The fossils studied here come from the Están de Colón (RD-34) site, which is in the municipality of Riodeva (province of Teruel, Aragón, Spain) (Fig. 1A, B).

Figure 1. 

A, B location and geological setting of the Están de Colón (RD-34) fossil site in Riodeva (Teruel, Spain). C stratigraphic section of the Villar del Arzobispo Formation in the Riodeva area. Cartography and stratigraphic section modified from Campos-Soto et al. (2019).

Geologically, RD-34 is included in the South-Iberian Basin (Fig. 1A, B). The South-Iberian Basin presents a NW–SE orientation and is part of the Mesozoic Iberian Extensional System (e.g., Campos-Soto et al. 2021). This basin was developed in eastern Iberia during the late Oxfordian–middle Albian and inverted during the Cenozoic Alpine Orogeny (e.g., Mas et al. 2004; Campos-Soto et al. 2019, 2021).

RD-34 is located in the upper half of the Villar del Arzobispo Formation (upper Kimmeridgian–Tithonian sensu Campos-Soto et al. 2017, 2019) section in the Riodeva area (Campos-Soto et al. 2019) (Fig. 1C). It is a detrital-carbonate lithostratigraphic unit formed by sandstone and clay levels with intercalations of limestone and marls (Mas et al. 1984). This unit comprises two informal parts (Campos-Soto et al. 2019): (1) an essentially carbonate lower part (CLP, upper Kimmeridgian) and an essentially siliciclastic upper part (SUP, upper Kimmeridgian–Tithonian). Deposits of the SUP have been interpreted as deposited in a coastal and alluvial plain (Campos-Soto et al. 2019, 2021). Concretely, the fossils studied here were located in a sandstone layer.

The Villar del Arzobispo Formation overlies the oncolitic limestone unit Higueruelas Formation (Kimmeridgian sensu Campos-Soto et al. 2016; Pacios et al. 2018).

Están de Colón (RD-34) fossil site was discovered in 2004 and partially excavated in 2006 and 2007. This site was first mentioned at a conference in 2008 (Cobos et al. 2008). RD-34 was found in a crop field with an area of approximately 24 m². During the two excavation campaigns, around 200 elements were recovered. The freshwater bivalve Margaritifera cf. valdensis Mantell, 1844 was reported at this site (Delvene et al. 2013). The preliminary study of this huge fossil site reveals that most of these fossils correspond to at least two stegosaurian specimens at different ontogenetic stages. In addition to the stegosaurs, RD-34 also yielded several theropod, sauropod, ornithopod, crocodylomorph, and osteichthyan fossils. The fossils studied here comprise a partial cranium (MAP-9029) and a mid cervical vertebra (MAP-9030). MAP-9029 and MAP-9030 are considered to belong to the same specimen because they came from the same layer, they were found associated (distance between them around 5 cm), and the size and features of the pieces are consistent. Additional postcranial fossils of this specimen were recovered; however, their systematic study is beyond the scope of this paper and some of them are still unprepared.

The studied fossils are deposited in the Museo Aragonés de Paleontología (Teruel, Spain).

The requirements of the International Code of Phylogenetic Nomenclature (PhyloCode) were met and its recommendations followed to formally establish the clade names. For each clade the following information was included: clade name, designation (new or converted), registration number, phylogenetic definition, reference phylogeny, hypothesized composition, and comments. Furthermore, an etymology section was also added in the case of new clade names. All clade names were registered in the Regnum repository (www.phyloregnum.org).

The data matrices (Files S2 and S3) were built using the MESQUITE v.3.81 software (Maddison and Maddison 2023).

The character list was primarily based on the stegosaurian-focused analyses by Sereno and Dong (1992), Galton and Upchurch (2004), Maidment et al. (2008), and Raven and Maidment (2017), with additional input from broader studies that focus on ornithischians, thyreophorans, and ankylosaurs (Sereno 1999; Butler et al. 2008; Soto-Acuña et al. 2021; Raven et al. 2023). These datasets were thoroughly evaluated, resulting in the exclusion or modification of several characters as required (File S1 [1.4]). Furthermore, 20 new characters were added (File S1 [1.4]). The final character list contains 115 characters: 40 cranial, 27 axial, 38 appendicular, and 10 osteodermal (Files S1 [1.4], S2, S3). Character scorings were based on a strict bibliographic revision, photographs, 3D models, and, when possible, on first-hand observations (Table 1; File S1 [1.3]).

The unarmoured taxon Lesothosaurus diagnosticus Galton, 1978 was used as the outgroup. It is known from several specimens, some of which are highly complete, and was recovered as an early-diverging thyreophoran in a few analysis (Butler et al. 2008; Boyd 2015). However, these results are poorly supported by evidence and L. diagnosticus has recently been recovered as an early-diverging ornithischian, genasaurian, or neornithischian (e.g., Baron et al. 2017; Han et al. 2018; Dieudonné et al. 2021; Fonseca et al. 2024). Following Baron et al. (2017), ‘Stormbergia dangershoeki’ Butler, 2005 is regarded as a subjective junior synonym of L. diagnosticus, and consequently, the latter taxon was coded (Thulborn 1972; Galton 1978; Sereno 1991; Butler 2005; Porro et al. 2015; Baron et al. 2017; pers. obs.). Four early-diverging thyreophoran taxa were included: Scutellosaurus lawleri Colbert, 1981 (Colbert 1981; Rosenbaum and Padian 2000; Breeden and Rowe 2020; Breeden et al. 2021), Emausaurus ernsti Haubold, 1990 (Haubold 1990; Norman et al. 2004), Yuxisaurus kopchicki Yao et al., 2022 (Yao et al. 2022), and Scelidosaurus harrisonii Owen, 1861 (Norman 2020a, 2020b, 2020c, 2021; pers. obs.), as well as three representative ankylosaurian taxa: the nodosaurid (struthiosaurid sensu Raven et al. 2023) Europelta carbonensis Kirkland et al., 2013 (Kirkland et al. 2013; pers. obs.) and the ankylosaurids Euoplocephalus tutus Lambe, 1902 (Vickaryous and Russell (2003); Arbour and Currie 2013) and Ankylosaurus magniventris Brown, 1908 (Carpenter 2004; Arbour and Mallon 2017). Regarding stegosaurs, non-valid or dubious taxa (e.g., ‘Craterosaurus pottonensis’ Seeley, 1874, ‘Chialingosaurus kuani’ Young, 1959, ‘Dravidosaurus blanfordi’ Yadagiri & Ayyasami, 1979, ‘Monkonosaurus lawulacus’ Zhao in Dong, 1990) were excluded (Maidment and Wei 2006; Maidment et al. 2008) and 20 taxa were added: Isaberrysaura mollensis Salgado et al., 2017 (Salgado et al. 2017), H. taibaii (Zhou 1984; Sereno and Dong 1992; Galton and Upchurch 2004; Peng et al. 2005; Maidment et al. 2006; Li et al. 2024a), Bashanosaurus primitivus Dai et al., 2022 (Dai et al. 2022), Baiyinosaurus baojiensis Li et al., 2024b (Li et al. 2024b), Gigantspinosaurus sichuanensis Ouyang, 1992 (Peng 2005; Maidment et al. 2008; Hao et al. 2018; Li et al. 2024a), Chungkingosaurus jiangbeiensis Dong, Zhou & Chang, 1983 (Dong et al. 1983; Dong 1990; Galton and Upchurch 2004; Maidment and Wei 2006; Li et al. 2024a), Mongolostegus exspectabilis Tumanova & Alifanov, 2018 (Tumanova and Alifanov 2018), Tuojiangosaurus multispinus Dong, Li, Zhou & Chang, 1977 (Dong et al. 1977, 1983; Galton and Upchurch 2004; Maidment and Wei 2006; Li et al. 2024a), Paranthodon africanus Broom, 1910 (Galton and Coombs 1981; Raven and Maidment 2018; pers. obs.), Loricatosaurus priscus Nopcsa, 1911a (Galton 1985, 1990, 2016; pers. obs.), He. mjosi (Carpenter et al. 2001; Galton and Upchurch 2004; Siber and Möckli 2009; Biyon-Bruyat et al. (2010); Maidment et al. 2018), S. stenops (Gilmore 1914; Ostrom and McIntosh 1966; Galton and Upchurch 2004; Escaso et al. 2007a; Maidment et al. 2015; pers. obs.), Jiangjunosaurus junggarensis Jia, Foster, Xu & Clark, 2007 (Jia et al. 2007; Li et al. 2024a), Wuerhosaurus homheni Dong, 1973 (Dong 1973, 1990, 1993; Maidment et al. 2008; Li et al. 2024a), Yanbeilong ultimus Jia et al., 2024 (Jia et al. 2024), Alcovasaurus longispinus Gilmore, 1914 (Gilmore 1914; Galton and Carpenter 2016), Kentrosaurus aethiopicus Hennig, 1915a (Hennig 1915a, 1916, 1925; Galton 1982, 1988; Galton and Upchurch 2004; Mallison 2010; Pereda-Suberbiola et al. 2013; pers. obs.), Thyreosaurus atlasicus Zafaty et al., 2024 (Zafaty et al. 2024), Adratiklit boulahfa Maidment, Raven, Ouarhache & Barrett, 2020 (Maidment et al. 2020; pers. obs.), and D. armatus (Owen 1875; Nopcsa 1911b; Galton 1985, 1991; Casanovas-Cladellas et al. 1995; Escaso et al. 2007b; Mateus et al. 2009; Cobos et al. 2010; Escaso 2014; Costa and Mateus 2019; Sánchez-Fenollosa et al. 2022, 2025; pers. obs.; this paper). Moreover, the Qiketai stegosaur (Li et al. 2024a) and the Zhongpu stegosaur (Li et al. 2024c) were also included. Therefore, 30 Operational Taxonomic Units (OTUs) were considered (Table 1; Files S1 [1.3], S2).

To further test the taxonomic assignment of the stegosaurian specimen from RD-34 (Figs 2–4), it was also coded as an independent OTU (File S3).

Figure 2. 

Cranium (MAP-9029) of Dacentrurus armatus Owen, 1875 from the Están de Colón (RD-34) fossil site (Riodeva, Teruel, Spain). 3D model of MAP-9029 in dorsal (A) and ventral (B) views. Produced and visualized using the Scantech iReal 2E scanner and IREAL 3D 2023 v. 3.3.3.3 software.

Figure 3. 

Cranium (MAP-9029) of Dacentrurus armatus Owen, 1875 from the Están de Colón (RD-34) fossil site (Riodeva, Teruel, Spain). Photographs (A, C, E) and interpretative drawings (B, D, F) of MAP-9029 in dorsal (A, B), ventral (C, D), and proximal (E, F) views. Abbreviations: f, frontal; fo, fossa; p, parietal; pocp, paroccipital process; po, postorbital; pf, prefrontal; sc, sagittal crest; soc, supraoccipital; sq, squamosal; stfe, supratemporal fenestra.

Figure 4. 

Mid cervical vertebra (MAP-9030) of Dacentrurus armatus Owen, 1875 from the Están de Colón (RD-34) fossil site (Riodeva, Teruel, Spain). Photographs (A–E) of MAP-9030 in anterior (A), left lateral (B), posterior (C), dorsal (D), and ventral (E) views. Abbreviations: cr, cervical rib; crc, cervical rib canal; dia, diapophysis; nc, neural canal; ns, neural spine; para, parapophysis; poz, postzygapophysis; prz, prezygapophysis; spozl, spinopostzygapophyseal lamina.

Maximum Parsimony (MP) analyses were conducted to infer the phylogenetic relationships of stegosaurs (File S1 [table S2]). The new data matrices (Files S2, S3) were analysed utilizing the TNT v.1.6 software (Goloboff and Morales 2023) and executing an initial search with ‘New Technology’ applying default options for ‘Sectorial Search’ and ‘Tree Fusing’ and setting 50 iterations of ‘Ratchet’ and 30 cycles of ‘Drift’. The Most Parsimonious Tree(s) (MPT) obtained was used as the starting point for a round of the Tree Bisection Reconnection (TBR) algorithm using the ‘Traditional search’ option. The IterPCR method (Pol and Escapa 2009) using the command ‘pcrprune’ (Goloboff and Szumik 2015) was employed to identify the rogue taxa. The resolution of the analyses was improved by excluding the rogue taxa (both a priori and a posteriori). Bremer Support values were calculated using the Bremer Support script and the Standard Bootstrap values (absolute frequencies) were calculated by applying the ‘Resampling’ function with 1000 replicates. Consistency (CI) and Retention (RI) indices were calculated utilizing the WSTATS script. Synapomorphies were obtained using the ‘Map Common synapomorphies’ option (File S1 [table S3]). We applied Equal Weighting (EW), and Implied Weighting (IW) setting the concavity constant ‘k’ to three and 12 (File S1 [table S2]). The analyses were conducted under two treatments (File S1 [table S2]): (1) considering all characters as unordered and (2) with multistate characters ordered (except char. 47 and 57).

Time-scaling was performed a posteriori using the function ‘bin_timePaleoPhy’ of the package ‘paleotree’ (Bapst 2012) in the R library (R Core Team 2023). Default arguments were applied except for ‘type = equal’, ‘vartime = 0.1’, ‘ntrees = 1’, and ‘add.term = T’. The a posteriori time-scaling method used (type = equal) was developed by Brusatte et al. (2008). This method was originally outlined by Ruta et al. (2006) but modified later (Brusatte et al. 2008; Brusatte 2011). The time data of each taxon was treated as a column of precise first and last appearances (dateTreatment = firstLast). The tree topology is provided in Newick format (File S4). The interval times data used were obtained from the International Chronostratigraphic Chart v2023/09 (File S5). Finally, the first and last interval times of each OTU (FAD and LAD) (File S6) were based on the available bibliography (File S1 [1.3, and supplementary references therein]).

The new data matrix comprised 115 morphological characters and 30 OTUs (File S2), increasing to 31 OTUs when the specimen from RD-34 was coded as an independent OTU (File S3). The IterPCR method identified M. exspectabilis and J. junggarensis as rogue taxa. Multiple analyses (File S1 [table S2]) were conducted with the primary objective of identifying unstable regions of the tree (Fig. 5; File S1 [figs S3–S8]) and variations in the synapomorphies of the clades (File S1 [1.6]).

In the absence of rogue taxa exclusions, MP analyses recovered 10 MPTs (characters not ordered, multistate characters ordered, and IW k = 12) and 16 MPTs (IW k = 3). The Strict Consensus Trees (SCTs) derived from these analyses exhibited poor resolution at the base of Stegosauria (File S1 [figs S3, S4]). However, when M. exspectabilis was excluded a posteriori, the resulting Reduced Strict Consensus Tree (RSCT) (File S1 [fig. S2]) was highly resolved and had a topology very similar to that of other analyses (Fig. 5; File S1 [1.5]).

MP analyses, excluding M. exspectabilis a priori, recovered a single MPT (Fig. 5) using unordered characters (194 steps, CI = 0.660, and RI = 0.799) and ordered multistate characters (196 steps, CI = 0.653, and RI = 0.803). When the analysis was conducted applying IW with k = 12, a single MPT was also recovered, but with a slightly different topology (File S1 [fig. S6]). However, when applying IW with k = 3, two MPTs were obtained, resulting in some changes to the topology and composition of Huayangosauridae and Stegosauridae (File S1 [fig. S5]). The same pattern is observed when the analyses were performed excluding both rogue taxa (File S1 [figs S7, S8]).

The topology (Fig. 5) obtained from the analyses 5 and 6 (File S1 [table S2]) was the most reliable for understanding stegosaurian evolutionary history due to its balance between high resolution (single MPT), minimal exclusion of rogue taxa (only the extremely fragmentary M. exspectabilis), and consistency with other alternative phylogenetic analyses (File S1 [1.5, 1.6]). Although some authors have suggested that IW parsimony outperforms EW (Goloboff et al. 2018; Ezcurra et al. 2023), others have pointed out that IW propagate errors and lead to reduced topological accuracy (Congreve and Lamsdell 2016; O’Reilly et al. 2016). In this context, EW parsimony remains the most widespread methodology applied in palaeontology, including in studies on stegosaurian dinosaurs.

Sc. lawleri was recovered as the most early-diverging thyreophoran included in these analyses (Fig. 5; File S1 [figs S2–S8]. Sc. lawleri, E. ernsti, Y. kopchicki, and Sce. harrisonii were excluded from Eurypoda (Ankylosauria + Stegosauria) (Fig. 5; File S1 [figs S2–S8]). The ankylosaurs Eu. carbonensis, Euo. tutus, and A. magniventris were recovered in a monophyletic group (Ankylosauria) and set as a sister group of Stegosauria (Fig. 5; File S1 [figs S2–S8]). When the extremely fragmentary M. exspectabilis was excluded (both a priori and a posteriori), Stegosauria was highly resolved and formed by two major sister clades: Huayangosauridae and Stegosauridae (Fig. 5; File S1 [figs S2, S5–S8]). M. exspectabilis was recovered as a huayangosaurid or an early-diverging stegosaurid (File S1 [figs S3, S4]).

According to the topologies obtained applying EW and IW with k = 12, Huayangosauridae included I. mollensis, H. taibaii, B. primitivus, Ba. baojiensis, and G. sichuanensis (Fig. 5; File S1 [figs S2, S6, S7]). I. mollensis was recovered as the most early-diverging huayangosaurid (Fig. 5; File S1 [figs S2, S6, S7]). H. taibaii and B. primitivus, and Ba. baojiensis and G. sichuanensis were recovered as sister taxa, respectively ([H. taibaii + B. primitivus] + [Ba. baojiensis + G. sichuanensis]) (Fig. 5; File S1 [figs S2, S6, S7]). Stegosauridae consisted of C. jiangbeiensis as the most early-diverging member (Fig. 5; File S1 [figs S2, S6, S7]), followed by a clade formed by T. multispinus and P. africanus (Fig. 5; File S1 [figs S2–S8]), and the newly defined clade Neostegosauria (Dacentrurinae + Stegosaurinae) (Fig. 5; File S1 [figs S2–S8]). However, in the IW analyses with k = 3, Huayangosauridae only included I. mollensis, H. taibaii, and B. primitivus in a polytomy (File S1 [figs S5, S8]), while Ba. baojiensis and G. sichuanensis were placed in Stegosauridae, forming a polytomy with C. jiangbeiensis and the clade formed by T. multispinus and P. africanus and Neostegosauria ([T. multispinus + P. africanus] + Neostegosauria) (File S1 [figs S5, S8]).

Stegosaurinae comprised two sister clades (Fig. 5; File S1 [figs S2–S8]). The first clade included the Qiketai stegosaur, Lo. priscus, He. mjosi, and S. stenops (Fig. 5; File S1 [figs S2–S8]). The second clade included J. junggarensis, the Zhongpu stegosaur, W. homheni, and Ya. ultimus (Fig. 5; File S1 [figs S2–S8]).

Finally, Dacentrurinae included Al. longispinus, K. aethiopicus, Th. atlasicus, Ad. boulahfa, and D. armatus (Fig. 5; File S1 [figs S2–S8]). In the analysis in which the specimen from RD-34 was treated as an independent OTU, it was grouped with D. armatus (File S1 [fig. S1]).

MAP-9029 (Figs 2, 3) exhibits a general morphology and a combination of characters typical of stegosaurian dinosaurs, including evidence of some cortical remodelling, the exclusion of the frontal from the dorsal orbital margin by the supraorbitals, and the presence of open supratemporal fenestrae. Stegosaurian cranial material from the Upper Jurassic of Europe is very scarce. Previous to this work, only two stegosaurian specimens preserve cranial elements (Fig. 6A–G; Mateus et al. 2009; Costa and Mateus 2019) and they have been referred to D. armatus (Sánchez-Fenollosa et al. 2025). ML 433 preserves a right premaxilla, a left maxilla, a left nasal, a postorbital fragment, and right and left angulars (Fig. 6 A–G; Mateus et al. 2009). MG 4863 preserves an anterior fragment of the left dentary and an incomplete left quadrate (Costa and Mateus 2019). MAP-9029 (Figs 2, 3) consists of the most complete skull from the European stegosaurian fossil record. No overlapping cranial material exists between specimens, so they cannot be compared. However, this makes D. armatus one of the stegosaurs with a better-known skull anatomy.

Figure 6. 

Cranial material of Dacentrurus armatus Owen, 1875 (ML 433, ‘Miragaia longicollum’ Mateus, Maidment & Christiansen, 2009 holotype) from the Upper Jurassic of Portugal (A–G) and Stegosaurus stenops Marsh, 1887 (NHMUK PV R36730) from the Upper Jurassic of USA (H–K). Right premaxilla (A, B, H, I) in lateral (A, H), and ventral (B, I) views. Left maxilla (C, D, J) in lateral (C, J), and ventral (D) views. Left nasal (E, F, K) in lateral (E), and dorsal (F, K) views. Right angular (G) in lateral view. Abbreviations: en, external naris; ts, tooth socket. Numbers indicate the morphological character and its scoring.

D. armatus (Figs 2, 3A–D) and other stegosaurs such as the huayangosaurids H. taibaii (Fig. 7G), I. mollensis (Salgado et al. 2017), and Ba. baojiensis (Li et al. 2024b) and the stegosaurids T. multispinus (Maidment and Wei 2006), He. mjosi (Carpenter et al. 2001; Maidment et al. 2018), S. stenops (Fig. 7D, H), and K. aethiopicus (Galton 1988) exhibit the plesiomorphic character of open supratemporal fenestrae (char. 2.0) also present in early-diverging ornithischians such as L. diagnosticus (Fig. 7A) and early-diverging thyreophorans such as Sc. lawleri (Breeden and Rowe 2020), E. ernsti (Haubold 1990), Y. kopchicki (Yao et al. 2022), and Sce. harrisonii (Fig. 7B). This contrast with the apomorphic closed and/or covered supratemporal fenestrae (char. 2.1) observed in ankylosaurs (e.g., Fig. 7E; Arbour and Currie 2013; Leahey et al. 2015; Arbour and Mallon 2017; Xing et al. 2024).

Figure 7. 

Skulls of ornithischian dinosaurs (A–E) and interpretative drawings in dorsal view (F–H). A, Lesothosaurus diagnosticus Galton, 1978 (NHMUK PV RU B23). B Scelidosaurus harrisonii Owen, 1861 (NHMUK PV R1111). C, F Dacentrurus armatus Owen, 1875 (MAP-9029). D Stegosaurus stenops Marsh, 1887 (NHMUK PV R36730). E Europelta carbonensis Kirkland et al., 2013 (AR-1-544/10). G Huayangosaurus taibaii Dong, Tang & Zhou, 1982 (modified from Sereno and Dong 1992). H S. stenops (modified from Galton and Upchurch 2004). Abbreviations: aso, anterior supraorbital; f, frontal; l, lacrimal; mso, medial supraorbital; n, nasal; o, orbit; p, parietal; pf, prefrontal; pm, premaxilla; po, postorbital; pso, posterior supraorbital; sq, squamosal; stfe, supratemporal fenestra. Numbers indicate the morphological character and its scoring.

Premaxillary teeth are present (char. 10.0) in huayangosaurids such as H. taibaii (Sereno and Dong 1992) and I. mollensis (Salgado et al. 2017). These retain the plesiomorphic condition of this character present in early-diverging ornithischians such as L. diagnosticus (Sereno 1991; Porro et al. 2015; pers. obs. [NHMUK PV R8501]), and early-diverging thyreophorans such as Sc. lawleri (Breeden et al. 2021), E. ernsti (Haubold 1990), and Sce. harrisonii (Norman 2020a). Edentulous premaxilla (char. 10.1) is present in stegosaurids such as C. jiangbeiensis (Dong et al. 1983), P. africanus (Raven and Maidment 2018; pers. obs. [NHMUK PV R47338]), He. mjosi (Siber and Mockli 2009), S. stenops (Fig. 6H, I), and D. armatus (Fig. 6A). Loss of premaxillary teeth allowed the development of a keratinous rhamphotheca (Czerkas 1999). The acquisition of this feature has probably been homoplastic between stegosaurs and ankylosaurs because early-diverging taxa of both groups have toothed premaxilla (e.g., Carpenter et al. 1998). Neostegosaurs such as D. armatus (Fig. 6B) and S. stenops (Figs 6I, 7G) have a broad notch between premaxillae on the midline (char. 5.1). However, it is absent (char. 5.0) in the early-diverging ornithischian L. diagnosticus (Sereno 1991; pers. obs. [NHMUK PV R8501]), in the early-diverging thyreophorans Sc. lawleri (Breeden et al. 2021) and Sce. harrisonii (Norman 2020a), and in the huayangosaurid H. taibaii (Fig. 7G; Sereno and Dong 1992). Therefore, huayangosaurids retain the plesiomorphic condition. The premaxilla of D. armatus (Fig. 6A) has an elongated nasal process that forms the anterodorsal margin of the external naris (char. 7.1) similar to the stegosaurids C. jiangbeiensis (Dong et al. 1983; Galton and Upchurch 2004) and S. stenops (Figs 6H, 7G). This contrast to the condition observed in the early-diverging ornithischian L. diagnosticus (Sereno 1991), the early-diverging thyreophoran Sc. lawleri (Breeden et al. 2021), and the huayangosaurid H. taibaii (Sereno and Dong 1992), which nasal process only forms the anterior margin of the external naris (char. 7.0). Moreover, the huayangosaurid H. taibaii has an abbreviated subnarial portion (char. 9.1) (Fig. 7G), whereas neostegosaurs such as D. armatus (Fig. 6A) and S. stenops (Figs 6H, 7G) have an elongated subnarial portion (char. 9.0). Therefore, probably huayangosaurids displayed massive and short premaxilla and stegosaurids acquired a slender and elongated premaxilla in their evolutionary history.

Most of stegosaurs, including D. armatus (Fig. 6C, D), exhibit a maxillary tooth row inset medially (char. 13.1). This condition derived from the tooth row in line with the lateral edge of the premaxilla (char. 13.0) present in early-diverging ornithischians such as L. diagnosticus (Sereno 1991; Porro et al. 2015; pers. obs. [NHMUK PV RU B17, NHMUK PV RU B23, NHMUK PV R8501, and NHMUK PV R11956]), and in early-diverging thyreophorans such as Sc. lawleri (Breeden et al. 2021) and E. ernsti (Haubold 1990). The early-diverging thyreophorans Y. kopchicki (Yao et al. 2022) and Sce. harrisonii (Norman 2020a; pers. obs. [NHMUK R1111]) also has tooth row inset medially like most of stegosaurs revealing that this feature was acquired at an early stage of the evolution of thyreophorans. In general, ankylosaurs exhibit deep buccal emarginations (char. 13.2). Therefore, the condition observed in T. multispinus (Maidment and Wei 2006) and P. africanus (Raven and Maidment 2018; pers. obs. [NHMUK PV R47338]) probably consist of a derived trait and is homoplastic (char. 13.0).

In D. armatus (Figs 2, 3A–D), as well as in other stegosaurs such as H. taibaii and S. stenops (Fig. 7D, G, H), the frontal is excluded from the orbital rim (char. 15.1). This also occurs in early-diverging thyreophorans such as Y. kopchicki (Yao et al. 2022) and Sce. harrisonii (Fig. 7B) and ankylosaurs (e.g., Fig. 7E; Arbour and Currie 2013; Leahey et al. 2015; Arbour and Mallon 2017; Xing et al. 2024). However, early-diverging ornithischians such as L. diagnosticus (Fig. 7A) and the early-diverging thyreophorans Sc. lawleri (Breeden and Rowe 2020) and E. ernsti (Haubold 1990) exhibit a frontal that form the dorsal margin of the orbit. Moreover, they have a free and rod-like palpebral bone (char. 15.0). The incorporation of the palpebral bone into the skull roof as supraorbital elements is responsible for the exclusion of the frontal from the dorsal margin of the orbit in Y. kopchicki, Sce. harrisonii, ankylosaurs, and stegosaurs (Maidment and Porro 2010). The frontal of D. armatus (Figs 2A, 3A, B) is longer than wide (char. 16.0) and is excluded from the anterior margin of the supratemporal fenestra (char. 18.1) similar to that in S. stenops (Fig. 7D, H). These probably represent apomorphic characters derived from the equidimensional or wider than long frontal (char. 16.1) and a postorbital that does not contact the parietal (char. 18.0) present in huayangosaurids (Fig. 7G; Sereno and Dong 1992; Salgado et al. 2017; Li et al. 2024b) and in non-neostegosaurian stegosaurids such as T. multispinus (Maidment and Wei 2006).

The parietal of D. armatus (Figs 2A, 3A, B) is convex (char. 19.0) with a remarkably sagittal crest. This condition is more similar to that in T. multispinus (Dong et al. 1983) than to the flat dorsal surface (char. 19.1) of the huayangosaurid H. taibaii (Sereno and Dong 1992), and the neostegosaurs K. aethiopicus (Galton 1988), He. mjosi (Carpenter et al. 2001; Maidment et al. 2018), and S. stenops (Fig. 7D). In extant mammals, the sagittal crest has been evolved independently in several lineages. Generally, it allows attachment for larger temporalis muscles, providing higher bite force and/or increased masticatory processing of tough food (e.g., Tanner et al. 2008; Figueirido et al. 2014; DeSantis et al. 2020). Therefore, the presence of a sagittal crest might have enabled D. armatus to chew for longer periods, which would be particularly beneficial when consuming tough foods with low nutritional value. However, in this case, the sagittal crest is relatively small, and a detailed functional study will be necessary to confirm this hypothesis.

The supraoccipital of D. armatus (Figs 2, 3) is posteroventrally oriented and clearly visible in dorsal view (char. 27.0). This condition contrast with the ventrally oriented supraoccipital that forms an angle of 90° with the dorsal plane of the skull roof (char. 27.1) present in other stegosaurs (Fig. 7D; Dong et al. 1983; Galton 1988; Sereno and Dong 1992; Galton and Upchurch 2004; Maidment et al. 2018). Therefore, it is likely to be an autapomorphic character for D. armatus. The extremely elongated neck and the axial muscles attachment may be related functionally with this cranial feature. Furthermore, this feature probably has implications for the position and orientation of the skull.

Finally, the dorsoventral ridge present in the supraoccipital (char. 28.0) of early-diverging ornithischians such as L. diagnosticus (Sereno 1991; Porro et al. 2015; pers. obs. [NHMUK PV RU B23]), early-diverging thyreophorans such as Y. kopchicki (Yao et al. 2022) and Sce. harrisonii (Norman 2020a; pers. obs. [NHMUK PV R1111]), and some stegosaurs (Fig. 7D; Sereno and Dong 1992; Dong et al. 1983; Galton and Upchurch 2004; Maidment et al. 2018), disappears (char. 28.1) in the dacentrurines K. aethiopicus (Galton 1988) and D. armatus (Figs 2A, 3E, F).

Madzia et al. (2021) formalized three pre-existing stegosaurian clade names (Stegosauria, Huayangosauridae, and Stegosauridae). However, they did not formalize other pre-existing clade names like Stegosaurinae (Sereno 2005) and Dacentrurinae (Mateus et al. 2009). These clades have been recovered in many phylogenetic analyses but with different compositions and topologies (e.g., Carpenter et al. 2001; Galton and Upchurch 2004; Maidment et al. 2008; Raven and Maidment 2017; Jia et al. 2024), making it challenging to compare results among studies. Establishing formal definitions for these groups helps standardize terminology and improves clarity when discussing phylogenetic relationships. For this reason, Stegosaurinae and Dacentrurinae are formalized by following the requirements and recommendations of the PhyloCode and keeping the original and traditional concept. Moreover, a new clade name, Neostegosauria, is proposed. Neostegosauria includes the late-diverging members of Stegosauridae and it has been also recovered in other analyses but with varying composition and topology (e.g., Carpenter et al. 2001; Raven and Maidment 2017; Dai et al. 2022; Jia et al. 2024; Li et al. 2024a, 2024c). According to the MP analyses, neostegosaurs are characterized by the possession of the following synapomorphies (File S1 [table S3]): (1) a frontal that is longer than wide (char. 16.0), (2) a medial process of the postorbital that contacts with the parietal (char. 18.1), (3) anterior and mid caudal vertebrae with a dorsal process on the transverse processes (char. 62.1), a scapula with (4) a subquadrangular acromial process with a posterodorsal corner (char. 72.1) and (5) a parallel sided blade (char. 73.1), and a sacral yoke with (6) large foramina between ribs (char. 86.1). An additional synapomorphy (char. 33.0) is recognized when the rogue taxon J. junggarensis was excluded from the analyses (File S1 [table S3]). Defining these phylogenetic names not only ensures consistency but also facilitates communication, discussion, and comparison between stegosaurian evolutionary hypotheses.

Raven and Maidment (2017) incorporated continuous characters into their data matrix to improve tree resolution. However, they did not account for intraspecific variation or variation within the same series in the case of the axial skeleton. Continuous characters significantly influence inferred evolutionary relationships, over 50% of apomorphic characters correspond to continuous characters (e.g., Raven and Maidment 2017; Jia et al. 2024; Li et al. 2024b), despite these representing only 20% of the total characters. Therefore, misapplication of continuous characters, along with other occasional errors, likely explains why MP analyses based on this data matrix and its derivatives (Maidment et al. 2020; Dai et al. 2022; Jia et al. 2024; Li et al. 2024b, 2024c; Zafaty et al. 2024), produce notably longer and more homoplastic trees than those presented here. This also explain certain inconsistencies with fossil evidence, direct comparisons, and taxonomy. For instance, recent phylogenetic analyses (e.g., Raven and Maidment 2017; Maidment et al. 2020; Dai et al. 2022; Li et al. 2024b) did not recover the holotypes of D. armatus and ‘Mi. longicollum’ as sister taxa because the continuous characters related to the proportions of dorsal vertebrae for ‘Mi. longicollum’ were coded solely based on the preserved first and second dorsal vertebrae (see Sánchez-Fenollosa et al. 2025 for details).

CI and RI values from MP analyses are higher than those reported in recent analyses (e.g., Raven and Maidment 2017; Dai et al. 2022; Jia et al. 2024; Li et al. 2024a, 2024b). This implies that these analyses potentially have greater coherence between morphological characters and inferred phylogenetic relationships. However, Bremer Support and Bootstrap values are similar to those in other analyses (e.g., Raven and Maidment 2017; Jia et al. 2024; Li et al. 2024a, 2024b). These low values indicate weak robustness and confidence for most stegosaurian clades. This trend across stegosaurian phylogenetic analyses may result from the fact that most species are represented by a single partial skeleton and highly incomplete material. Therefore, fieldwork and new fossil discoveries will be crucial for advancing the understanding of stegosaurian evolutionary history.

Several synapomorphies support the major clades of Thyreophora, Ankylosauria, and Stegosauria (File S1 [table S3]). However, this new data matrix only includes a few non-stegosaurian taxa (Table 1; File S2). Therefore, these synapomorphies (File S1 [table S3]) are limited and should be considered with caution. The taxa Sc. lawleri (Kayenta Formation, Sinemurian–Toarcian, USA), E. ernsti (Posidonienschiefer Formation, early Toarcian, Germany), Y. kopchicki (Fengjiahe Formation, late Sinemurian–Toarcian, China), and Sce. harrisonii (Charmouth Mudstone Formation, late Sinemurian, UK) are clearly outside of Eurypoda and are considered early-diverging thyreophorans (Fig. 5; File S1 [figs S2–S8]). Other authors obtained similar results studying the evolutionary relationships of ornithischians and thyreophorans (e.g., Sereno 1999; Thompson et al. 2012; Baron et al. 2017; Raven and Maidment 2017; Han et al. 2018; Dieudonné et al. 2021; Raven et al. 2023; Fonseca et al. 2024). However, these results differ from those of Wiersma and Irmis (2018), in which Sce. harrisonii is recovered as an ankylosaur, and Norman (2021), in which Sc. lawleri, E. ernsti and Sce. harrisonii are recovered as ankylosauromorphans (= ankylosaurs sensu Madzia et al. 2021).

The time-scaled tree suggests that first stegosaurs may have appeared during the Early Jurassic and declined, coincident with the rise of ankylosaurs, in the Early Cretaceous (Fig. 5). Cretaceous stegosaurs are represented by a handful of valid taxa: M. exspectabilis, W. homheni, Ya. ultimus from Asia (Dong 1990; Tumanova and Alifanov 2018; Jia et al. 2024) and P. africanus from Africa (Raven and Maidment 2018). Fragmentary stegosaurian remains have been reported from at least the Lower Cretaceous of Europe (Galton 1981; Pereda-Suberbiola et al. 2003; Allain et al. 2022) and South America (Pereda-Suberbiola et al. 2013), and the Upper Cretaceous of India (Yadagiri and Ayyasami 1979).

I. mollensis was originally interpreted as an early-diverging ornithopod (Salgado et al. 2017). However, it is here recovered as a stegosaur (Fig. 5; File S1 [figs S2–S8]) and is included within the clade Huayangosauridae (Fig. 5; File S1 [figs S2, S5–S8]). This taxon exhibits several features shared with other huayangosaurids and early-diverging stegosaurids: (1) large antorbital fossa (char. 3.0; shared with H. taibaii), (2) presence of premaxillary teeth (char. 10.0; shared with H. taibaii), (3) presence of an anterior maxillary foramen (char. 12.1; shared with H. taibaii), (4) asymmetrical tooth crown (char. 36.0; shared with H. taibaii, G. sichuanensis, and C. jiangbeiensis), (5) > 25 maxillary/dentary teeth (char. 38.2; shared with H. taibaii and G. sichuanensis), and (6) weakly developed cingulum (char. 39.0; shared with H. taibaii). Other authors have also suggested that I. mollensis is a stegosaur (Han et al. 2018; Raven and Maidment 2018; Raven et al. 2023; Fonseca et al. 2024). The type specimen of I. mollensis is not fully prepared, so further work on this specimen and the discovery of new fossils may allow a better understand of the evolutionary relationships of this species. I. mollensis is especially important because it is the oldest known stegosaurian taxon and the first described from South America (lower Bajocian, Argentina). Fragmentary stegosaurian specimens have been also found in the Upper Jurassic and Lower Cretaceous of South America (Pereda-Suberbiola et al. 2013; Rauhut et al. 2021).

M. exspectabilis has been included for the first time in a stegosaurian phylogenetic analysis and it is recovered as a huayangosaurid or an early-diverging stegosaurid (File S1 [figs S3, S4]). This taxon was described from very fragmentary material: six caudal vertebrae, two partial pubes, and a fragment of a sacral rib (Tumanova and Alifanov 2018). However, M. exspectabilis exhibits some features exclusively present in huayangosaurids and early-diverging stegosaurids: anterior and mid caudal vertebrae with (1) transverse processes without dorsal process (char. 62.0) and (2) neural spines with unexpanded apex (char. 66.0), and a pubis with (3) a dorsal edge of the postpubis without a kink (char. 94.0). It was found in beds of the Dzundain Formation (Mongolia, Asia) with an age of Lower Cretaceous (Aptian–Albian). Therefore, M. exspectabilis is one of the youngest stegosaurs currently known and these results suggest that a lineage of huayangosaurids or early-diverging stegosaurids persisted in Asia until at least the late Early Cretaceous.

The MP analyses recovered Huayangosauridae and Stegosauridae with small variations in composition and topology depending on the methodology applied (Fig. 5; File S1 [figs S2, S5–S8]). In most analyses (Fig. 5; File S1 [table S2; figs S2, S6, S7]), Huayangosauridae includes I. mollensis (Los Molles Formation, early Bajocian, Argentina), H. taibaii (lower Shaximiao Formation, Bathonian–early Oxfordian, China), B. primitivus (lower Shaximiao Formation, Bathonian, China), Ba. baojiensis (Wangjiashan Formation, late Bathonian, China), and G. sichuanensis (upper Shaximiao Formation, ?Kimmeridgian–?Tithonian, China) and it is supported by two synapomorphies (File S1 [table S3]): (1) presence of an anterior maxillary foramen (char. 12.1) and (2) > 25 maxillary/dentary teeth (char. 38.2). These results clearly differ from those of recent phylogenies (e.g., Hao et al. 2018; Maidment et al. 2020; Dai et al. 2022; Jia et al. 2024; Li et al. 2024b; Zafaty et al. 2024). However, they resemble those obtained by Li et al. (2024a) although C. jiangbeiensis and T. multispinus are recovered here as early-diverging members of Stegosauridae (Fig. 5; File S1 [figs S2, S5–S8]). The clade Stegosauridae is supported by two synapomorphies (File S1 [table S3]): a premaxilla with (1) an elongated nasal process that forms the anterodorsal margin of the external naris (char. 7.1) and a femur with (2) a fourth trochanter extremely reduced or inappreciable (char. 99.2). An additional synapomorphy (char. 51.2) is recognized when the multistate characters were treated as ordered (File S1 [table S3]). C. jiangbeiensis (upper Shaximiao Formation, ?Kimmeridgian–?Tithonian, China) is the earliest-diverging member and it retains some simplesiomorphic characters such as (1) asymmetrical tooth crowns (char. 36.0) and (2) neural spines of anterior and mid caudal vertebrae with unexpanded apex (char. 66.0).

When applying IW with k = 3, Huayangosauridae is more restricted, including only I. mollensis, B. primitivus, and H. taibaii (File S1 [figs S5, S8]), and is supported solely by the presence of an anterior maxillary foramen (char. 12.1) (File S1 [table S3]). In contrast, Ba. baojiensis and G. sichuanensis are recovered as early-diverging members of Stegosauridae (File S1 [figs S5, S8]), a clade supported in this case by the presence of (1) maxillary and dentary teeth with a greatly developed and ring-like cingulum (char. 39.1) and (2) cervical vertebrae with postzygapophyses elongated that slightly overhang the centrum facet (char. 45.1) (File S1 [table S3]).

T. multispinus (upper Shaximiao Formation, ?Kimmeridgian–?Tithonian, China) and P. africanus (Kirkwood Formation, ?Berriasian–?Valanginian, South Africa) exhibit a single synapomorphy: maxillary tooth row in line with the lateral edge of the premaxilla (char. 13.0). Both species have been recovered as sister taxa in some recent phylogenetic analyses (e.g., Raven and Maidment 2017; Dai et al. 2022; Jia et al. 2024; Zafaty et al. 2024). However, as noted Raven et al. (2023), this is likely due to a combination of missing data (P. africanus is extremely fragmentary) and similarities in tooth morphology (also present in other stegosaurs).

Neostegosauria is supported by several synapomorphies (see above and File S1 [table S3]) and is formed by Stegosaurinae and Dacentrurinae (Fig. 5; File S1 [figs S2–S8]). The first neostegosaurs (Fig. 5) appeared in the Middle Jurassic, suggesting that stegosaurs diversified rapidly during this epoch.

Stegosaurinae, comprising exclusively Laurasian taxa, exhibits a stable composition across all analyses (Fig. 5; File S1 [figs S2–S8]), although its synapomorphies vary depending on the methodology applied (File S1 [table S3]). This clade is invariably supported by two synapomorphies (File S1 [table S3]): dermal armour with (1) a parasagittal arrangement of rows alternating on each side of the midline (char. 110.1) and (2) dorsal plates with a generally thin structure (char. 112.1). A third synapomorphy (char. 37.0) is present in most analyses but is either absent or substituted (char. 64.1) under IW with k = 3 (File S1 [table S3]). When the rogue taxa J. junggarensis was excluded a priori, Stegosaurinae is supported by a single synapomorphy (char. 112.1) (File S1 [table S3]). Some stegosaurs exhibit variations in plate morphology that have been interpreted as potential evidence of sexual dimorphism (Saitta 2015). However, the synapomorphies supporting this clade, as well as the dermal plate characters incorporated in this data matrix, do not pertain to plate morphology. If these interpretations are correct, such variations likely reflect interspecific differences rather than phylogenetic signals.

In turn, Stegosaurinae is divided into two main lineages (Fig. 5; File S1 [figs S2–S8]). The first of them includes J. junggarensis (Shishugou Formation, early Oxfordian, China), W. homheni (late Valanginian–?Barremian, Lianmuqin and Jingchuan Formations, China), Ya. ultimus (Albian, Zuoyun Formation, China), and the Zhongpu stegosaur (Aptian–Albian, upper Hekou Group, China). In most analyses, this clade is supported by a single synapomorphy: some cervical centra with a remarkably foramen on lateral surface (char. 43.1). This character is present in J. junggarensis (Jia et al. 2007) and in the Zhongpu stegosaur (Li et al. 2024c), but unknown in W. homheni (Dong 1990) and Ya. ultimus (Jia et al. 2024). Additionally, W. homheni, Ya. ultimus, and the Zhongpu stegosaur share the presence of dorsal neural arches with anterior and posterior surfaces deeply excavated above the neural canal (char. 52.1, [?] in J. junggarensis). This character supports this clade under analyses 4 and 7 (File S1 [table S2]). Maidment et al. (2008) synonymized the genera Wuerhosaurus Dong, 1973 and Stegosaurus and diagnosed S. homheni based on a single autapomorphy (char. 52.1 here). However, the results of this study (Fig. 5; File S1 [figs S2–S8]) clearly distinguish the species W. homheni and S. stenops, leading us to recognize Wuerhosaurus and Stegosaurus as separate genera. Therefore, the Zhongpu stegosaur, originally referred to Stegosaurus sp. (Li et al. 2024c), requires taxonomic re-evaluation. Moreover, the presence of dorsal neural arches with anterior and posterior surfaces deeply excavated above the neural canal (char. 52.1; autapomorphy of W. homheni sensu Maidment et al. 2008) in both Ya. ultimus and the Zhongpu stegosaur supports the need for a taxonomic revision of Early Cretaceous Chinese stegosaurs.

The second lineage of stegosaurines (Fig. 5; File S1 [figs S2–S8]) includes the Qiketai stegosaur (Kimmeridgian–Tithonian, Qigu Formation, China), Lo. priscus (Callovian, Oxford Clay Formation and unnamed unit, UK and France), He. mjosi (Kimmeridgian–Tithonian, Morrison Formation, USA), and S. stenops (Kimmeridgian–Tithonian, Morrison and Lourinhã Formations, USA and Portugal) and it is supported by two synapomorphies: anterior and mid caudal vertebrae (posterior to Cd2) with (1) transverses processes greatly ventrally projected (char. 61.1) and a coracoid with (2) subcircular morphology (char. 68.1). In most analyses, the Qiketai stegosaur and Lo. priscus are recovered as sister OTUs (Fig. 5; File S1 [figs S2, S3, S6, S7]) consistent with Li et al. (2024a). The Morrison Formation stegosaurs, He. mjosi and S. stenops, are recovered as sister taxa (Fig. 5; File S1 [figs S2–S8]). The genus Hesperosaurus was proposed as a subjective junior synonym of Stegosaurus (Maidment et al. 2008, 2015). However, recent studies regard it as a separate genus (e.g., Carpenter 2010; Raven and Maidment 2017; Maidment et al. 2018), as originally proposed by Carpenter et al. (2001). Further osteological descriptions based on several relatively complete specimens housed in Sauriermuseum Aathal (Siber and Mockli 2009) may shed light on the stegosaurian taxonomy of the Morrison Formation.

The neostegosaurs Al. longispinus (Kimmeridgian–Tithonian, Morrison Formation, USA), K. aethiopicus (Kimmeridgian–Tithonian, Tendaguru Formation, Tanzania), Th. atlasicus (Bathonian–?Callovian, El Mers III Formation, Morocco), Ad. boulahfa (Bathonian, ?El Mers II Formation, Morocco), and D. armatus (Kimmeridgian–Tithonian, Kimmeridge Clay, Argiles d’ Octeville, Lourinhã, and Villar del Arzobispo Formations, UK, France, Portugal, and Spain) are included in Dacentrurinae (Fig. 5; File S1 [figs S2–S8]). This clade is supported by a single synapomorphy: presence of two pairs of long terminal caudal dermal spines (char. 114.1) (File S1 [table S3]). These phylogenetic analyses (Fig. 5; File S1 [figs S2–S8]) are the first to include Al. longispinus and K. aethiopicus within Dacentrurinae. Moreover, they also clearly distinguish the species Al. longispinus and D. armatus, thus supporting the separation of the genera Alcovasaurus Galton and Carpenter 2016 and Dacentrurus (Sánchez-Fenollosa et al. 2025; contra Costa and Mateus 2019). In recent phylogenetic analyses, Al. longispinus has been recovered as an early-diverging thyreophoran outside Eurypoda (Raven and Maidment 2017) or as an early-diverging stegosaurid (e.g., Maidment et al. 2020; Jia et al. 2024; Li et al. 2024a; Zafaty et al. 2024). Raven and Maidment (2017) erroneously coded several characters of Al. longispinus that explain the early-diverging position: (1) the presence of eight cervical vertebrae (char. 5 in Raven and Maidment 2017), (2) an acetabular length of ilium/dorsoventral height of pubic peduncle of ilium ratio of 1.8 (char. 18 in Raven and Maidment 2017), and (3) the presence of elongated posterior caudal vertebrae (char. 74 in Raven and Maidment 2017). According to Galton and Carpenter (2016), Al. longispinus preserves only one (or possibly two) cervical vertebrae, the ilium is extremely badly figured (field photos) and possibly broken, and the posterior caudal vertebrae are short. These errors were retained in later analyses based on the same data matrix (e.g., Maidment et al. 2020; Jia et al. 2024; Li et al. 2024a; Zafaty et al. 2024).

Th. atlasicus, Ad. boulahfa, and D. armatus are recovered as closely related taxa (Fig. 5; File S1 [figs S2–S8]). The clade that includes only these species is supported by a single synapomorphy: dorsal centra wider than long (char. 50.1). This character is also observed in several stegosaurian specimens from the Upper Jurassic and Berriasian (Lower Cretaceous) of Europe (e.g., Pereda-Suberbiola et al. 2003; Cobos et al. 2010; Company et al. 2010; Allain et al. 2022).

Regarding ichnological evidence, tracks with stegosaurian affinities like the ichnogenus Deltapodus Whyte & Romano, 1994 have been reported from the Middle Jurassic of Europe (Whyte and Romano 2001), the Upper Jurassic–Berriasian of almost world-wide (e.g., Cobos et al. 2010; Belvedere and Mietto 2010; Mateus et al. 2011; Pascual et al. 2012; Zhang et al. 2012; Lockley et al. 2017; Castanera et al. 2024), the Lower Cretaceous of China (Xing et al. 2013, 2021), and the Upper Cretaceous of India (Mohabey 1986). Moreover, tracks with thyreophoran (probably stegosaurian) affinities such as the ichnogenera Shenmuichnus Li et al., 2012 and Moyenisauropus Ellenberg 1974 have been reported from at least the Early Jurassic of Europe and Asia (Gierliński 1999; Li et al. 2012; Xing et al. 2016; Figueiredo et al. 2024). Therefore, the ichnological evidence appears consistent with the osteological record and the origin and evolutionary history of stegosaurs suggested here.