xu2017). Although often not visible, a narrow olfactory tract and elongate olfactory bulb, as seen in Albertavenator, Sinornithoides youngi Russell and Dong, Reference Russell and Dong1993, and T. formosus, may also characterize the group (Currie and Dong, Reference Currie and Dong2001; Evans et al., Reference Evans, Cullen, Larson and Rego2017; van der Reest and Currie, Reference van der Reest and Currie2017).
Among troodontids, T. formosus appears most similar to taxa such as Xixiasaurus, Gobivenator, and Zanabazar. These taxa share a roughly triangular outline in dorsal view due to a prominent lateral extension of the postorbital process (Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009; Lü et al., Reference Lü, Xu, Liu, Zhang, Jia and Ji2010; Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014; van der Reest and Currie, Reference van der Reest and Currie2017), in contrast to the stouter or more elongate frontals of Albertavenator (Evans et al., Reference Evans, Cullen, Larson and Rego2017) and Mei, Linhevenator, and Almas (Xu and Norell, Reference Xu and Norell2004; Xu et al., Reference Xu, Tan, Sullivan, Han and Xiao2011; Pei et al., Reference Pei, Norell, Barta, Bever, Pittman and Xu2017), respectively. As in Zanabazar and Albertavenator (Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009; Evans et al., Reference Evans, Cullen, Larson and Rego2017; van der Reest and Currie Reference van der Reest and Currie2017), a distinct transverse ridge marks the dorsal surface of the frontal in T. formosus where it slopes steeply posteriorly as the anterior margin of the supratemporal fossa (Fig. 3). Gobivenator has a similar ridge but a much shallower slope to the fossa (Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014). Finally, the frontals of Zanabazar (Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009) and T. formosus share a more angled contact for the nasal, a parasagittal trough near the midline, and a shallow depression just anteromedial to the postorbital contact (Currie, Reference Currie1985).
Using the length and width dimensions (as defined by Cullen et al., Reference Cullen, Zanno, Larson, Todd, Currie and Evans2021), the two most complete Two Medicine specimens largely plot with L-shaped frontals (Fig. 4). However, as noted by Cullen et al. (Reference Cullen, Zanno, Larson, Todd, Currie and Evans2021), there is considerable overlap among the L-shaped and triangular morphs. Additionally, the posteroventral contact for the laterosphenoid consists of a shallow slot bordered by thin anterior and squat posterior walls (Fig. 3), rather than a more groove-like form as present in “L. mcmasterae”-type frontals (Currie, Reference Currie1985, fig. 3b). Whether this is a taxonomically significant feature remains to be determined.
Ilium
Only partial ilia are preserved in MOR 748, MOR 430, and MOR 563, with the best specimen represented by the embryonic material of MOR 246 (#11). In the latter, the right ilium, femur, and pubis remain in articulation (Fig. 5). A fracture splits the egg and cuts through the ilium and femur in a parasagittal plane. This affords a view of the overall shape but not of any surface features. The anterior end is not observable, remaining buried within the matrix. The ilium appears long and low with a dorsal–ventral height being shorter than the length of acetabular region from anterior of pubic peduncle to posterior of ischial peduncle. The dorsal margin remains relatively horizontal, dipping slightly above the acetabulum then slopping ventrally posterior to ischial peduncle. The post-acetabular region is nearly as long as the acetabular region and tapers to a rounded point. Its ventral margin is nearly horizontal and slightly concave. The pubic peduncle is longer anteroposteriorly and extends slightly farther ventrally than the ischiadic peduncle.
The small juvenile, MOR 430, preserves portions of both ilia, including the post-acetabular and acetabular regions, the ischiadic peduncle, and some of the pubic peduncle (Fig. 5). The preserved outline matches that of the embryo. The preserved post-acetabular blade is slightly concave in lateral aspect. A medially directed horizontal shelf runs the length of the post-acetabular portion about mid-height, thus dividing this portion into roughly equivalent dorsal and ventral halves. This shelf’s medial edge hangs down slightly and helps define a slightly concave, ventrally located depression, the brevis fossa. A small concavity sits just dorsal to the ischial peduncle on the medial side of the ilium. The ilium’s contribution to the acetabulum begins anteriorly with a slightly concave surface. As it arches upwards it becomes strongly concave and remains so through its apex. It then rapidly widens to a planar surface rotated laterally outwards. Thus, the posterior portion of the acetabulum forms a broad anterolateral and ventrally directed planar surface as the antitrochanter occupying most of the ischial peduncle. The entire surface has a smooth glossy finish, indicating it was covered in life with articular cartilage. The ischiadic peduncle thus seems to lack a distinct face for contact with the ischium and instead appears to be relatively small and rounded. However, the roundness may reflect the immaturity of the specimen.
Ilium comparison
The incomplete condition of all the Two Medicine ilia make extensive comparisons difficult. Based on the embryonic (MOR 246) and small juvenile (MOR 430) specimens, T. formosus had a relatively longer ilium with an ilium/femur ratio well above 0.60. This would make it more similar to Daliansaurus (0.63) (Shen et al., Reference Shen, Lue, Liu, Kundrat, Brusatte and Gao2017) and Almas (0.83) (Pei et al., Reference Pei, Norell, Barta, Bever, Pittman and Xu2017) and unlike the shorter ilia (≤ 0.60) of Sinornithoides, Jianianhualong, Sinovenator, and Mei long (Currie and Dong, Reference Currie and Dong2001; Xu et al., Reference Xu, Norell, Wang, Makovicky and Wu2002, Reference Xu, Currie, Pittman, Xing, Meng, Lü, Hu and Yu2017; Gao et al., Reference Gao, Morschhauser, Varricchio, Liu and Zhao2012) Similarly, the ilium appears dolichoiliac, as in Almas, “L. mcmasterae,” and Sinornithoides (Currie and Dong, Reference Currie and Dong2001; Pei et al., Reference Pei, Norell, Barta, Bever, Pittman and Xu2017; van der Reest and Currie, Reference van der Reest and Currie2017), but with possibly a slightly concave dorsal margin to the iliac blade above the acetabulum.
The ventral margin of the postacetabular blade appears nearly horizontal, like that of Gobivenator, Sinovenator, and “L. mcmasterae” (Xu et al., Reference Xu, Norell, Wang, Makovicky and Wu2002; Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014; van der Reest and Currie, Reference van der Reest and Currie2017). A brevis shelf, as in Almas (Pei et al., Reference Pei, Norell, Barta, Bever, Pittman and Xu2017), runs the length of the posterior blade on its medial face.
The pubic peduncle extends farther ventrally than the ischiadic peduncle, as in most troodontids (Currie and Dong, Reference Currie and Dong2001; Gao et al., Reference Gao, Morschhauser, Varricchio, Liu and Zhao2012; Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014; Shen et al., Reference Shen, Lue, Liu, Kundrat, Brusatte and Gao2017; van der Reest and Currie, Reference van der Reest and Currie2017; Xu et al., Reference Xu, Currie, Pittman, Xing, Meng, Lü, Hu and Yu2017) but provides few other details. As with Gobivenator (Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014), Troodon formosus possesses a supracetabular rim overhanging the acetabulum over much of its anterior half. Similar rims are also observed in Sinornithoides (Currie and Dong, Reference Currie and Dong2001) but are more extensive in Mei long and Anchiornis huxleyi, Xu et al., Reference Xu, Zhao, Norell, Sullivan, Hone, Erickson, Wang, Han and Guo2009 (Gao et al., Reference Gao, Morschhauser, Varricchio, Liu and Zhao2012).
The antitrochanter of T. formosus, like that of Sinornithoides (Russell and Dong, Reference Russell and Dong1993), occupies much of the lateral aspect of the ischiadic peduncle behind the acetabulum and differs from the more dorsal location in Daliansaurus (Shen et al., Reference Shen, Lue, Liu, Kundrat, Brusatte and Gao2017). As with Mei long (Gao et al., Reference Gao, Morschhauser, Varricchio, Liu and Zhao2012), the ischial peduncle of T. formosus lacks a clear facet for the ischium.
Pubis
In contrast to the ilium, much better samples of Troodon pubis specimens exist (Fig. 5). The samples include: a complete adult pubis and partial juvenile pubis from MOR 553, a shaft fragment from MOR 748, a nearly complete pubis from MOR 430, and visible portions of a pubis in the embryo MOR 246 (#11).
The medio-laterally narrow proximal pubis end is concave laterally and convex medially. It possesses a small, oval, and posterior-facing ischial contact; a slightly longer and gently concave acetabular region; and a longer ilium contact oriented perpendicular to the ischial contact. The ilium contact, which makes up a majority of the proximal end, tapers anteriorly and curves laterally. This contributes to the concave lateral side of the proximal end. A ventrally facing obturator notch sits adjacent and just anterior to the ischial suture. The surfaces for the ilium and ischium articulations lay perpendicular to one another with the shaft of the pubis running parallel to the latter.
The proximal end quickly tapers to form the shaft. The shaft has a laterally compressed oval cross-section initially. About one-third down its length, a medially directed ridge begins. This increases in size and likely contacted a similar ridge from the opposite pubis over the last one-third of the its length forming the pubic symphysis and apron. The pubis is weakly concave behind this ridge. Notably, a flat, striated and posteromedially facing surface marks the shaft just behind (posterior) to the proximal portion of this ridge. The shaft arcs inwards towards its distal end but is otherwise straight in lateral view.
In the embryo, a shaft emerges beneath the right femur; its long axis paralleling the pubic peduncle of the ilium (Fig. 5.7). This has a diameter of ~1.5 mm and appears as a hollow cylinder. The upper (proximal) end is not observable, possibly hidden by the femur. The distal end does not bear a boot, but porous bone and extremely hollow shafts suggest the boot may not have been ossified at this early ontogenetic stage. Although it is impossible to orient the ilium perfectly, its position suggests that the pubic shaft runs close to vertical. The pelvis likely was neither strongly pro-pubic nor opisthopubic.
Both the small juvenile and the largest MOR 553 specimens possess a large distal boot. The latter shows a small posterior projection and a large anterior projection. This contrasts with the juvenile pubis, MOR 430. Although the MOR 553 specimen is crushed, the greater development of the anterior portion of the boot seems real. The dorsal surface of this portion of the boot is complete and relatively damage free. Presumably, the anterior portion of the boot developed more rapidly than the posterior one through ontogeny.
The boot has a flattened medial side, where it would have abutted the boot of the opposite pubis. Bone on the dorsal surface bears irregular grooves, suggesting some type of soft-tissue attachment. The ventral (distal) surface of the boot is slightly convex overall, bears irregular pits and bumps, and on a fine scale has a grainy finish. The surface contrasts sharply from the smooth bone of the shaft and resembles that of the ilium and ischium contacts. In life, the bottom portion may have born some type of cartilage or ligamentous/tendonous attachment.
Pubis comparison
The long straight pubis of Troodon formosus is common to most troodontids. The relative length proportions of pubis to ischium, although not clearly known, appears closer to those of Gobivenator, Almas, and Jianianhualong (Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014; Pei et al., Reference Pei, Norell, Barta, Bever, Pittman and Xu2017; Xu et al., Reference Xu, Currie, Pittman, Xing, Meng, Lü, Hu and Yu2017) than the more extreme ratios of Sinornithoides and Sinovenator (Currie and Dong, Reference Currie and Dong2001; Xu et al., Reference Xu, Norell, Wang, Makovicky and Wu2002). Troodon formosus, Gobivenator, “L. mcmasterae,” and Saurornithoides mongoliensis distinctively share an anteroposteriorly expanded articulation for the ilium (Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009; Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014). This expansion likely would have joined with a corresponding one in the ilium of T. formosus to form an ambiens process, as in “L. mcmasterae” (van der Reest and Currie, Reference van der Reest and Currie2017), but absent in other troodontids (e.g., Mei long and Sinovenator) (Xu et al., Reference Xu, Norell, Wang, Makovicky and Wu2002; Gao et al., Reference Gao, Morschhauser, Varricchio, Liu and Zhao2012). Other attributes shared by T. formosus and the isolated partial pelvis assigned to “L. mcmasterae” (UALVP 55804) include the overall cross-sectional form of the pubic shaft, a vertically oriented ischial peduncle, and the anterolateral curvature of the long iliac articulation. The latter appears more pronounced in T. formosus.
The poorly preserved pubic apron may have extended over only the distal 40% of the element in T. formosus. As in Gobivenator (Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014), T. formosus possessed a distal boot, although it appears more extensive anteriorly than posteriorly in the adult. Further, like Gobivenator (Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014), the pubis in T. formosus was oriented nearly vertically. This condition differs from the retroverted or propubic orientation of Sinovenator (Xu et al., Reference Xu, Norell, Wang, Makovicky and Wu2002) and Mei long and S. mongoliensis (Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009; Gao et al., Reference Gao, Morschhauser, Varricchio, Liu and Zhao2012). Throughout ontogeny, the pubic shaft of T. formosus lacks the distinct anterior curvature found in “L. mcmasterae” (van der Reest and Currie, Reference van der Reest and Currie2017).
Metatarsal III
Metatarsal III, an unusual element for the shaft, becomes markedly narrow; at its minimum, the shaft diameter is only nearly 1.0% that of the greatest length. Unsurprisingly, nearly all specimens are incompletely preserved. Three MOR 553 specimens include a significant portion of the shaft whereas MOR 748 preserves metatarsal III in its entirety (Fig. 6).
The metatarsal begins with a wedge-shaped proximal end that fits between the posterior portions of metatarsals II and IV. These two metatarsals meet anteriorly and obscure the third metatarsal from the extensor surface proximally. The shaft rapidly narrows from the proximal end. Over the first 40% of its length the shaft is at its thinnest both mediolaterally and anteroposteriorly. In MOR 748, the third metatarsal has a greatest length of 212 mm, but over this proximal portion, the shaft is only 2–3 mm wide and not much more than 1 mm deep. Throughout this stretch, the strap-like shaft lies at the posterior edge of metatarsals II and IV. This position and the deep shafts of the adjacent metatarsals create a deep longitudinal furrow on the extensor surface of the proximal metatarsus.
Beginning at about one-third down its length, the metatarsal angles to the anterior surface of the metatarsus. This corresponds to a change in the shaft cross-section from strap-like to triangular with a flat to slightly convex anterior face and an attenuated apex pointing posteriorly. Because of these changes, over much of the final third of its length, metatarsal III sits wedged between and slightly anterior to metatarsals II and IV (Fig. 6).
The shaft also exhibits a slight lateral bend, so that the distal articular surface sits slightly more lateral than the proximal. This curvature is readily visible in flexor view of isolated metatarsals, here the generally straight posterior ridge arcs laterally just prior to the distal articular surface. This marks the point where metatarsal III emerges from between the other two metatarsals and the posterior ridge, which had been very narrow and sharp, becomes much broader and rounded.
The distal articular surface is ginglymoid, with just a shallow groove separating the two condyles. Although the condyles have similar mediolateral widths, the larger medial condyle has a greater radius of curvature. On the flexor surface, the articulation continues proximolaterally as an asymmetric tongue-like projection.
On the extensor aspect, a distinct but shallow depression lies just proximal to the articular surface. The central portion of this hollow remains smooth throughout ontogeny, but its proximal and lateral borders gain a roughened texture in the largest individuals. In all specimens, this lateral border also projects laterally from the edge of the bone. Well-developed collateral ligament pits occur on both sides of the distal end. Typically, a second shallower and more longitudinally oriented pit lies proximoposteriorly to the lateral ligament pit. All have a roughened surface.
Throughout ontogeny the proportions of the distal articular surface change. In the smallest specimen, MOR 430, the ratio of the mediolateral to anteroposterior dimensions is only 0.56; however, as the mediolateral dimension increases from 5.9, 16.1, 20.3, 21.9, to 24.3 mm, the ratio also increases from 0.56, 0.78, 0.82, 0.85, to 0.87 in the best-preserved specimens (MOR 430, MOR 553S-7-29-92-113, MOR 748, MOR 553S-7-16-91-79, and MOR 553-11-1-01-7).
In comparison to previous metatarsus reconstructions for Troodon, metatarsals II and IV have only a short proximal contact on the extensor surface (contra Wilson and Currie, Reference Wilson and Currie1985, fig. 5d) and a long distal contact along the flexor surface. Also, the Troodon metatarsus bears a longitudinal trough running the proximal half on the flexor surface.
Metatarsal comparison
Metatarsal III of Troodon formosus, like those of most troodontids, is a highly modified element with a reduced proximal end, very narrow proximal shaft that expands with a triangular cross-section distally, and a robust distal articulation (Makovicky and Norell, Reference Makovicky, Norell, Weishampel, Dodson and Osmólska2004). Metatarsal III is longer than both metatarsal II and IV and its proximal end is only visible posteriorly. This form fits into the derived arctometatarsalian condition characteristic of most troodontids (Zanno et al., Reference Zanno, Varricchio, O’Connor, Titus and Knell2011), with perhaps the exception of Sinovenator and Mei long (Xu et al., Reference Xu, Norell, Wang, Makovicky and Wu2002, Gao et al., Reference Gao, Morschhauser, Varricchio, Liu and Zhao2012).
Among troodontids, Troodon formosus has a relatively short metatarsus. The metatarsal III to femur length ratio in MOR 748 of 0.66 exceeds only that of Linhevenator (0.63) (Xu et al., Reference Xu, Tan, Sullivan, Han and Xiao2011) but is surpassed by those of Sinovenator (0.69), Mei long (0.75), Sinornithoides (0.79), Anchiornis (0.83), Daliansaurus (0.84), and Philovenator (1.25) (Shen et al., Reference Shen, Lue, Liu, Kundrat, Brusatte and Gao2017). Much of this variation likely reflects scaling, as the negative allometry of theropod metatarsus length is well established (Holtz, Reference Holtz1995) and only Linhevenator approaches the large size of T. formosus among these troodontid taxa.
The distal articulating surface extends posteroproximally in Tochisaurus and Sinornithoides (Kurzanov and Osmólska, Reference Kurzanov and Osmólska1991; Currie and Dong, Reference Currie and Dong2001), but only in T. formosus, Borogovia, and IGM 100/44 does it take on a long, broad, tongue-like form (Russell, Reference Russell1969). Several troodontids (e.g., Philovenator, Tochisaurus, and Talos) (Kurzanov and Osmólska, Reference Kurzanov and Osmólska1991; Xu et al., Reference Xu, Tan, Sullivan, Han and Xiao2011, van der Reest and Currie, Reference van der Reest and Currie2017) possess an extensor fossa just proximal to the distal articulation. This was described as semi-circular to sub-circular in “L. mcmasterae” and sub-triangular in T. formosus by van der Reest and Currie (Reference van der Reest and Currie2017), but Two Medicine material shows some variation in the fossa outline. In two smaller specimens (MOR 553S-7.29.92.113, MOR 553S-8.6.9.406) with mediolateral articular widths (ML) of 16.1 mm and 17.4 mm, respectively, the fossa has a triangular shape pointing back up the shaft. In larger specimens, MOR 553S-7.16.91.79 (ML = 21.9) and 11.1.01.7 (ML = 24.3), a rugose ridge develops across the proximal margin of the fossa and truncates the apex of the triangle. In these specimens, the suboval fossa has a long axis angled distolaterally to proximomedially.
In the Two Medicine Formation specimens, the anterior aspect of metatarsal III at its broadest part is flat to slightly convex. van der Reest and Currie (Reference van der Reest and Currie2017) described a concave face as diagnostic for “L. mcmasterae”, but Cullen et al. (Reference Cullen, Zanno, Larson, Todd, Currie and Evans2021) thought this was a variable feature among North American troodontids. In most troodontids where known and in TMP 1992.036.0575 (van der Reest and Currie, Reference van der Reest and Currie2017, fig. 13b), the shaft of metatarsal III is exposed posteriorly over nearly its entire length, except where metatarsal IV exhibits a distinct medial kink just proximal to the distal expansion of metatarsal III (Zanno et al., Reference Zanno, Varricchio, O’Connor, Titus and Knell2011). In contrast, in MOR 748, the distal shaft remains obscured over some 40% of its length.
Remarks
Among the phylogenetic results, our parsimony analysis including “Latenivenatrix mcmasterae”, CMN 8539, and our revised Troodon formosus scores yielded discordant results (Figs. 7.1, S6–S9). The topology of Troodontinae differs from the one inferred from van der Reest and Currie (Reference van der Reest and Currie2017). The strict consensus tree of the 12 most parsimonious trees placed T. formosus in a clade with CMN 8539, Talos, Linhevenator, and Philovenator, rather than with Saurornithoides and Zanabazar (Figs. 7.1, S6). However, this relationship was not recovered after bootstrapping (Fig. S8). Instead, most of the Troodontinae fell into a polytomy, but T. formosus grouped with CMN 8539 with low bootstrap support (bs = 20), consistent with the synonymy of T. formosus and “Stenonychosaurus inequalis.” The new reconstruction only persisted for an additional step before the clade collapsed (Bremer = 1; Fig. S9). The strict consensus trees also plotted “L. mcmasterae” as a sister taxon to Gobivenator, which both formed a clade with Saurornithoides and Zanabazar (Figs. 7.1, S6). Although, “L. mcmasterae” formed a clade with Linhevenator and Philovenator after bootstrapping but with low support (bs = 8; Fig. S8). The new clade persisted only for one additional step under a Bremer analysis (Bremer = 1; Fig. S9). Unlike previous reconstructions, we also found that Urbacodon plots rootward to Sinornithoides and MPC-D 100/44, an unnamed troodontid (Figs. S6, S7, S9; Bremer = 1). However, this relationship did not hold after bootstrapping (Fig. S8; bs = 2). We found consistent results after excluding CMN 8539 (Figs. S11–S14).
Figure 7. Phylogenetic results from the analysis of the van der Reest and Currie (Reference van der Reest and Currie2017) character matrix with revised Troodon formosus scoring. (1) Troodontidae portion of the strict consensus tree from the parsimony analysis with Troodon formosus in bold font. Node labels represent the Bremer support values. Bootstrap support values were not included due to topological differences. (2) Troodontidae portion of the majority rules consensus tree from the Bayesian analysis with Troodon formosus in bold font. Node labels represent the percent posterior probability. Scale bar measures the amount of character evolution. CMN 8539 is the type of Stenonychosaurus inequalis.
Our Bayesian analysis yielded low resolution across most of Troodontidae (Figs. 7.2, S10). Troodontidae showed low node support with a 67% posterior probability (pp). Although the support for Troodontinae increased (pp = 82%) compared to that inferred from the original matrix (pp = 56%; Fig S5), most taxa within Troodontinae fell into a polytomy following the rescoring of T. formosus. CMN 8539 grouped with T. formosus with moderate support (pp = 85%), consistent with the parsimony results. Notably, the clade uniting Saurornithoides, T. formosus, and Zanabazar supported by the original matrix (pp = 85%) fell into a polytomy with Gobivenator, “L. mcmasterae,” Talos, and Urbacodon. Analyses excluding CMN 8539 yielded concordant results (Fig. S15).
To test for the effect of including “L. mcmasterae” on our phylogenetic results, we removed it in a subsequent analysis along with the three characters established by van der Reest and Currie (Reference van der Reest and Currie2017). In the parsimony analysis excluding “L. mcmasterae,” T. formosus fell into a polytomy with Talos and Urbacodon (Figs. S16, S17, S19; Bremer = 1). Troodon formosus formed a clade with Linhevenator and Philovenator after bootstrapping but with low support (Fig. S18; bs = 1). Removing “L. mcmasterae” did not affect the topology in the Bayesian analysis. Most relationships within Troodontinae remained unresolved (Fig. S20), consistent with our analysis including “L. mcmasterae.” The posterior node support for Troodontinae increased after removing “L. mcmasterae” with moderately low support (pp = 80%). Overall, these results highlight the poor resolution within Troodontinae.
Discussion
Phylogenetic analysis consistently placed Troodon formosus in a clade that includes Talos, Linhevenator, Philovenator, Saurornithoides mongoliensis, Zanabazar, Gobivenator, and “Latenivenatrix mcmasterae,” the Troodontinae of van der Reest and Currie (Reference van der Reest and Currie2017). Our osteological observations bear this out. Troodon formosus shares with several of these taxa a roughly triangular frontal with a prominent lateral extension of the postorbital process, a sharp demarcation of the anterior limit of the supratemporal fenestra, and a parasagittal trough near the midline. Other characters in common are an expanded iliac articulation on the pubis and the highly modified condition of metatarsal III. Both parsimony and Bayesian analyses found a close relationship between T. formosus and CMN 8539, the holotype specimen of “Stenonychosaurus inequalis,” consistent with their synonymy. Nevertheless, despite the additional scoring of more than 100 characters for T. formosus, the relationships among the Troodontinae remain fairly ambiguous. Our parsimony topology, grouping T. formosus, CMN 8539, and Talos close to Philovenator and Linhevenator (Fig. 7.1), departs from that of van der Reest and Currie (Reference van der Reest and Currie2017), which supported a relationship among T. formosus, S. mongoliensis, and Zanabazar. An analysis by Pei et al. (Reference Pei, Pittman, Goloboff, Dececchi, Habib, Kaye, Larsson, Norell, Brusatte and Xu2020) similarly recovered a close relationship among T. formosus, S. mongoliensis, and Zanabazar but with relatively low group support. Another recent study placed T. formosus as a sister taxon to Linhevenator, with both closer to S. mongoliensis and Zanabazar than Philovenator (Wang et al., Reference Wang, Zhang, Tan, Jiangzuo, Zhang and Tan2022). Overall, the revised scoring for T. formosus has reduced our confidence in the general relationships within Troodontinae and suggests caution is warranted when formulating biogeographic and evolutionary hypotheses from these trees.
Three recent studies have examined whether troodontid teeth of the Late Cretaceous of Montana and Alberta are taxonomically diagnostic. These studies used a sample of seven, eight, and ten teeth as representative of the “Judith River Formation” but all included six teeth from MOR 553, Jack’s Birthday Site in the Two Medicine Formation of Montana. The remaining (one, two, or four) teeth came from the Judith River Formation with all but Larson and Currie (Reference Larson and Currie2013) including the Troodon formosus holotype, ANSP 9259. Results varied. Both Larson and Currie (Reference Larson and Currie2013) and Evans et al. (Reference Evans, Cullen, Larson and Rego2017) considered the “Judith River” sample as broadly overlapping that from Dinosaur Park Formation with both samples representing the same quantitative morphotype, T. formosus (Larson and Currie, Reference Larson and Currie2013). Two studies (Larson and Currie, Reference Larson and Currie2013; Larson et al., Reference Larson, Brown and Evans2016) found these combined Campanian teeth distinct from those of the younger Horseshoe Canyon Formation, whereas two (Torices et al., Reference Torices, Funston, Kraichy and Currie2014; Evans et al., Reference Evans, Cullen, Larson and Rego2017) did not.
Given the great heterogeneity within some troodontid individuals (e.g., Zanabazar junior) (Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009), it is not surprising that the utility of separating troodontid taxa by teeth alone may be challenging or ambiguous. Given this potential ambiguity, Evans et al. (Reference Evans, Cullen, Larson and Rego2017, p. 824) suggested that “T. formosus may be a nomen dubium.”
The Two Medicine troodontid material described here, in expanding the hypodigm, helps to clarify and reconfirm the species Troodon formosus as originally established by Leidy (Reference Leidy1856) and later revised in part by Currie (Reference Currie1987) and more recently under the name “Stenonychosaurus inequalis” by van der Reest and Currie (Reference van der Reest and Currie2017). van der Reest and Currie (Reference van der Reest and Currie2017, p. 933) wrote that the “similarities between frontals and metatarsals of “S. inequalis” from the Dinosaur Provincial Park Formation and those of the Two Medicine troodontid suggest that the two taxa may in fact be the same species.” This observation is confirmed by both our osteological descriptions and the recent tooth comparisons among the relevant formations (Larson and Currie, Reference Larson and Currie2013; Larson et al., Reference Larson, Brown and Evans2016; Evans et al., Reference Evans, Cullen, Larson and Rego2017). Below we outline past issues with Troodon formosus and our reasoning for retaining this species name over its synonym, Stenonychosaurus inequalis.
Due to the confusion surrounding T. formosus, it is prudent to first clearly state that we do not think that all the specimens previously assigned to T. formosus are necessarily correct in their assignment (Table 1). Instead, we argue that the name T. formosus is valid as outlined by the International Code of Zoological Nomenclature (ICZN, 1999) and that this name is linked to a legitimate species. Troodon formosus has been synonymized with “S. inequalis,” as per Article 61.3 (ICZN, 1999), by several authors (Russell, Reference Russell1969; Carpenter, Reference Carpenter1982), although most thoroughly by Currie (Reference Currie1987). Currie (Reference Currie, Currie and Koppelhus2005, p. 375) said, “Whereas most tooth genera from that period have proven to be nomina dubia, the tooth of this small theropod was distinctive enough to eventually lead to the identification of the rest of the skeleton.” Despite an initial description from a single tooth, the hypodigm, as outlined above, has since grown to include substantially more material. The current interpretation of the T. formosus species is bolstered by the synonymy (with “Stenonychosaurus inequalis”) set forth by Currie (Reference Currie1987), but exclusive of material from Pectinodon bakkeri Carpenter, Reference Carpenter1982. Additionally, the status of “Latenivenatrix mcmasterae” van der Reest and Currie, Reference van der Reest and Currie2017, remains in question (Cullen et al., Reference Cullen, Zanno, Larson, Todd, Currie and Evans2021). It is plausible that further material could be removed from the species concept T. formosus and incorporated into other troodontids, in agreement with Fiorillo et al. (Reference Fiorillo, Tykoski, Currie, McCarthy and Flaig2009), Zanno et al. (Reference Zanno, Varricchio, O’Connor, Titus and Knell2011), Evans et al. (Reference Evans, Cullen, Larson and Rego2017), and van der Reest and Currie (Reference van der Reest and Currie2017).
On the validity of the name Troodon formosus
Over the last 15 years, there has been some discontent regarding the taxon Troodon formosus. These issues have largely stemmed from the holotype specimen, ANSP 9259, being a single, potentially undiagnostic tooth.
The name Troodon formosus is based on a tooth
The fact that T. formosus is attached to a tooth holotype is the most common argument against the validity of the name (Fiorillo et al., Reference Fiorillo, Tykoski, Currie, McCarthy and Flaig2009; Zanno et al., Reference Zanno, Varricchio, O’Connor, Titus and Knell2011; Larson and Currie, Reference Larson and Currie2013; Torices et al., Reference Torices, Funston, Kraichy and Currie2014; Evans et al., Reference Evans, Cullen, Larson and Rego2017; van der Reest and Currie, Reference van der Reest and Currie2017). There has been a large push to revise tooth taxa across Dinosauria because teeth are often seen as inadequate type specimens. The tooth-based argument against the name T. formosus can be broken down into two main concerns. Either (1) there is an inherent problem with naming an organism based on its dentition, or (2) any tooth specimen itself is invalid because it is non-diagnostic.
The claim that teeth are invalid types is sometimes presented as an a-priori assumption. Fiorillo (Reference Fiorillo2008, p. 323) stated, “Taxonomic distinctions within the Dinosauria are based on the skeleton rather than on dentition,” and Zanno et al. (Reference Zanno, Varricchio, O’Connor, Titus and Knell2011, p. 2) stated, “Troodon formosus, since based on a single tooth, may represent a nomen dubium.” However, Article 17.3 of the Code (ICZN, 1999, p. 21) explicitly states that an animal name is valid even if “it is based on only part of an animal.” Article 72.5 further reiterates this with the specific statement that a type specimen may be made up of “an animal, or any part of an animal” (ICZN, 1999, p. 77). The allowances in the Code make it clear that there are no intrinsic issues with naming an animal based on a tooth or any other fragment of its anatomy. This does not mean that tooth taxa ought to be exempt from scrutiny. On the contrary, suspicious taxa originally based on meager material, such as T. formosus, must be investigated thoroughly. Critics of T. formosus are correct in their assertions of the dangers posed by illegitimate taxa. Such taxa can confuse cladistics, obscure ecology, and otherwise interfere with our understanding of ancient life.
The type of T. formosus is undiagnostic
Even though dinosaur teeth are specifically allowed as type specimens, it is often argued that they are inappropriate choices due to their lack of diagnostic features. Types are defined as, “standards of reference that provide objectivity in zoological nomenclature…” (ICZN, 1999, p. 79). Specifically, a holotype is, “The single specimen designated or otherwise fixed as the name-bearing type of a nominal species or subspecies when the nominal taxon is established” (ICZN, 1999, p. 120). A type specimen is meant to be a name-bearer. The Code makes it clear that it cannot rule on matters of biology, and, as such, a type specimen cannot define a species concept—diagnostic power belongs to researchers, as guaranteed by the Code.
When described by Leidy (Reference Leidy1856), the holotype tooth was certainly diagnostic because it clearly could not be ascribed to any then-known organism. This alone does not guarantee modern validity, but the name must be considered because it met all requirements when erected. In a practical sense, type specimens in paleontology often become less diagnostic over time, as once diagnostic features are found to be more widely distributed and become “obsolescent characters” (Wilson and Upchurch, Reference Wilson and Upchurch2003). Nevertheless, additional specimens and observations can expand the hypodigm while also adding features that cannot be inferred from the type alone.
The name Troodon formosus is a nomen dubium
Although Troodon formosus remains in common use (e.g., Holtz et al., Reference Holtz, Brinkman and Chandler2000; Horner et al., Reference Horner, Schmitt, Jackson and Hanna2001a, Reference Horner, Padian and de Ricqlèsb; Sankey et al. Reference Sankey, Brinkman, Guenther and Currie2002; Fiorillo et al., Reference Fiorillo, Tykoski, Currie, McCarthy and Flaig2009; Norell et al., Reference Norell, Makovicky, Bever, Balanoff, Clark, Barsbold and Rowe2009; Turner et al., Reference Turner, Makovicky and Norell2012; Larson and Currie, Reference Larson and Currie2013; Torices et al., Reference Torices, Funston, Kraichy and Currie2014; Tsuihiji et al., Reference Tsuihiji, Barsbold, Watabe, Tsogtbaatar, Chinzorig, Fujiyama and Suzuki2014; Sellés et al., Reference Sellés, Vila, Brusatte, Currie and Galobart2021), it also has been deemed a nomen dubium by several authors (Zanno et al., Reference Zanno, Varricchio, O’Connor, Titus and Knell2011; Evans et al., Reference Evans, Cullen, Larson and Rego2017; van der Reest and Currie, Reference van der Reest and Currie2017). Currie (Reference Currie, Currie and Koppelhus2005) raised doubts even earlier, but did not specifically refer to T. formosus as a nomen dubium, instead mentioning that other tooth taxa have proven to be doubtful while T. formosus has been successfully associated with additional skeletal material. Still, Zanno et al. (Reference Zanno, Varricchio, O’Connor, Titus and Knell2011) argued that T. formosus is a nomen dubium specifically because the type specimen is a tooth, a sentiment that van der Reest and Currie (Reference van der Reest and Currie2017) echo.
Nomen dubium is a term often invoked by paleontologists, but it is a subjective declaration of doubt that always invites further discussion. Nothing can truly “qualify” as a nomen dubium because the Code (ICZN, 1999) does not list any objective qualifications. Instead, it is defined as, “a name of unknown or doubtful application” (ICZN, 1999, p 111). As pointed out by van der Reest and Currie (Reference van der Reest and Currie2017), the term is mentioned in Article 75.5. This is the only place, other than the glossary, where it appears. Nomen dubium is meant only to indicate an author’s skepticism about the validity of a given name, it has no power in and of itself.
Given the above issues, several researchers (Currie Reference Currie, Currie and Koppelhus2005; Evans et al., Reference Evans, Cullen, Larson and Rego2017; van der Reest and Currie, Reference van der Reest and Currie2017) have advocated abandoning T. formosus and instead recognizing “S. inequalis” as the better taxon for this Judithian troodontid of Montana and Alberta. However, there has been no recent discussion regarding the viability of the “S. inequalis” holotype, CMN 8539 (Sternberg, Reference Sternberg1932), despite the advocation for abandonment of T. formosus and its type. CMN 8539 was originally described by Sternberg (Reference Sternberg1932, p. 102) as having been “completely exposed and badly weathered.” The damage is to such a degree that, as pointed out by Russell (Reference Russell1969), metatarsal III was frequently misinterpreted as, “the longest bone of the foot not reduced in its proximal portion but is of about the same dimensions as Mts. II and IV (Sternberg, Reference Sternberg1932, p. 105).” Metatarsal III morphology is a key diagnostic skeletal feature encompassed by CMN 8539, and the specimen is of such a quality that this detail was originally misinterpreted. Additionally, when discussing “S. inequalis,” van der Reest and Currie (Reference van der Reest and Currie2017, p. 932) said, “A subtriangular fossa and a slightly convex anterior face of metatarsal III are also observed in Talos sampsoni…”. This further raises the question of how diagnostic CMN 8539 is because this is the only feature observable in the type specimen of “S. inequalis.” Cullen et al. (Reference Cullen, Zanno, Larson, Todd, Currie and Evans2021) concurred that this feature is not diagnostic. Although it may encompass more elements, CMN 8539 offers no more diagnostic power than ANSP 9259.
On the validity of the species Troodon formosus
Some grievances raised about T. formosus concern perceptions about the biological species rather than the name itself. Such issues are beyond the scope of the ICZN, which only rules on taxonomy. These doubts come from a number of authors, but can be summarized as (1) the T. formosus species concept is based on a tooth, (2) the stratigraphic range of T. formosus is too broad, (3) the geographic range of T. formosus is too broad, and (4) the T. formosus hypodigm may incorporate more than one taxon (Currie, Reference Currie, Currie and Koppelhus2005; Fiorillo, Reference Fiorillo2008; Fiorillo et al., Reference Fiorillo, Tykoski, Currie, McCarthy and Flaig2009; Zanno et al., Reference Zanno, Varricchio, O’Connor, Titus and Knell2011; Larson and Currie, Reference Larson and Currie2013; Torices et al., Reference Torices, Funston, Kraichy and Currie2014; Evans et al., Reference Evans, Cullen, Larson and Rego2017; van der Reest and Currie, Reference van der Reest and Currie2017).
The T. formosus species concept is based on a tooth
Several studies have sought to differentiate and organize North American troodontids based on tooth morphotype (Larson and Currie, Reference Larson and Currie2013; Torices et al., Reference Torices, Funston, Kraichy and Currie2014; Evans et al., Reference Evans, Cullen, Larson and Rego2017). Larson and Currie’s (Reference Larson and Currie2013) original analysis found that teeth from coeval formations generally could not be differentiated. Their data supported the claim that troodontid teeth from the upper Two Medicine (MOR 553, mislabeled as Judith River Formation) and Dinosaur Park formations were of the same T. formosus morphotype and distinct from other forms, such as those from the Horseshoe Canyon Formation. However, Torices et al. (Reference Torices, Funston, Kraichy and Currie2014) did not recover the same morphotypes and instead argued that troodontid teeth found in stratigraphically distinct units from Alberta are undifferentiable. Notably, their study excludes material from the Two Medicine or Judith River formations. Evans et al. (Reference Evans, Cullen, Larson and Rego2017) also concluded that troodontid tooth morphotypes are difficult to distinguish given current methods and material. Their re-analysis corroborated Larson and Currie (Reference Larson and Currie2013), providing support for the inclusion of Two Medicine Formation material (MOR 553, still mislabeled as Judith River) in the Dinosaur Park Formation morphotype.
Each of these studies allowed the possibility that sampled troodontid teeth simply are not diagnostic between species. However, Torices et al. (Reference Torices, Funston, Kraichy and Currie2014) also mentioned that it is possible the teeth in their study came from the same species. That Larson and Currie (Reference Larson and Currie2013) and Evans et al. (Reference Evans, Cullen, Larson and Rego2017) found that teeth from the upper Two Medicine and Dinosaur Park formations matched could alternatively be interpreted as further evidence supporting the T. formosus/S. inequalis synonymy. However, Evans et al. (Reference Evans, Cullen, Larson and Rego2017) interpreted this overlap as evidence that T. formosus is a nomen dubium. van der Reest and Currie (Reference van der Reest and Currie2017) echoed this sentiment. They noted that the teeth of “L. mcmasterae” cannot be distinguished from those of S. inequalis, however, no tooth-bearing elements occur in associated “L. mcmasterae” specimens. The lack of tooth-bearing specimens for North American troodontids will likely exacerbate isolated tooth assignments, which are already complicated by heterodonty and ontogeny (Currie, Reference Currie1987; Larson and Currie, Reference Larson and Currie2013; Torices et al., Reference Torices, Funston, Kraichy and Currie2014; Evans et al., Reference Evans, Cullen, Larson and Rego2017). However, the effect of these tooth-related complications could be assuaged through neotype designations.
The stratigraphic range of Troodon formosus is too broad
Zanno et al. (Reference Zanno, Varricchio, O’Connor, Titus and Knell2011) mentioned that T. formosus material has been described from numerous formations potentially covering 20 Myr, but this is somewhat misleading. The proposed time frame is drastically reduced if the upper Two Medicine is used instead of the entire formation. Additionally, the vast majority of T. formosus material comes from the Dinosaur Park and upper Two Medicine formations. These formations are coeval (Fowler, Reference Fowler2017; Ramezani et al., Reference Ramezani, Beveridgem, Rogers, Eberth and Roberts2022), their division more political than geological. Also, the upper Judith River Formation approximately matches this same timeframe (Fowler, Reference Fowler2017; Ramezani et al., Reference Ramezani, Beveridgem, Rogers, Eberth and Roberts2022) (Fig. 1). It is extremely unlikely that a single species existed across a 20-million-year period, but that is not the current claim.
The geographic range of Troodon formosus is too broad
The first concerns over the geographic range of T. formosus were perhaps expressed by Currie (Reference Currie, Currie and Koppelhus2005) when opting to use the new combination Troodon inequalis. Currie (Reference Currie, Currie and Koppelhus2005) suggested that T. inequalis is more conservative and uses it to describe troodontid material from the Dinosaur Park Formation exclusive of the Two Medicine Formation. Nevertheless, Currie (Reference Currie, Currie and Koppelhus2005) did not describe any morphological differences between T. formosus and T. inequalis, indeed the material was still considered synonymous with “S. inequalis.” According to the Code (ICZN, 1999, Art. 61.3) these three names are synonyms, and it is most correct and conservative to continue to use T. formosus over either junior name.
Fiorillo (Reference Fiorillo2008) and Fiorillo et al. (Reference Fiorillo, Tykoski, Currie, McCarthy and Flaig2009) described troodontid material from the Prince Creek Formation of the North Slope, Alaska. Although concerned that the material might extend beyond T. formosus, the authors nonetheless assigned the specimens to T. formosus since the new material could not be differentiated morphologically. This is the conservative choice, and any of the specimens can simply be removed from the T. formosus concept if shown to fit better elsewhere. The Prince Creek Formation material greatly expanded the known geographic range of T. formosus, and Zanno et al. (Reference Zanno, Varricchio, O’Connor, Titus and Knell2011) expressed concern that the geographic range had become too widespread. Zanno et al. (Reference Zanno, Varricchio, O’Connor, Titus and Knell2011) indicated a geographic range of 4000 km for T. formosus; however, the ranges of some modern terrestrial macrovertebrates (Ursus americanus Pallas, Reference Pallas1780, 5500 km; Canis latrans Say in James, Reference James, Long, Say and Adams1823, 7000 km; Odocoileus virginianus [Zimmermann, Reference Zimmermann1780], 8000 km; Puma concolor [Linnaeus, Reference Linnaeus1771], 11,000 km) show that the proposed 4000-km range for T. formosus is not unrealistic (ranges estimated with data from IUCN, 2020).
The T. formosus hypodigm may incorporate more than one taxon
The concern that T. formosus may contain more than one taxon is central to the preceding three arguments and is mentioned by numerous authors (Fiorillo, Reference Fiorillo2008; Fiorillo et al., Reference Fiorillo, Tykoski, Currie, McCarthy and Flaig2009; Zanno et al., Reference Zanno, Varricchio, O’Connor, Titus and Knell2011; Larson and Currie, Reference Larson and Currie2013; Torices et al., Reference Torices, Funston, Kraichy and Currie2014; Evans et al., Reference Evans, Cullen, Larson and Rego2017; van der Reest and Currie, Reference van der Reest and Currie2017). As previously stated, and in agreement with the above authors, it is quite possible that there are additional troodontids represented by the material that has been assigned to T. formosus. Despite the possibility of these undescribed troodontids, it is correct to continue to use T. formosus. As conducted by Fiorillo (Reference Fiorillo2008) and Fiorillo et al. (Reference Fiorillo, Tykoski, Currie, McCarthy and Flaig2009), material that is morphologically undifferentiable from T. formosus should be placed in T. formosus. Specimens should be reassigned only when the necessary evidence has been described. For example, van der Reest and Currie (Reference van der Reest and Currie2017) argued that some frontals of T. formosus better fit into “L. mcmasterae.” Cullen et al. (Reference Cullen, Zanno, Larson, Todd, Currie and Evans2021) countered that this simply reflects morphologic variation within one species. Resolving this will only make the T. formosus concept more precise. Revision of species concepts is the natural course of biology. A modern case can be seen in the recent extraction of two additional glider taxa (Petauroides minor [Collett, Reference Collett1887] and Petauroides armillatus Thomas, Reference Thomas1923) from the species Petauroides volans (Kerr, Reference Kerr1792) (McGregor et al., Reference McGregor, Padovan, Georges, Krockenberger, Yoon and Youngentob2020). The discovery that P. minor and P. armillatus had previously been included in P. volans has no effect on the validity of P. volans (name or concept), and the species concept for P. volans has accordingly become more precise with the distinction.
Moving forward with Troodon formosus
Below are three courses of action that could be taken apropos T. formosus:
Make no taxonomic changes regarding Troodon formosus: maintain current type and continue prevailing usage
Although it can be argued that prevailing usage of T. formosus need not change (name is valid according to the Code, species concept is coherent and adheres to universality), discussion of the taxon has become problematic. Taking no action regarding T. formosus would be dissatisfactory to a number of parties, and discussion surrounding T. formosus would continue to suffer.
Abandon Troodon formosus: allow it to lapse into taxonomic obsolescence
As proposed by van der Reest and Currie (Reference van der Reest and Currie2017), the taxon T. formosus could be abandoned. The authors indicated that this would reestablish the taxon S. inequalis, but that outcome is neither guaranteed nor simple in its implementation. Would Two Medicine Formation material then be assigned to S. inequalis? If so, then there has been no change in the morphology of the species concept (i.e., the concept that was associated with T. formosus would just be transferred to S. inequalis). Essentially, a junior synonym would supplant a senior synonym. Furthermore, given the damaged and nonspecific nature of the type CMN 8539, this action would leave S. inequalis beleaguered by the same weaknesses as its predecessor.
Alternatively, would S. inequalis material be extracted from what had been T. formosus, and the Two Medicine Formation material consolidated into a new, unnamed taxon? This is similarly problematic since no morphological distinctions have been identified between the two taxa; they would be synonymous from the start. Each of these scenarios would threaten stability and universality, thus conflicting with the Code’s objective.
Institute a neotype for Troodon formosus: conserve prevailing usage by defining a new type specimen
The central concern harrying T. formosus is that the species concept has aggregated around a type tooth considered to possess questionably diagnostic morphology. According to the conditions supplied by the Code, the name T. formosus is valid. Additionally, there is general agreement concerning the taxon’s biology. Importantly, even if the type specimen is undiagnostic, there is no convincing evidence for a second skeletal taxon from these geologic units of Alberta and Montana that would confound the synonymy of Troodon formosus and Stenonychosaurus inequalis. Rather, a more illuminative and diagnostic type specimen would anchor prevailing usage and grant needed stability to T. formosus, thus preserving a popular and long held dinosaur name. However, the institution of a neotype over an existing holotype can only be enacted by the Commission through its plenary power (ICZN, 1999, Art. 75.5, 81).
We judge that the option of formally instituting a neotype for T. formosus is the appropriate course of action. It is most in line with the goals of the Code, and a new type specimen would alleviate the majority of concerns regarding the taxon. To this end, a petition to the Commission is being drafted (Hogan et al., in review). We would propose that MOR 553, the material from Jack’s Birthday Site and including the material described here as well as the teeth used in multiple tooth studies would make the best neotype. Although van der Reest and Currie (Reference van der Reest and Currie2017, p. 934) argued that, “For any specimens that are positively identified as Troodon formosus, however, they must originate from the Judith River Formation in the region from where the holotype was recovered,” these geologic units (Judith River, Two Medicine, Dinosaur Park) are contemporary and confluent. Establishing a neotype will also facilitate further taxonomic comparisons between the Two Medicine Formation troodontid material and those of other geologic formations (e.g., “Stenonychosaurus inequalis” and “Latenivenatrix mcmasterae” material from the Dinosaur Park Formation). Shifting the type locality from a vague position within the Judith River to a known site in the Two Medicine Formation would also provide more context moving forward.
Conclusions
Analysis of troodontid material from the upper Two Medicine Formation of Montana indicates that it is conspecific with “S. inequalis” of the Dinosaur Park Formation of Alberta, as van der Reest and Currie (Reference van der Reest and Currie2017) suspected. We recognize T. formosus as the senior synonym for this taxon, which is distinguished by a maxilla with a larger, more broadly rounded maxillary fenestra, 23 teeth, and a large palatal shelf; more pronounced basioccipital tubera; an L-shaped to triangular frontal; and a relatively shorter metatarsal III with a convex to flat anterior face at its maximum breadth. The consistency of the Two Medicine material with “S. inequalis” lends some support for the validity of “L. mcmasterae” (van der Reest and Currie, Reference van der Reest and Currie2017); however, the validity of the latter will require discovery of more complete specimens where hypothesized apomorphies can be recognized together.
Phylogenetic analysis places T. formosus within the Troodontinae; however, within-group resolution remains ambiguous. Our parsimony and Bayesian analyses, respectively, place T. formosus (and CMN 8539, the holotype of “S. inequalis”) in a clade consisting of Talos, Philovenator, and Linhevenator or, alternatively, in a polytomy with the rest of Troodontinae. The troodontid phylogeny has been regarded as one of the least resolved paravian groups (Pittman et al., Reference Pittman, O’Connor, Field, Turner, Ma, Makovicky and Xu2020), and the past decade has seen several variations of our two tree topologies (see Wang et al., Reference Wang, Zhang, Tan, Jiangzuo, Zhang and Tan2022, fig. 8). Our recoding of T. formosus supports Troodontinae as a group but, unfortunately, does not clarify the relationships within it.
The name Troodon formosus has not been shown, through reasoning based in the Code (ICZN, 1999) to be invalid, inappropriate, or otherwise incorrect. The T. formosus species concept has become robust with the synonymization of “S. inequalis” (Currie, Reference Currie1987), consequentially T. formosus is now known from substantially more material than a lone tooth. Although concerns have been expressed about the T. formosus species concept, none carries enough weight to legitimize abandonment.
We propose that T. formosus would benefit from the introduction of a neotype, namely MOR 553. This Two Medicine material, described in part here, would greatly expand the species hypodigm and would promote stability, a key tenant of the Code, by conserving prevailing usage while ameliorating most concerns voiced over the last 15 years. However, since the T. formosus type has not been lost or destroyed, only the International Commission on Zoological Nomenclature has the power to implement a neotype (Art. 75.5). Accordingly, a petition is being drafted for the Commission to review. Troodon is a well-known dinosaur taxon in both the academic and public spheres. The name has been in effect for over 150 years and is the oldest North American dinosaur name still in use. Additionally, many important specimens have been assigned to this group—material that has been integral to our understanding of theropod reproduction and the evolution of avian reproduction.
Acknowledgments
We would like to thank J. Horner for the opportunity to study and collect at Jack’s Birthday Site long ago, D. Strosnider and H. Olson for assistance in generating bone figures, J. Wilson for discussion, and M. Ivie for his assistance with all things taxonomical. We also thank the Willi Hennig Society for sponsoring the free use of the program TNT.
Competing interests
The authors declare none.
Data availability statement
All data and code used in this study are available on Zenodo (https://doi.org/10.5281/zenodo.13305737).
Open access
