This is the accepted version of the journal article: Van den Brandt, Marc J.; Abdalaa, Fernando; Benoit, Julien; [et al.]. «Taxonomy, phylogeny and stratigraphical ranges of middle Permian pareiasaurs from the Karoo Basin of South Africa». Journal of systematic palaeontology, Vol. 19, Issue 19 (2021), p. 1367-1393. DOI 10.1080/14772019.2022.2035440 This version is available at https://ddd.uab.cat/record/259563 under the terms of the license 1 Taxonomy, phylogeny and stratigraphic ranges of middle Permian 2 pareiasaurs from the Karoo Basin of South Africa 3 Marc J. Van den Brandta*, Fernando Abdalaab, Julien Benoita, Michael O. Dayac, David P. 4 Groenewaldad and Bruce S. Rubidgea. 5 a Evolutionary Studies Institute and School of Geosciences, University of the Witwatersrand, 6 Johannesburg, PO WITS 2050, South Africa; bUnidad Ejecutora Lillo, CONICET-Fundación 7 Miguel Lillo, Miguel Lillo 251, Tucumán, Argentina; cNatural History Museum, Cromwell 8 Road, London SW7 5BD, United Kingdom; dInstitut Català de Paleontologia Miquel 9 Crusafont, Edifici ICTA-ICP, c/ Columnes s/n, Campus de la UAB, 08193 Cerdanyola del Vallès, 10 Barcelona, Spain. 11 *Corresponding author. Email: marcvandenbrandt@gmail.com 12 13 Suggested running head: Middle Permian pareiasaur taxonomy, phylogeny and 14 stratigraphic ranges 15 16 Abstract 17 18 Pareiasaurs were relatively abundant, globally distributed, herbivorous parareptiles of the 19 mid to late Permian. The basal-most forms, all members of the Bradysauria, are restricted to 20 the Guadalupian (mid-Permian) of South Africa and went extinct in the late Capitanian near 21 the top of the Tapinocephalus Assemblage Zone. Currently four species are recognised in 1 22 this group: Bradysaurus seeleyi, B. baini, Embrithosaurus schwarzi and Nochelesaurus 23 alexanderi. Those taxa have historically been poorly defined and based on a limited number 24 of specimens, leaving the taxonomic diversity of the group open to doubt and limiting their 25 utility in biostratigraphy. Here we present our fourth and final contribution designed to 26 improve the understanding of this group of pareiasaurs by providing a taxonomic and 27 phylogenetic review, updated stratigraphic ranges and updated diagnoses for each taxon of 28 the Bradysauria. Bradysaurus seeleyi is synonymised with Bradysaurus baini, resulting in 29 three valid mid-Permian pareiasaur taxa: Bradysaurus baini, Embrithosaurus schwarzi and 30 Nochelesaurus alexanderi. Our cladistic analysis of cranial and postcranial characters 31 supports the monophyly of Bradysauria with five synapomorphies. Embrithosaurus schwarzi 32 is recovered as the sister taxon to a clade containing Bradysaurus baini and Nochelesaurus 33 alexanderi. By identifying 157 pareiasaur specimens in fossil collections we show that the 34 Bradysauria are stratigraphically restricted to the Abrahamskraal Formation of the Beaufort 35 Group and suggest a staggered appearance. Bradysaurus baini is first to appear, followed by 36 Nochelesaurus alexanderi, and lastly by Embrithosaurus schwarzi. All three taxa perished 37 during the Capitanian mass extinction, and have their highest occurrences near the top of 38 the Abrahamskraal Formation. 39 Keywords: Bradysauria; Capitanian; Guadalupian; Pareiasauria; Tapinocephalus Assemblage 40 Zone. 41 42 Introduction 43 2 44 Pareiasaurs were a group of heavily-built, relatively abundant herbivorous parareptiles that 45 flourished world-wide in the middle to late Permian (Boonstra 1969; Lee 1993, 1994, 1997a, 46 1997b). They are characterised by an expanded ‘cheek flange’ on the side of the skull, 47 osteoderms, and cranial ornamentation. The South African Beaufort Group records the 48 highest diversity of pareiasaurian species, hosting eleven of the 24 species currently 49 recognised. In South Africa, pareiasaur remains are restricted to two distinct stratigraphic 50 ranges: one cohort of relatively large and lightly armoured genera is found in the 51 Guadalupian (mid-Permian) Tapinocephalus Assemblage Zone (AZ), while a second, 52 generally smaller and more heavily armoured cohort is found in the Lopingian (late Permian) 53 Endothiodon, Cistecephalus and Daptocephalus AZs (Boonstra 1932a; Kitching 1995; 54 Rubidge 1995; Smith & Keyser 1995a, 1995b, 1995c, 1995d; Lee 1997a; Day 2013; Day et al. 55 2015; Day & Rubidge 2020; Smith et al. 2020). Pareiasaurs were victims of both the late 56 Capitanian (Day et al. 2015) and end-Permian mass extinction events (Smith & Botha-Brink 57 2014). 58 The Guadalupian pareiasaurs, from the Tapinocephalus AZ, are the most basal genera of the 59 group and are therefore essential to understanding the origin, diversification, and 60 subsequent global dispersal of pareiasaurs (Lee 1997a, 1997b; Jalil & Janvier 2005; Tsuji 61 2010; Tsuji et al. 2013; Turner et al. 2015; Benton 2016; Liu & Bever 2018; Van den Brandt et 62 al. 2020). The taxonomy of Guadalupian pareiasaurs was chaotic until the late 20th Century, 63 when an alpha-taxonomic review of all pareiasaurs recognised only four species from the 64 Tapinocephalus AZ: Bradysaurus baini, B. seeleyi, Embrithosaurus schwarzi and 65 Nochelesaurus alexanderi (Lee 1994, 1997a). 3 66 However, this review considered relatively few specimens and was limited by the nature of 67 the fossils themselves, including poor historic preparation, damage, and deformation of 68 many specimens recovered from the exceptionally hard matrix of the Tapinocephalus AZ 69 (see Van den Brandt et al. 2020). Broader anatomical distinctions among the taxa and, 70 particularly their stratigraphic ranges, have remained poorly understood. Pareiasaurs were 71 an important component of the tetrapod fauna in the Karoo Basin during the late 72 Guadalupian and knowledge of their ranges and occurrences is necessary to understand 73 both the early evolution of the clade and ecosystem changes before and during the 74 Capitanian mass extinction on land. 75 To better understand the group, we initiated a project in 2015 to study middle Permian 76 pareiasaurs, resulting in an M.Sc, a Ph.D and a series of three descriptive papers (Van den 77 Brandt 2016, 2020; Van den Brandt et al. 2020, 2021a, 2021b). In our first descriptive paper, 78 we produced an updated cranial description and diagnosis for Embrithosaurus schwarzi, and 79 an updated phylogenetic analysis on cranial characters for pareiasaurs in which we 80 established the new clade Bradysauria, comprising the four species from the Tapinocephalus 81 AZ (Van den Brandt et al. 2020). The second paper produced the first detailed cranial 82 description and updated diagnosis for Nochelesaurus alexanderi (Van den Brandt et al. 83 2021a). In our third paper we produced the first modern comparative postcranial analysis 84 for the three genera of middle Permian South African pareiasaurs; Bradysaurus, 85 Embrithosaurus, and Nochelesaurus (Van den Brandt et al. 2021b). 86 This fourth paper integrates all new anatomical information into an updated taxonomy for 87 the Tapinocephalus AZ pareiasaurs, including updated diagnoses, phylogenetic analysis, 4 88 geographic distribution and biostratigraphic ranges for all three valid species within 89 Bradysauria. 90 Historical Background 91 The discovery of large numbers of pareiasaurian specimens in South Africa during the late 92 1800s and early 1900s led to the erection of many new species by early palaeontologists 93 (Owen 1876; Seeley 1888, 1892; Broom 1912, 1913, 1914, 1924). Those early descriptions 94 were often based on fragmentary material and included only short, ambiguous diagnoses, 95 causing much initial taxonomic confusion and disagreement. Some specimens were lost (eg. 96 Owen 1876, NHMUK PV R 4063/R1710/R1710a, holotype skull of Pareiasaurus serridens) or 97 mistakenly referred to other taxa (eg. Owen 1856, 1862, NHMUK PV R 4063/R1710/R1710a 98 referred to Dicynodon tigriceps). 99 The holotype of Pareiasaurus bombidens Owen 1876 (NHMUK PV R 1714/43525), the first 100 described pareiasaur from the Tapinocephalus AZ, has been ignored or considered 101 indeterminate by most researchers as it is fragmentary, not well preserved, and lacks 102 diagnostic characters (Watson 1914; Haughton & Boonstra 1929; Boonstra 1934; Lee 1994, 103 1997a; Van den Brandt 2016, 2020). NHMUK PV OR 49426 (holotype of Bradysaurus seeleyi) 104 was initially described by Seeley (1888) and referred to Owen’s P. bombidens, and NHMUK 105 PV R 1971 was described by Seeley (1892) as the holotype of Pareiasaurus (=Bradysaurus) 106 baini. Broom (1903) described SAM-PK-8034 (holotype of Embrithosaurus schwarzi) and 107 mistakenly referred it to Pareiasaurus serridens. 108 Concerned by the taxonomic confusion existing at the time, Watson (1914) produced the 109 first taxonomic reassessment of pareiasaurs, after he carefully re-examined museum 5 110 catalogues, reassigned incorrectly referred specimens, and re-evaluated holotypes held at 111 the Natural History Museum, London. Watson’s major contribution was to differentiate the 112 South African pareiasaurs into the large, basal, sparsely armoured forms from the 113 Tapinocephalus AZ (proposing the new genera Bradysaurus and Embrithosaurus), and the 114 small- and medium-sized, extensively armoured members of the later Cistecephalus AZ 115 (retaining the names of Pareiasaurus, Propappus, and Anthodon). In the process he correctly 116 observed that the genus Pareiasaurus occured only in the Cistecephalus AZ. 117 Broom (1916) disagreed with Watson’s (1914) new generic names for pareiasaurs from the 118 Tapinocephalus AZ and continued to create new species of Pareiasaurus and Propappus for 119 new specimens recovered from the Tapinocephalus AZ. Between 1912 and 1924, Broom 120 named and described four such species: Propappus rogersi Broom 1912, Pareiasaurus 121 acutirostris Broom 1913, Pareiasaurus whaitsi Broom 1914 and Pareiasaurus strubeni 122 Broom 1924. 123 The next major pareiasaurian revision was completed by Haughton and Boonstra between 124 1929 and 1934 in their eleven-part ‘Pareiasaurian Studies’ series of papers, which included 125 two taxonomic reassessment works based on cranial characteristics (Haughton & Boonstra 126 1929; Boonstra 1934). In their first taxonomic paper, Haughton & Boonstra (1929) noted 127 very brief diagnostic features. They focused primarily on dentition, separating the 128 pareiasaurs into three primary divisions by the number of cusps on their teeth. Secondarily, 129 considering the ratio of the width between the orbits relative to the width between the 130 tabulars, they further subdivided the taxa into three secondary subdivisions. This created, 131 by their own admission, a somewhat compromised classification scheme. For the 132 Tapinocephalus AZ they retained the two genera recognised by Watson (Bradysaurus and 6 133 Embrithosaurus), added two new species of Bradysaurus (B. seeleyi and B. vanderbyli), 134 named four new species based on new material (Dolichopareia angusta, Platyoropha 135 broomi, Nochelesaurus alexanderi and Brachypareia watsoni) and reassigned the four taxa 136 named by Broom between 1912 and 1924: Propappus rogersi to Brachypareia rogersi, 137 Pareiasaurus whaitsi to Bradysuchus whaitsi, Pareiasaurus strubeni to Nochelesaurus 138 strubeni, and Pareiasaurus acutirostris to Koalemasaurus acutirostris. Thus Haughton & 139 Boonstra (1929) recognised eight genera and twelve species from the Tapinocephalus AZ. 140 Boonstra (1934) later considered Platyoropha broomi a junior synonym of Bradysaurus baini 141 and so retained seven genera and eleven species recognised by Haughton & Boonstra 142 (1929). He also produced a detailed cranial description for a generalised pareiasaurian skull, 143 including comprehensive accounts of all individual cranial elements. He added more 144 detailed diagnostic cranial features to each species and illustrated comprehensively, 145 although roughly, all individual species in all major cranial views. However, he did not 146 include separate cranial descriptions for individual species and his updated species level 147 diagnoses included many shared features across many taxa (Lee 1994, 1997a; Van den 148 Brandt 2016, 2020). 149 The two landmark contributions of Haughton & Boonstra (1929) and Boonstra (1934) 150 resulted in a plethora of new pareiasaurian taxa with few clearly identifiable features to 151 separate species (Fig. 1). 152 <> 153 Broom (1935) produced a short summary classification in which he briefly discussed the 154 history of the important South African pareiasaur specimens discovered and species 7 155 described. By now he had accepted the two genera named by Watson (1914) for specimens 156 from the Tapinocephalus AZ. He reduced the number of taxa recognised by Haughton & 157 Boonstra (1929) and Boonstra (1934) by recognising only five species of Bradysaurus (B. 158 baini, B. acutirostris, B. whaitsi, B. strubeni and B. bombidens) and Embrithosaurus schwarzi. 159 Boonstra (1969) lumped all species with fewer than nine cusps on their teeth into 160 Bradysaurus (recognising four species: B. baini, B. seeleyi, B. strubeni and B. angusta) and 161 kept Embrithosaurus schwarzi for individuals with nine cusps on their teeth. He removed 162 several species that he had previously considered valid (Boonstra 1934) as they were either 163 based on immature specimens (Brachypareia) or on pathological features (Bradysuchus). 164 Kuhn (1969) likewise considered all basal South African taxa as either Bradysaurus (for those 165 with fewer than nine cusps per tooth) or Embrithosaurus (for those with nine cusps per 166 tooth). He considered as valid the full set of twelve species originally diagnosed by 167 Haughton & Boonstra (1929) and resurrected B. bombidens which had been rejected by all 168 previous workers except for Broom (1935), resulting in thirteen recognised species. Kuhn 169 (1969) therefore recognised nine species of Bradysaurus (B. baini, B. bombidens, B. 170 vanderbyli, B. seeleyi, B. whaitsi, B. acutirostris, B. watsoni, B. rogersi, and B. broomi) and 171 four species of Embrithosaurus (E. schwarzi, E. strubeni, E. alexandri, and E. angusta). 172 Lee (1994, 1997a) performed an alpha taxonomic reassessment of all pareiasaur species 173 around the world. Through a detailed study of all holotypes, he reduced the number of valid 174 species from 49 to 17. In the Tapinocephalus AZ, Lee considered only four of the original 175 twelve species recognised by Haughton & Boonstra (1929) to be valid: Bradysaurus seeleyi, 176 B. baini, Embrithosaurus schwarzi and Nochelesaurus alexanderi. Amended diagnoses for 8 177 each valid taxon were produced and he also undertook the first phylogenetic analysis for 178 Pareiasauridae (Lee 1994, 1995, 1996, 1997a, 1997b). 179 Based on crania alone, Van den Brandt (2016) conducted a taxonomic reassessment of the 180 Tapinocephalus AZ pareiasaurs and agreed with the four taxa put forward by Lee as valid 181 (1994, 1997a). However, he concluded that Bradysaurus seeleyi is the least secure member 182 of the group because of its lack of cranial autapomorphies and distinguishing characters 183 apart from small cheeks. Based on our anatomical study of crania and postcrania, Van den 184 Brandt (2020) and this paper propose synonymising Bradysaurus seeleyi with Bradysaurus 185 baini, resulting in three valid mid-Permian pareiasaur taxa: Bradysaurus baini, 186 Embrithosaurus schwarzi and Nochelesaurus alexanderi. 187 188 Material and Methods 189 Supplemental material comprises six appendices: Supplemental material, Appendix 1A-C, 190 contains character modifications, character scores and character descriptions; synonym 191 justifications and Bradysauria indet. specimens; and 240 pareiasaur specimen identifications 192 with justifications. Supplemental material, Appendix 2, contains the updated data matrix 193 (tnt) file; Supplemental material, Appendix 3, cranial and postcranial variability in middle 194 Permian species of Pareiasauria, was used to inform the Diagnoses, which was developed 195 from various sources; Supplemental material, Appendix 4, locality and stratigraphic data for 196 240 pareiasaur specimens; Supplemental material, Appendix 5, a specimen flow chart, 197 showing a summary of previous taxonomic revisions of middle Permian South African 198 pareiasaurs and the changes proposed in this study; and Supplemental material, Appendix 9 199 6, phylogenetic strict and majority consensus trees with synapomorphies identified for each 200 monophyletic group of pareiasaurians. 201 Specimen identification methods 202 Many authors have recognised that pareiasaur remains from the Tapinocephalus AZ are 203 often distorted, altering cranial proportions and complicating taxonomic assessments. As 204 such our identification criteria focus on the absence or presence of discrete dental, cranial, 205 and post-cranial features. 206 For biostratigraphic purposes, our goal was to identify as many Guadalupian South African 207 pareiasaur specimens as possible to species level. This was not always possible due to 208 deformation, ontogenetic stage of development, distortion, and the degree of preservation 209 of the fossils but we were able to identify sixty-five percent or 157 out of 240 specimens 210 (Supplemental material, Appendix 1C: numbers 1 to 157, Tables 2, 3). Our justifications for 211 every specimen identified are provided in Supplemental material, Appendix 1C: numbers 1 212 to 157. 213 The holotypes of Bradysaurus baini, Embrithosaurus schwarzi and Nochelesaurus alexanderi 214 (Figs. 2, 3) reveal obvious autapomorphies and morphological differences, such as the shape 215 of the pelvis of Embrithosaurus schwarzi and the ventromedially oriented (lingually- 216 pointing) upper teeth of Bradysaurus baini. However, in accessing a large sample set of over 217 200 pareiasaurs, many comprising relatively complete skeletons, we found that the discrete 218 diagnostic features or combination of features of the holotypes are sometimes not upheld 219 in particular specimens and some specimens may show features of more than one holotype. 220 When a specimen exhibits a combination of characters from more than one species, we 10 221 have (in some instances) been able to narrow its identification to one of two species 222 (Supplemental material, Appendix 1C: numbers 158 to 180). For example, in specimens 223 FMNH UC 1533 and BP/1/7165, the skull points to a particular species and the postcranium 224 to another. We have also noted specimens as unidentifiable, for example, specimen SAM- 225 PK-9168 has features of Bradysaurus, Nochelesaurus and Embrithosaurus (Supplemental 226 material, Appendix 1C: numbers 181 to 240). It is important to note that most of the 23 227 specimens identified to one of two species and the 60 unidentifiable specimens are poorly 228 preserved, poorly prepared, and as such often have no diagnostic parts preserved. 229 To avoid the scenario whereby a diagnostic model cannot securely identify a specimen, early 230 workers created several species, each with only slightly different features, which led to an 231 inflated number of species (eg. Haughton & Boonstra 1929). To avoid an inflated number of 232 very similar yet slightly different species, we had to: 233 • prioritise diagnostic features that represent major anatomical differences that are 234 less susceptible to deformation: orientation of the upper jaw teeth; shapes of the 235 premaxilla and maxilla, postfrontal, and postparietal; morphology of the posterior 236 cheek bosses and the overall cranial ornamentation; shape of the pelvis; features of 237 the femur; twisting of the humerus; and ornamentation of the osteoderms; 238 • identify the species by the majority of the features represented in a specimen when 239 that specimen exhibits traits of more than one species; 240 • broaden the taxonomic definition of features or measurements within each species 241 by increasing the intraspecific variability (e.g. Bradysaurus: cheek size varies from 242 small to large, posterior cheek bosses large or medium-sized; Embrithosaurus: tooth 11 243 cusp numbers seven to nine; Nochelesaurus: lateral condyle of the femur extends 244 distally moderately to far beyond the medial condyle). 245 Phylogenetic methods 246 The character matrix of Van den Brandt et al. (2020) was used for initial cranial character 247 scoring (characters 1-78) and the matrix of Liu & Bever (2018) for the initial postcranial 248 character scoring (characters 79-139). 249 We scored the cranial and postcranial characters for all valid middle Permian South African 250 species based on direct studies of the holotypes and the best-preserved referred specimens. 251 We reworded character statements 14, 31, 49 and 102 to improve clarity (Supplemental 252 material, Appendix 1A). For character 69, state (1) (teeth with 2-7 cusps) was no longer 253 applicable and the character was modified to: state (0) conical, single cusp; state (1) 7-9 254 cusps on each maxillary tooth; state (2) 9-11 cusps; and state (3) more than 11 cusps. The 255 states of all taxa were updated accordingly. For character 74, we swapped the definitions of 256 state (0) and (1), as the previous state definition did not match the data (all known 257 pareiasaurs have teeth on the transverse flange of the pterygoid). 258 Characters 2, 3, 11, 117, and 129 were excluded from the analysis as they are 259 uninformative. Bradysaurus seeleyi (because of its synonymy with B. baini) was removed 260 from the taxon list resulting in 21 pareiasaurian species and eight non-pareiasaur 261 parareptilian outgroups. All of the modified statements and conditional states for characters 262 44, 48, 50, 51 and 68, introduced by Van den Brandt et al. (2020), were retained. A 263 significant number of changes in the scoring were introduced given our redescriptions of the 264 material (see Van den Brandt 2016, 2020; Van den Brandt et al. 2020, 2021a, 2021b): 12 265 changes in scoring for Embrithosaurus were introduced in characters 54, 76, 79, 80, 92, 99, 266 113, 115, 124, 136, 139; for Bradysaurus in characters 10, 11, 12, 38, 39, 44, 54, 69, 70, 71, 267 79, 80, 92, 98, 102, 107, 111, 113, 115, 125; and for Nochelesaurus in characters 3, 4, 5, 10, 268 11, 12, 14, 19, 23, 33, 42, 43, 44, 54, 64, 68, 69, 70, 71, 72, 76, 79, 81, 82, 83, 88, 99, 102, 269 103, 107, 108, 110, 112, 114, 116, 117, 122, 127, 135, and 136 (Supplemental material, 270 Appendix 1A, 2). For character 54, we changed the scores from (1) to (0) for all 271 pareiasaurian taxa, except for Shihtienfenia, Sanchuansaurus, Honania and Obirkovia which 272 we left unchanged as ?. 273 The program TNT (Tree Analysis Using New Technology) version 1.5 was used to find the 274 most parsimonious trees (Goloboff et al. 2008; Goloboff & Catalano 2016). Considering the 275 relatively large size of the data matrix (29 taxa and 139 characters) the search strategy for 276 the most parsimonious trees (mpt) entailed a heuristic search of 10 random addition 277 sequences with TBR (tree bisection reconnection), saving 10 trees per replication, and an 278 additional search using the trees from RAM as the starting point and implementing TBR on 279 those trees. The default setting for branch collapsing in TNT, rule 1 (minimum length = 0), 280 which collapses branches if the support is ambiguous, was used (Goloboff et al. 2008). 281 Millerettidae was selected as the root and all characters were run as unordered and equally 282 weighted. 283 Stratigraphic methods 284 Using the updated diagnoses provided below, we examined 240 pareiasaur specimens for 285 re-identification (Tables 2, 3). Where provenance information was available, the 286 stratigraphic position of individual specimens was determined using a dataset of fossil 287 occurrences within the Abrahamskraal Formation and lower Teekloof Formation at the 13 288 resolution of approximately 50 m bins, first produced by Day (2013) and most recently used 289 by Day et al. (2018) (Supplemental material, Appendix 4). The total stratigraphic range for 290 each species was determined from those occurrences. For each valid species, the 291 stratigraphic ranges presented by Day (2013), Van den Brandt (2016), and this study are 292 presented and discussed (Fig. 5). All middle Permian pareiasaur specimens were recovered 293 west of 24° East longitude, where the Abrahamskraal Formation comprises up to six 294 lithostratigraphic members (Day & Rubidge 2014; Cole et al. 2016). 295 Institutional abbreviations 296 AM, Albany Museum, Makhanda (Grahamstown); AMNH, American Museum of Natural 297 History, New York; BP, Evolutionary Studies Institute, University of the Witwatersrand, 298 Johannesburg (formerly Bernard Price Institute for Palaeontological Research); CGP, Council 299 for Geosciences, Pretoria; FMNH, Field Museum of Natural History, Chicago; GPIT, 300 Paläontologische Sammlung, Eberhard-Karls-Universität Tübingen, Germany; NHMUK, 301 Natural History Museum, London, United Kingdom; NMQR, National Museum, 302 Bloemfontein; PIN, Paleontological Institute, Moscow, Russia; SAM, Iziko South African 303 Museum, Cape Town. 304 Results 305 Systematic Palaeontology 306 307 Superorder Reptilia Laurenti, 1768 308 Order Parareptilia Olson, 1947 14 309 Suborder Pareiasauria Seeley, 1888 310 Pareiasauridae Cope, 1896 311 Bradysauria Van den Brandt, Abdala, Rubidge, 2020 312 Definition 313 The clade that includes Bradysaurus baini, Embrithosaurus schwarzi, Nochelesaurus 314 alexanderi, their common ancestor, and its descendants. 315 Diagnosis 316 Clade of large-bodied pareiasaurs diagnosed by five synapomorphies: (1) character 21, 317 presence of regular radiating ridges on the skull roof from the centre of the dermal skull 318 roof bones (also the condition in Scutosaurus and Provelosaurus); (2) character 39, absence 319 of supernumerary bones on the skull (unambiguous within Pareiasauria, also the condition 320 in Owenetta); (3) character 77, medial rows of palatal denticles widely separated, 321 converging anteriorly (unambiguous); (4) character 79, pleurocentrum of the atlas 322 separated from the axial intercentrum (unambiguous); and (5) character 115, notched 323 anterior acetabulum (also the condition in Pareiasuchus nasicornis and P. peringueyi). 324 The Bradysauria are the only pareiasaurs to exhibit: a basisphenoid body that is wide and 325 strongly constricted, giving it an hourglass shape in ventral view (character 13); width of the 326 humerus ulnar fossa (intercondylar depression) much wider than the olecranon process 327 (character 104); and the absence of a central boss or boss complex on the lateral squamosal 328 (Van den Brandt personal observation, not a listed as a phylogenetic character). 15 329 The Bradysauria differ from other pareiasaurs by lacking bosses on the ventral margin of the 330 quadratojugal, except for Parasaurus (character 48); their cheek ornamentation that 331 consists of low, rounded bosses (except for Nochelesaurus which has more distinct, taller, 332 more pointed bosses) (character 50); and their irregularly spaced tooth cusps (character 66) 333 (Haughton & Boonstra 1929, Boonstra 1934, Lee 1997a). 334 Genus Bradysaurus Watson, 1914 335 Species Bradysaurus baini Seeley, 1892 336 (Figs 2A, 2B, 3D-F) 337 Synonyms 338 1892 Pareiasaurus baini Seeley: 311, pls. 17-23 339 1913 Pareiasaurus acutirostris Broom: 397. 340 1914: Bradysaurus baini Watson: 101. 341 1929 Bradysaurus vanderbyli Haughton and Boonstra: 82. 342 1929 Bradysaurus seeleyi Haughton and Boonstra: 82. 343 1929 Koalemasaurus acutirostris Haughton and Boonstra: 84. 344 1929 Platyoropha broomi Haughton and Boonstra: 84. 345 1935 Bradysaurus bombidens Broom: 40, figs 2, 4. 346 1935 Bradysaurus acutirostris Broom: 43, fig. 3. 347 1969 Bradysaurus broomi Kuhn: 75. 16 348 Holotype. NHMUK PV R 1971, complete skull and separate lower jaw with very complete 349 postcranium including a complete shoulder girdle and pelvis, nearly complete vertebral 350 column, missing only the end of the caudal vertebrae, complete right fore- and hind-limbs, 351 left humerus, radius, ulna, femur, tibia, and fibula partially preserved, both 352 astragalocalcanea, and a few ribs. 353 Diagnosis. 354 Autapomorphies. Bradysaurus baini has one cranial autapomorphy (modified from 355 Lee 1994, 1997a): Distal paroccipital process of the opisthotic anteroposteriorly long 356 with a large, dorsomedially elongated, posteriorly projecting, knob-like process on 357 the medial edge of the posterodorsal corner of the distal paroccipital process. 358 Bradysaurus has one postcranial autapomorphy: anterior dorsal osteoderms smooth 359 and strongly convex, with an incipient central boss, and very light ornamentation 360 (character 135). 361 Unique combinations of features. Bradysaurus differs from both Embrithosaurus 362 and Nochelesaurus in having the following combination of characters: maxilla with a 363 large, wide and rugose posterodorsal maxillary boss, lower in juveniles (character 364 31); mediolaterally wide and curved ventral premaxilla and maxilla (up to 20 mm 365 wider than the tooth row), accommodating horizontally to ventromedially oriented, 366 lingually-pointing upper jaw teeth, that are not visible in lateral view (character 67); 367 posterodorsal maxillary-lacrimal ridge is wide, low, swollen, bulbous and continuous; 368 maxillary teeth with seven to nine marginal cusps; 20 upper jaw teeth (character 68); 369 mandibular teeth with seven to nine marginal cusps; most dorsoventrally elongated 370 and mesiodistally narrow, a minority short and wide, all with a central three-cusped 17 371 trident, symmetrical or non-symmetrical mesial and distal marginal cusp 372 arrangements, without horizontal cingulum, some mandibular teeth with medial 373 cusps; lingual surface of mandibular teeth bulged or with vertical ridges (character 374 71); relatively small angular boss; frontal central boss large, wide and low, or high 375 and distinct; ventrolateral quadratojugal swollen, bulging and not flat, covered with 376 irregular rugose ridges, and sometimes bulged to form a shelf; corner cheek boss 377 either large, low, wide and bulbous, or medium-sized, round, as wide as high; wide 378 or narrow cheek flanges; cheek flange highly variable, extending between 20° and 379 30° below the tooth row (character 44); smallest cheek flange amongst co-occurring 380 pareiasaurs (extending only 20° below the tooth row); bosses on the posterior 381 margin of the quadratojugal either large, low, bulbous, wider than high or medium- 382 sized, round, as wide as high; epicondyles of the humerus extend distally far beyond 383 the radial and ulnar articulation surfaces, appearing forked (character 98); wider 384 proximal and distal humerus; proximal internal trochanter of the femur high, wide 385 and strongly curved, forming a semi-circular structure, with a perpendicular proximal 386 edge; distal internal trochanter with a short anterior (preaxial) process/projection; 387 parapophysial and diapophysial processes fuse into a single process, creating a single 388 vertical lateral facet from the 4th or 5th vertebra posteriorly; second sacral rib is the 389 largest; posterior dorsal osteoderms flattened, low, thin and disc-like, most with 390 shallow circular central hollow and very light ornamentation. 391 Differential diagnosis. Bradysaurus differs from Embrithosaurus in having: all upper 392 jaw teeth dorsoventrally elongated, mesiodistally narrow, with a three-cusped 393 central trident, and irregularly arranged marginal cusps; postfrontal roughly square, 18 394 slightly longitudinally elongated, and posteriorly curved medially; low and wide 395 posteroventral process on the clavicle, lacking a distinct notch just medial to the 396 process; humerus with deep intercondylar fossa; delto-pectoral crest large, robust, 397 and ‘heart-shaped’, with a notch on the ventrodistal corner; short, crescent-shaped, 398 insertion point for muscle attachment on the humeral shaft of the dorsal proximal 399 humerus; presence of a distinct, vertical ridge on the medial radius; pelvis low and 400 wide; posterior iliac blade vertical, oriented parallel to the sagittal plane, and with 401 dorsal margins medially curved and convex a short distance; anterior Iliac blades 402 deeply concave laterally, upwardly bent (everted) 55°-60° (character 111) to present 403 a large surface in dorsal view, strongly curved with the anterior process tip pointing 404 45° off the sagittal plane and extending out far laterally from the main body of the 405 blade; Iliac shaft oriented anterodorsally, approximately 60° off the vertical plane 406 (character 113); puboischiatic plate wide, long, dorsally and ventrally flat; pelvis 407 symphysis dorsoventrally low and anteroposteriorly long (character 118); ventral 408 keel on the puboischiatic plate low and narrow and extends obliquely at a shallow 409 angle; median pubic process small, extending a short distance ventrally, with sharp 410 square edges; major trochanter posterior projection short; small 40° angle between 411 the lateral and medial facets of the proximal articulation surface of the tibia; distal 412 tibia ventromedially expanded and wide; wide tibial shaft; caudal vertebrae with L- 413 shaped lateral projections (character 85). 414 Bradysaurus differs from Nochelesaurus in having: maxillary teeth with symmetrical 415 mesial and distal marginal cusp arrangement; maxillary teeth lingual surface with a 416 low, smooth bulge; snout wider than high; transversely narrow postparietal; cranial 19 417 ornamentation comprising low, indistinct bosses (especially on the prefrontal, 418 postfrontal and postorbital); absence of gap (vertical separation) on the posterior 419 edge of the cheek between the dorsal-most quadratojugal boss and the large 420 squamosal boss on most specimens; short scapula blade, mid scapula blade 421 anteroposteriorly wide, distal portion of scapula blade flared; preaxial and postaxial 422 margins of distal portion of scapula blade curved; medial process of scapula blade 423 low and wide; proximal and distal expansions of the humerus twisted 60°-80° apart 424 and almost perpendicular; symmetrical distal and proximal radius; postaxial shaft of 425 femur notable bulged/convex; lateral condyle of the femur extends distally 426 moderately (20-45 mm) beyond the medial condyle; proximal and distal notches of 427 the major trochanter of the femur. 428 Derivation of name. Brady, slow, heavy (Greek bradys); saurus, lizard (Latin); baini, after 429 Mr. Thomas Bain (Seeley 1892). 430 Referred material. Seventy-four specimens listed in Supplemental material, Appendix 1C: 431 numbers 84 to 157. 432 Locality, stratigraphy and age. NHMUK PV R 1971 was collected on 12 August 1889 by 433 Seeley, T. Bain, J.S. Marais, and S. Marais from the farm De Bad in the Beaufort West district 434 of the Western Cape Province (Seeley 1892). Diictodon – Styracocephalus Subzone of the 435 Tapinocephalus AZ (Day & Rubidge 2020), lower to mid Moordenaars Member of the 436 Abrahamskraal Formation. 437 Remarks. Regarding Lee’s (1994, 1997a) two other cranial autapomorphies proposed for 438 this taxon: 1) the huge, rounded lump on the maxilla (i.e. a maxillary boss) is shared with 20 439 Embrithosaurus, Sanchuansaurus, Pareiasaurus, Scutosaurus, and E. mirabilis (character 31), 440 although it is very large in Bradysaurus baini, and; 2) the ventromedially oriented, lingually- 441 pointing, upper jaw teeth (character 67) are shared with Pareiasuchus nasicornis and 442 Sanchuansaurus. Since a maxillary boss and ventromedially oriented upper jaw teeth occur 443 in more than one pareiasaur species, these two features are removed as autapomorphies of 444 Bradysaurus baini, but are included in the comparative diagnosis of the taxon. 445 <> 446 <> 447 Genus Embrithosaurus Watson, 1914 448 Species Embrithosaurus schwarzi Watson, 1914 449 (Figs 2C, 2D, 3A-C) 450 Synonyms 451 1903 Pareiasaurus serridens Broom: 123, pl. 15, figs 1-24, pl. 16. 452 1914 Pareiasaurus whaitsi Broom: 138, fig. 453 1914 Embrithosaurus schwarzi Watson: 101. 454 1929 Bradysuchus whaitsi Haughton & Boonstra: 82. 455 1935 Bradysaurus whaitsi Broom: 43, figs 3, 5. 456 Holotype. SAM-PK-8034, a distorted skull with occluded lower jaw, vertebrae 1 to 27 with 457 articulated osteoderms, two small articulated caudal vertebrae, partial left scapulocoracoid, 458 cleithrum, both clavicles, and interclavicle, complete right humerus, partial left humerus, 21 459 complete right radius, partial left radius, both ulnae, both femora, both tibiae, right fibula, 460 partial left fibula, complete pelvis, two digits of the forelimb, two digits of the hindlimb, and 461 several small fragments of vertebrae, ribs and phalanges labeled as “doubtful association”. 462 Diagnosis. 463 Autapomorphies. Embrithosaurus schwarzi has three unambiguous pelvic 464 autapomorphies (modified from Lee 1994, 1997a): 1) anterior iliac blade flat and 465 vertical/dorsoventral (not everted or upturned) (character 111); 2) iliac blades 466 diverge anteriorly, oriented 45°-60° off the sagittal plane, and; 3) pelvic symphysis 467 very thick dorsoventrally (character 119). 468 Unique combinations of features. Embrithosaurus differs from both Bradysaurus 469 and Nochelesaurus in having all of the following: maxillary teeth with seven or nine 470 marginal cusps; nine-cusped maxillary teeth are dorsoventrally short, mesiodistally 471 wide, lacking an elongated, central, three-cusped trident, and with marginal cusps 472 arranged more regularly around the tooth crown than seven-cusped teeth; 473 mandibular teeth with seven or nine marginal cusps; small, distinct maxillary boss 474 (character 31); posterodorsal maxillary-lacrimal ridge narrow and formed by a small 475 distinct maxillary boss and several small and distinct lacrimal bosses; prefrontal with 476 a central, shallow, wide depression on some specimens; frontal central boss very 477 small and low; postfrontal mediolaterally elongated (internal sutures); largest cheek 478 flange in the group of co-occurring taxa, extending 30°-40° below the tooth row 479 (character 44); large, low, indistinct bulbous bosses, wider than high, on the 480 posterior edge of the quadratojugal; corner cheek boss large, low, wide and bulbous, 481 distal paroccipital process of the opisthotic anteroposteriorly short with a small, 22 482 dorsomedially elongated, posteriorly projecting, knob-like process on the medial 483 edge of its posterodorsal corner; high and wide posteroventral process on the 484 clavicle and presence of a deep notch medial to the posteroventral process in most 485 specimens; humerus with shallow intercondylar fossa (character 103); delto-pectoral 486 crest small, anteroposteriorly narrow, and with a long, shallow groove on the ventral 487 surface; shaft of the dorsal proximal humerus lacking a short, curved muscular ridge; 488 major trochanter projects far posteriorly (character 122); proximal internal 489 trochanter high, narrow and gently curved, forming a s-shaped structure, with a 490 perpendicular proximal edge; distal internal trochanter with an anterior process; 491 large 60°angle between the medial and lateral facets of the proximal articulation 492 surface of the tibia; narrow tibial shaft; distal tibia narrow and symmetrical; anterior 493 and posterior osteoderms with distinct, rugose, irregular and tab-like (L or E shaped) 494 central boss, and a thin and flat periphery with radial ridges (character 136). 495 Differential diagnosis. Embrithosaurus differs from Bradysaurus in having: vertical 496 and narrow premaxilla and maxilla (< 5 mm wider than the tooth row), 497 accommodating vertically oriented upper jaw teeth, visible in lateral view; 18 upper 498 jaw teeth (character 68); ventrolateral quadratojugal flat and with distinct high, 499 circular, ovoid, bosses arranged posterodorsally; relatively large angular boss; 500 epicondyles of the humerus do not extend distally far beyond the radial and ulnar 501 articulation surfaces; narrower proximal and distal humerus; pelvis high and narrow; 502 Iliac blades straight and laterally flat to very slightly shallowly concave; extreme 503 anteroventral margin of the anterior process of the iliac blade is very slightly curved 504 and bent upwards and laterally outwards; Iliac shaft oriented anterodorsally 23 505 approximately 30° off the vertical plane (character 113); puboischiatic plate narrow 506 and short; with a strongly and evenly concavely curved (not flat) dorsal surface; 507 pelvic symphysis anteroposteriorly short (character 118); keel on ventral 508 puboischiatic plate high and wide, with steep 50° oblique lateral sides; median pubic 509 process large, tongue shaped with rounded edges, extending far anteroventrally; 510 parapophysial and diapophysial processes fuse into a single process, creating a single 511 vertical lateral facet from the 6th vertebra posteriorly; first sacral rib is the largest; 512 caudal vertebrae with straight lateral projections (character 85). 513 Embrithosaurus differs from Nochelesaurus in: maxillary teeth with symmetrical 514 mesial and distal marginal cusp arrangement; all mandibular teeth dorsoventrally 515 short and mesiodistally wide, lacking a central three-cusped trident, with 516 symmetrical mesial and distal marginal cusp arrangements, with a bulged lingual 517 surface lacking a cingulum and medial cusps; snout wider than high; transversely 518 narrow postparietal; cranial ornamentation comprising low, indistinct bosses 519 (especially on the prefrontal, postfrontal and postorbital); wide cheek flanges; 520 absence of gap (vertical separation) on the posterior edge of the cheek between the 521 dorsal-most quadratojugal boss and the large squamosal boss; short scapula blade; 522 mid scapula blade anteroposteriorly wide, distal portion of scapula blade flared; 523 preaxial and postaxial margins of distal portion of scapula blade curved; medial 524 process of scapula blade low and wide; proximal and distal expansions of the 525 humerus twisted 60°to 80° apart and almost perpendicular; symmetrical distal and 526 proximal radius; postaxial shaft of femur notably bulged/convex; lateral condyle of 24 527 the femur extends distally moderately (30-45 mm) beyond the medial condyle; 528 proximal and distal notches of the major trochanter of the femur. 529 Derivation of name. Embritho, weighty (Greek); saurus, lizard (Latin); schwarzi, after Mr E. 530 H. L. Schwarz (Broom 1903). 531 Referred material. Forty specimens listed in Supplemental material, Appendix 1C: numbers 532 1 to 40. 533 Locality, stratigraphy and age. According to Broom (1903), SAM-PK-8034 was collected by 534 Mr A.W. Rogers and Mr E.H.L. Schwarz in 1902 on the farm Hoogeveld Lot A (Hoogeveld 535 270), in the Gouph region, near Knoflock’s Fontein, Van der Byl’s Kraal, in the Prince Albert 536 district of the Western Cape Province (Broom 1903). Diictodon – Styracocephalus Subzone 537 of the Tapinocephalus AZ (Day & Rubidge 2020), lower Moordenaars Member of the 538 Abrahamskraal Formation. 539 540 Genus Nochelesaurus Haughton and Boonstra, 1929 541 Species Nochelesaurus alexanderi Haughton and Boonstra, 1929 542 (Fig. 2E-F) 543 Synonyms 544 1912 Propappus rogersi Broom: 323, pl. 19, figs 1-11, pl. 20 , figs 12-21, pl. 21, figs 24-27. 545 1929 Nochelesaurus alexanderi Haughton and Boonstra: 83. 546 1929 Dolichopareia angusta Haughton and Boonstra: 83. 25 547 1929 Brachypareia watsoni Haughton and Boonstra: 84. 548 1929 Brachypareia rogersi Haughton and Boonstra: 84. 549 1969 Bradysaurus angusta Boonstra: 30. 550 1969 Embrithosaurus alexandri Kuhn: 80, pl. 39, fig. 7d, pl. 41, fig 11, pl. 44, fig. 3. 551 1969 Embrithosaurus angusta Kuhn: 80, pl. 38, figs 2, 6-8, pl. 39, figs 2, 6 7f, pl. 44, fig. 5. 552 1969 Bradysaurus watsoni Kuhn: 75, pl. 39, fig. 7h, pl. 41, fig. 12 553 1969 Bradysaurus rogersi Kuhn: 75, pl. 39, fig. 7g, pl. 40, figs 21-30, pl. 41, fig. 8. 554 Holotype. SAM-PK-6239, dorsal and right side of the skull and fragments of the right lower 555 jaw, braincase and palate, a partial right scapulocoracoid, cleithrum and clavicle, partial 556 interclavicle, damaged left humerus, seven articulated osteoderms, complete right femur 557 and tibia, fibula fragment, astragalocalcaneum, and several unnumbered, unidentified 558 fragments. 559 Diagnosis. 560 Autapomorphies. Nochelesaurus alexanderi has one cranial autapomorphy: 561 transversely wide postparietal, twice as wide as long. 562 Unique combinations of features. Nochelesaurus differs from both Bradysaurus and 563 Embrithosaurus in having: maxillary teeth with five or seven marginal cusps, creating 564 a symmetrical cusp arrangement or an 8th, low, small, basal cusp on the mesial 565 margin, creating a non-symmetrical cusp arrangement; maxillary teeth with three to 566 five vertical ridges on the lingual surface; up to ten cusps on mandibular teeth; 26 567 mandibular teeth with non- symmetrical mesial and distal cusp arrangements due to 568 extra, low, very small, mesial marginal cusp on certain teeth; low and narrow 569 maxillary-lacrimal ridge, has a sharp lateral edge on the maxilla and does not reach 570 the small, distinct lacrimal bosses; relatively narrow snout, as high as wide or very 571 slightly wider than high; maxilla without a maxillary boss; cranial ornamentation 572 comprising distinct, high, pointed bosses (especially on the prefrontal, postfrontal 573 and postorbital); central frontal boss large, high and distinct; narrowest cheek 574 flange; corner cheek boss large, elongated, pointed and distinct, with a narrow neck 575 proximally, a raised distal rim, and covered in rugose ridges; posterior edge of the 576 quadratojugal with distinct bosses that vary from horizontally flat, pointed and plate- 577 like to medium-sized, round, as wide as high; flat, vertical separation (gap) on the 578 posterior edge of the cheek, between the dorsal-most quadratojugal boss and the 579 large squamosal boss, through which the horizontal squamosal-quadratojugal suture 580 passes; medial edge of the posterodorsal corner of the distal paroccipital process 581 low and gentle convexly swollen; posteroventral edge of the distal paroccipital 582 process is dorsolaterally elongated and swollen in some specimens; tubercles 583 present on the ventral surface of the basisphenoid/ parabasisphenoid, just behind 584 the basipterygoid processes; most specimens with a long and narrow scapula blade; 585 mid scapula blade anteroposteriorly constricted and very narrow; distal portion of 586 scapula blade narrow with little anteroposterior flaring, preaxial and postaxial 587 margins of distal portion of scapula blade straight and only very slightly curved in 588 most specimens, expanding gradually (character 88); medial process of scapula blade 589 high; proximal and distal expansions of the humerus very flat, only 20°-40° apart and 590 almost on the same plane (character 92); tubercle sometimes present on the 27 591 postaxial margin of the intercondylar fossa of the humerus; ventrolaterally expanded 592 distal radius; postaxial shaft of femur straight (not notably bulged/convex); lateral 593 condyle of the femur extends distally far (70 mm) beyond the medial condyle in most 594 specimens; major trochanter posterior projection very short; absence of or very 595 shallow proximal and distal notches of the major trochanter of the femur; proximal 596 internal trochanter low and narrow, straight or very slightly curved, with an oblique 597 proximal edge (character 125); distal internal trochanter without an anterior 598 process; osteoderms with a wide central boss covering most of the scute, round or 599 comprising five or six rugose, flattened, radial, tab-like processes, with a high and 600 thick periphery. 601 Differential diagnosis. Nochelesaurus differs from Bradysaurus in having: vertical 602 and narrow premaxilla and maxilla (< 5 mm wider than the tooth row), 603 accommodating vertically oriented upper jaw teeth, visible in lateral view; 18 upper 604 jaw teeth (character 68); large cheek flanges extending approximately 30° below the 605 tooth row; ventrolateral quadratojugal flat and with distinct high, circular to ovoid, 606 bosses arranged posterodorsally; relatively large angular boss; epicondyles of the 607 humerus do not extend distally far beyond the radial and ulnar articulation surfaces 608 in most specimens; narrower proximal and distal humerus; parapophysial and 609 diapophysial processes fuse into a single process, creating a single vertical lateral 610 facet from the 6th vertebra posteriorly. 611 Nochelesaurus differs from Embrithosaurus in having: postfrontal roughly square, 612 slightly longitudinally elongated, and posteriorly curved medially; all upper jaw teeth 613 dorsoventrally elongated, mesiodistally narrow, with a three-cusped central trident, 28 614 and irregularly arranged marginal cusps; all mandibular teeth dorsoventrally 615 elongated and mesiodistally narrow, with a central three-cusped trident, with non- 616 symmetrical mesial and distal marginal cusp arrangements, with an incipient 617 horizontal cingulum carrying medially (lingually) placed cusp(s) on the lingual surface 618 of certain mandibular teeth; lingual surface of mandibular teeth with one to three 619 distinct vertical ridges; narrow cheek flanges; low and wide posteroventral process 620 on the clavicle, and lacking a distinct notch just medial to the process; humerus with 621 deep intercondylar fossa; delto-pectoral crest large, robust, and ‘heart-shaped’, with 622 a notch on the ventrodistal corner; short, crescent-shaped, insertion point for 623 muscle attachment on the humeral shaft of the dorsal proximal humerus; presence 624 of a distinct, vertical ridge on the medial radius; small 40° angle between the lateral 625 and medial facets of the proximal articulation surface of the tibia; distal tibia 626 ventromedially expanded and wide; wide tibial shaft. 627 Derivation of name. Nochele, sluggish, dull (Ancient Greek νωχελής) (Latinized nocheles); 628 saurus, lizard (Latin); alexanderi, after Dr. Alexander du Toit (Haughton & Boonstra 1929). 629 Referred material. Forty-three specimens listed in Supplemental material, Appendix 1C: 630 numbers 41 to 83. 631 Locality, stratigraphy and age. SAM-PK-6239 was collected in 1923 from the farm 632 Boesmanskop (Bushmans Kop 302), Beaufort West district of the Western Cape Province 633 (Haughton & Boonstra 1929). This locality is high in the Diictodon – Styracocepahlus 634 Subzone of the Tapinocephalus AZ (Day & Rubidge 2020), Karelskraal Member of the 635 Abrahamskraal Formation. 29 636 Remarks. Concerning Lee’s (1994, 1997a) three postcranial autapomorphies proposed for 637 Nochelesaurus alexanderi: 1) the subscapular fossa is not located close to the anterior 638 margin of the scapula blade and is removed as an autapomorphy; 2) the tubercle on the 639 humerus is manifestly present in only one specimen of Nochelesaurus, and it is removed as 640 an autapomorphy and considered an individual variation within the genus, and 3) the flange 641 on the dorsal femur that projects distally far beyond the postaxial tibial facet does not 642 consist of an obvious flange, instead, the distal extension of the lateral condyle of the femur 643 is furthest in most, but not all specimens of Nochelesaurus, and is therefore removed as an 644 autapomorphy and considered individual variation within the genus. 645 Bradysauria indet. 646 1876 Pareiasaurus bombidens Owen: 9, pl. 8, fig. 3, pl. 9. 647 1924 Pareiasaurus strubeni Broom: 507, figs 5, 6. 648 1929 Nochelesaurus strubeni Haughton and Boonstra: 83. 649 1935 Bradysaurus strubeni Broom: 45, fig. 5. 650 1969 Embrithosaurus strubeni Kuhn: 80, pl. 39, fig. 7e, pl. 43, fig. 4. 651 652 Phylogeny 653 The TNT analyses found 1065 most parsimonious trees (mpt) of length 263 (Retention Index 654 RI = 0.783, Consistency Index CI = 0.615). The strict consensus tree (Fig. 4A) shows 655 Pareiasauria recovered as a monophyletic group, comprising an extended basal polytomy of 656 twelve terminal taxa and three monophyletic groups: Pareiasuchus nasicornis and P. 30 657 peringueyi, the middle Permian pareiasaurs from the Karoo (Embrithosaurus + 658 Nochelesaurus + Bradysaurus) and the dwarf South African pareisaurs (Nanopareia + 659 Pumiliopareia + Anthodon) with Provelosaurus, in which Nanopareia and Provelosaurus are 660 in a polytomy, and Pumiliopareia and Anthodon form a clade. Synapomorphies listed for the 661 groups below are after unambiguous optimisation only on the majority rule consensus tree 662 (Fig. 4B). Unambiguous optimization synapomorphies on the strict and majority rule 663 consensus are presented in Supplemental material, Appendix 6. 664 <> 665 In agreement with Van den Brandt et al. (2020), South African middle Permian pareiasaurs 666 (Embrithosaurus + Nochelesaurus + Bradysaurus) form a monophyletic Bradysauria 667 supported by five synapomorphies: character 21, presence of regular radiating ridges on the 668 skull roof from the centre of the dermal skull roof bones (also the condition in Scutosaurus 669 and Provelosaurus); character 39, absence of supernumerary bone on the skull 670 (unambiguous within pareiasaurians, also the condition in Owenetta); character 77, medial 671 rows of palatal denticles widely separated, converging anteriorly (unambiguous); character 672 79, pleurocentrum of the atlas separate from the axial intercentrum (unambiguous); and 673 character 115, notched anterior acetabulum (also the condition in in P. nasicornis and P. 674 peringueyi). The condition of character 79 is unknown in most pareiasaurs, but has now 675 been updated for the basal South African forms. The first three synapomorphies (characters 676 21, 39 and 77) were previously presented by Van den Brandt et al. (2020). 677 In Bradysauria, Nochelesaurus and Bradysaurus from a monophyletic group supported by 678 two synapomorphies: character 44, quadratojugal (cheek) flange small, extending 0°-40° 679 below the level of the maxillary tooth row; and character 71, lingual surface of the 31 680 mandibular teeth with a distinct, triangular ridge, narrowing towards the crown of the 681 tooth. The two elginid taxa also have small cheeks, and several other pareiasaurs have a 682 triangular ridge on the inner surface of the mandibular teeth (i.e. Shihtienfenia, 683 Pareiasuchus peringueyi, Pareiasuchus nasicornis, and Scutosaurus). 684 After the Bradysauria, Bunostegos and then Honania form an unnamed grade between the 685 basal pareiasaurs and the forms with more derived features (Fig. 4B). Bunostegos and all 686 remaining pareiasaurs are united by four synapomorphies: characters 13, 48, 50 and 66, 687 which is the same result found by Van den Brandt et al. (2020): character 13, wide, not 688 strongly laterally constricted basisphenoid; character 48, presence of bosses on the ventral 689 margin of the quadratojugal; character 50, cheek ornamentation present on the posterior or 690 ventral margins of the cheek in the form of well-developed, more distinct, taller, more 691 pointed bosses; and character 66, cusps regularly spaced along the tooth crown. In the 692 members of the Bradysauria the basisphenoids are wide and strongly laterally constricted; 693 bosses are not present on the ventral margin of the quadratojugal; cheek ornamentation 694 consists of low, rounded bosses (except for Nochelesaurus which has more distinct bosses); 695 and cusps are irregularly spaced along the tooth crown. These basal features of members of 696 the Bradysauria are well established in the literature (Haughton & Boonstra 1929; Boonstra 697 1934; Lee 1997a). 698 The clade formed by Honania and all remaining pareiasaurs (Fig. 4B) is united by four 699 synapomorphies: character 69, 9–11 or more cusps on the maxillary teeth; character 90, 700 dorsal edge of the posterior coracoid oriented posteroventrally forming an angle of more 701 than 135° with the posterior border of the scapula; character 91, absence of the cleithrum; 702 and character 92, the torsion of the humerus being less than or equal to 45°. The 32 703 Bradysauria and Bunostegos have fewer cusps (7-9) on the maxillary teeth, the posterior 704 coracoid is more horizontal, meeting the posterior border of the scapula at an angle of less 705 than 135°, the cleithrum is present, and the torsion of the humerus is greater than or equal 706 to 60° (except in Nochelesaurus). 707 Crownward to Honania there is a polytomy formed by three terminal taxa (Parasaurus, 708 Sanchuansaurus and Deltavjatia, Fig. 4B) and three monophyletic groups, all united by two 709 synapomorphies: characters 122, larger, more distinct major trochanter; and character 124, 710 femur with a narrow postaxial flange at the midpoint of the femur. 711 The Chinese Shihtienfenia forms a monophyletic group with Pareiasuchus from southern 712 Africa (P. nasicornis and P. peringueyi, Fig. 4B) all united by four synapomorphies: character 713 71, lingual surface of the mandibular teeth with a distinct, triangular ridge, narrowing 714 towards the crown of the tooth; character 87, length of the scapula blade greater than or 715 equal to three times the glenoid fossa diameter; character 95, narrow and rounded 716 postaxial flange of the entepicondyle; and character 111, strongly everted, almost horizontal 717 anterior margin of the ilium. Character 71 is also a synapomorphy uniting Nochelesaurus 718 and Bradysaurus (see above). Pareiasuchus nasicornis and P. peringueyi are united by five 719 synapomorphies: character 72, absence of horizontal cingulum on the lingual surface of the 720 marginal teeth; character 93, narrow and rounded postaxial flange of the ectepicondyle; 721 character 108, strong and sheet-like second and third sacral ribs; character 115, notched 722 anterior acetabulum; and character 120, strong preaxial expansion of the head of the femur. 723 The conditional states of characters 72 and 93 in this group are also present in the three 724 South African dwarf pareiasaurs. 33 725 The second monophyletic group in the majority rule and the strict consensuses (Fig. 4A, 4B) 726 comprises the three dwarf South African pareiasaurs (Nanopareia, Pumiliopareia, and 727 Anthodon), and the South American Provelosaurus. Pumiliopareia and Anthodon are 728 monophyletic and Nanopareia and Provelosaurus form a polytomy. The three South African 729 dwarf forms and Provelosaurus share ten synapomorphies: character 25, v-shaped anterior 730 interpterygoid vacuity; character 72, absence of a horizontal cingulum on the lingual surface 731 of the marginal teeth; character 92, torsion of the humerus less than or equal to 20°; 732 character 93, narrow and rounded preaxial flange of the ectepicondyle; character 97, 733 entepicondylar foramen exposed in ventral (flexor) view; character 100, radial condyle of 734 the humerus located terminally, encroaching onto the distal end of the humerus; character 735 101, ulnar articulation surface of the humerus forms a groove bordered posteriorly by a 736 prominent tubercle; character 110, no or only slight anterior expansion of the iliac blade; 737 character 113, shaft of the Ilium orientated vertical or posterodorsally; and character 136, 738 osteoderms ornamented with fine, straight, regularly spaced ridges radiating out from a 739 central boss to the edge. 740 Another clade (Fig. 4B) comprises Scutosaurus, Pareiasaurus, and a monophyletic Elginiidae 741 (Elginia wuyongae, Obirkovia, Arganaceras, and Elginia mirabilis) supported by five 742 synapomorphies: character 14, tubercles present on the ventral surface of each 743 basipterygoid process (basisphenoid), immediately posterior to the interpterygoid vacuity 744 and the transverse suture with the pterygoid; character 15, presence of a central boss on 745 the ventral surface of the basioccipital; character 31, prominent boss or horn present on the 746 maxilla immediately posterior to the external naris; character 33, snout as high as wide; and 747 character 111, strongly everted, almost horizontal anterior margin of the ilium. 34 748 The clade formed by the four elginids is supported by five synapomorphies (Fig. 4B): 749 character 13, basisphenoid narrow with relatively straight margin; character 35, presence of 750 a postfrontal ‘horn’; character 41, tabular (supernumerary element), contact, well- 751 developed, making contact posteriorly, excluding the postparietals from the posterior edge 752 of the skull table; character 50, ornamentation present on the posterior or ventral margins 753 of the cheek in the form of prominent, conical horns with sharp, pointed tips; and character 754 58, dermal bosses of cranial bones have a central long, pointed horn. 755 Resampling and support. Values of resampling for Pareiasauria varies from 74 with 756 jackknife (with 36 of removal possibility) to 91with symmetric resampling (with 33 of change 757 probability) and can be considered as well supported (Table 1). The situation with 758 Bradysauria is quite different as the support ranges from 54with poissonbootstrapping to 759 77using symmetric resampling. Bradysauria is therefore poorly supported and basically 760 remains with the same degree of support (bootstrapping of 58) previously reported by Van 761 den Brandt et al. (2020). 762 <> 763 Bremer support for Pareiasauria is 3 and for Bradysauria is 2, as well as for the clades 764 (Pareiasuchus nasicornis, Pareiasuchus peringueyi)and (Anthodon, Pumiliopareia). Other 765 pareiasaurian clades recovered in the strict consensus have a Bremer support of 1. The 766 relative Bremer support is 60 for Paraeiasauria and 43 for Bradysauria, suggesting again a 767 poor support of the latter group. 768 769 Discussion 35 770 771 Biostratigraphic discussion 772 The stratigraphic ranges of the mid-Permian pareiasaurs of South Africa’s Karoo Basin have 773 never been well established. Initially, when Seeley (1892) briefly described Karoo 774 stratigraphy as comprising five biozones, he considered that pareiasaurs occurred mostly in 775 one horizon that cropped out in the Koup Karoo and which he called the Zone of 776 Pareiasaurians. Though he did not venture to the area around Fort Beaufort, from where 777 the first known pareiasaurs were collected, Seeley assumed that two of these species 778 (Pareiasaurus serridens and Propappus omocratus; but not Anthodon) occurred in the same 779 zone. Later, Watson (1914) realised that the taxa from the Koup belonged to different 780 genera and that Pareiasaurus was only found in younger assemblages, leading him to 781 rename the Pareiasaurus zone as the Tapinocephalus zone. 782 Boonstra (1932a) briefly noted that the large-bodiedpareiasaurs first encountered very near 783 the base of the Tapinocephalus Zone underwent few changes and died out near the top of 784 the zone. Later, Boonstra (1969) suggested a decreasing abundance of pareiasaurs through 785 the Tapinocephalus Zone, with the majority of specimens (i.e. 189 out of 215) occurring in 786 his lower subdivision, and only one specimen in the upper (third) subdivision. 787 In their vertebrate biozonation of the Western Karoo Basin, Keyser & Smith (1979) proposed 788 renaming Boonstra’s lower and middle Tapinocephalus AZ together as the Dinocephalian 789 zone, for which both Bradysaurus bombidens and Embrithosaurus schwarzi were designated 790 as characteristic fossil taxa. Both genera were considered to also be present in the overlying 791 Pristerognathus/Diictodon AZ, though at much lower abundance. This was broadly 792 reiterated by Smith & Keyser (1995a), who suggested uncertain ranges for the genera 36 793 Bradysaurus and Embrithosaurus throughout the Tapinocephalus AZ and into the lower part 794 of the overlying renamed Pristerognathus AZ. 795 A more refined stratigraphic study of the Guadalupian part of the Beaufort Group by Day 796 (2013) determined the range of the three mid-Permian pareiasaur genera accepted by Lee 797 (1997a) to be mainly restricted to the Moordenaars and Karelskraal members of the upper 798 Abrahamskraal Formation (upper Tapinocephalus AZ), with the exception of Bradysaurus 799 (Fig. 5: Day 2013). However, this was based on a small number of specimens, mainly those 800 mentioned by Lee (1997a); of the large number of museum specimens, only those identified 801 as Bradysaurus were considered by Day (2013), likely contributing to its perceived longer 802 range. 803 More recently, Van den Brandt (2016) re-identified 39 out of 108 pareiasaur specimens from 804 the Tapinocephalus AZ (36%), ascribing the low percentage of positive identifications to the 805 study being restricted to crania only and to the poor condition of most of the material. 806 These new identifications nevertheless enabled the reassessment of stratigraphic ranges for 807 pareiasaur taxa in the Tapinocephalus AZ (Fig. 5: VdB 2016). 808 <> 809 <> 810 Bradysaurus has historically been a ‘waste basket’ for most large pareiasaurian specimens 811 collected from the Tapinocephalus AZ and so has generally been considered the most 812 abundant genus. We identified 74 specimens that could be reliably assigned to Bradysaurus 813 baini (Supplemental material, Appendix 1C: numbers 84 to 157). The stratigraphically lowest 814 occurring specimen is BP/1/7886 (from Banksdrif in the Laingsburg district). This locality is 37 815 well constrained to the mid/upper Koornplaats Member. The highest occurrences are 816 NHMUK PV OR 49426 (Palmietfontein, Beaufort West district) and BP/1/7213 (Witfontein, 817 Sutherland, Namakwa district) both of within the Karelskraal Member (Table 4). The 818 stratigraphic range of Bradysaurus baini therefore extends from the upper Koornplaats 819 Member to the Karelskraal Member (Fig. 5: This study), and thus occurs in both subzones of 820 the Tapinocephalus AZ (Day & Rubidge 2020). 821 <> 822 <> 823 We have identified 40 Embrithosaurus specimens, the stratigraphically lowest of which 824 (SAM-PK-9165) was found on Van Der Byls Kraal, Beaufort West district, at the boundary of 825 the Swaerskraal/Moordenaars Members. The highest occurrence is in the Karelskraal 826 Member, represented by specimens BP/1/7241 (from Wolwehoek, Beaufort West district) 827 and SAM-PK-012252 (from Gannakraal, Fraserburg, Namakwa district) (Table 4). The 828 stratigraphic range of Embrithosaurus thus corresponds closely with the Diictodon – 829 Styracocephalus Subzone of the Tapinocephalus AZ (Day & Rubidge 2020) (Fig. 5). 830 Forty-three specimens are referable to Nochelesaurus. The stratigraphically lowest 831 occurrence is specimen is BP/1/8011 (from De List, Laingsburg district) which was collected 832 from the upper Koornplaats Member. The highest definite occurrence is specimen 833 BP/1/7252 (from Wolwehoek, Hendriks Kraal 298, Beaufort West district), which is within 834 the Karelskraal Member (Table 4). Like Bradysaurus, Nochelesaurus therefore also occurs in 835 both subzones of the Tapinocephalus AZ (Day & Rubidge 2020) (Fig. 5). 38 836 Pareiasaur occurrences in the mid-Permian are restricted to a relatively small geographic 837 region of the south-western Karoo, and all the species are distributed throughout this 838 region; there is no variation in the distribution of the different species (Fig. 6). The majority 839 of specimens are located in a roughly triangular area between Prince Albert in the south, 840 Sutherland in the west, and Beaufort West in the east. A few specimens are scattered 841 further north, just north of Fraserburg. Towards the north of the basin (eg. Free State 842 Province) exposures of the lower Tapinocephalus AZ have not produced any fossils of large 843 animals, including pareiasaurs (Groenewald et al. 2019, 2022). 844 No Tapinocephalus AZ pareiasaurs are currently known to occur in the southeast of the 845 basin, likely the result of relatively little prospecting in this part of the basin, coupled with 846 less rock outcrop because of vegetation cover. 847 <> 848 Stratigraphic data presented here show Bradysaurus and Nochelesaurus both occur in the 849 Koornplaats Member, and thus their range begins in the Eosimops-Glanosuchus Subzone of 850 theTapinocephalus AZ, whereas Embrithosaurus is restricted to the overlying Diictodon- 851 Styracocephalus Subzone (Day & Rubidge 2020). The fact that Embrithosaurus is 852 phylogenetically the most basal genus of the group but stratigraphically the last to appear 853 suggests a ghost lineage for the genus through the lower Eosimops-Glanosuchus Subzone 854 (Day & Rubidge 2020). This may simply be the result of inadequate sampling from this lower 855 subzone, particularly below the Koornplaats Member. So far, no bradysaurian fossils have 856 been found in the Teekloof Formation, suggesting they perished (or at least were extirpated 857 from the Karoo Basin) during the Capitanian mass extinction (Day et al. 2018). 39 858 Anatomical discussion 859 It has long been known that the large basal South African pareiasaurs, the Bradysauria, 860 possess ‘leaf-shaped’ tooth crowns compared to the ‘fan-shaped’ tooth crowns of the 861 younger Lopingian pareiasaurs (Haughton & Boonstra 1929; Boonstra 1934). However, the 862 dental features resulting in ‘leaf-shaped’ or ‘fan-shaped’ pareiasaurian teeth are not well 863 explained. Accordingly we compiled a comparative list of characters separating the two 864 tooth types (Table 5). The most important and obvious evolutionary change is the increase 865 in the number of marginal tooth cusps between the members of the Bradysauria and the 866 younger pareiasaurs (characters 69 and 70), probably related to more efficient grasping of 867 plant material. 868 869 <> 870 871 Horizontal upper teeth of Bradysaurus baini. Boonstra (1934) described two contrasting 872 pareiasaurian jaw morphologies relating to the curvature of the maxilla and the orientation 873 of the upper jaw teeth, noting that maxillae with only slight ventral curvature hold vertical 874 teeth, whereas those that are strongly inwardly curved have lingually directed teeth. Most 875 pareiasaurs have a vertical maxilla that is only slightly or gently curved ventrally and the 876 upper jaw teeth point straight down. Lee (1994, 1997a) noted three taxa that show 877 exceptions: the slightly anteriorly or labially-pointing upper teeth of Scutosaurus (an 878 autapomorphy for this taxon), and the lingually-pointing or “medially inflected” upper teeth 879 of Pareiasuchus nasicornis (Lee et al. 1997:307) and Bradysaurus baini. He proposed a 880 character to score the orientation of the upper teeth (Lee 1997a, character 27). 881 40 882 Jalil & Janvier (2005) reported anteroventrally directed upper jaw teeth in Arganaceras to a 883 more exaggerated degree than in Scutosaurus. Tsuji (2013, Appendix 3) published a list of 884 autapomorphies per species in which she did not consider the lingually-pointing upper jaw 885 teeth of Pareiasuchus nasicornis and Bradysaurus baini as autapomorphies (presumably 886 since the condition occurs in more than one taxon). Turner et al. (2015, Supplemental data, 887 Appendix 1:8) termed the lingually-pointing condition for these two taxa as“ventromedially” 888 oriented upper teeth for character 67 in their phylogenetic analysis. 889 890 Scutosaurus specimens studied by us confirm that, despite the often severely distorted and 891 crushed condition of many skulls, this taxon consistently displays slightly outwardly-oriented 892 upper jaw teeth. This suggests that even severe cranial distortions do not significantly affect 893 the orientation of the teeth. Similarly, the ventromedially oriented (lingually-pointing) upper 894 jaw teeth of Bradysaurus baini are probably a true character of the species and not the 895 result of distortion. Moreover, the lingually-pointing upper jaw teeth of Bradysaurus baini 896 are always accompanied by a strongly medially curved ventral maxilla, with a large portion 897 curved and directed horizontally just above the tooth row to accommodate the long curved 898 roots of ventromedially oriented, lingually-pointing teeth. In lateral view, these horizontal 899 upper teeth are not visible in specimens of Bradysaurus baini, and in ventral view, the 900 inwardly curved maxilla and horizontal upper jaw teeth cover and obscure the anterior and 901 lateral portions of the palate. In Embrithosaurus and Nochelesaurus, the upper teeth are 902 visible in lateral view and in ventral view, the slightly curved ventral maxilla and vertical 903 teeth do not obscure the palate. We noticed that in all middle Permian pareiasaurs the 904 orientation of teeth in the upper jaw is not consistent along the entire tooth row as the 41 905 distal most upper teeth point slightly medially. The interspecific differences in tooth 906 orientation discussed here apply to only the premaxillary and anterior maxillary teeth. 907 Speculating on the validity of the horizontal upper teeth of Bradysaurus as a diagnostic 908 character we note that all three middle Permian species have lower jaws that are 909 significantly smaller than the upper jaws. When the jaws are occluded, the lower teeth pass 910 well inside the upper teeth, by several centimetres. No tooth wear is evident and no tooth 911 wear patterns have been described for any pareiasaur. The considerable distance between 912 upper and lower jaw teeth and the lack of tooth wear in pareiasaurs led Haughton & 913 Boonstra (1930: 264) to propose that pareiasaurian teeth do not occlude and so did not 914 directly cut or grind vegetation, but merely served as “rugosities to prevent the soft 915 vegetation from slipping out after a mouthful has been taken”. 916 Lee et al. (1997) explained how the uninterrupted even cutting edge of pareiasaurian 917 dentition, characteristic of herbivores, employs a cropping action which does not require 918 contact between upper and lower teeth but rather the upper and lower teeth move past 919 one another in a scissor-like action. Lee et al. (1997) also discussed how the depressed jaw 920 articulation of pareiasaurs (well below the tooth row) produces a posteriorly directed force 921 during mastication. This creates a shearing action that facilitates cropping of tough plant 922 matter, limiting pareiasaurs to cropping off bits of vegetation before swallowing in the 923 absence of propalinal movement. Lee’s explanations (no occlusion and posteriorly directed 924 upper forces) may account for the seemingly strange horizontal lingually-pointing upper 925 teeth of Bradysaurus baini (and Pareiasuchus nasicornis) being functional, and the unusual 926 orientation of the upper dentition is not the result of distortion. Niche partitioning (differing 927 diets) may be related to the unique upper tooth orientation of Bradysaurus baini compared 42 928 to Embrithosaurus and Nochelesaurus enabling three genera of similar size to coexist 929 without apparently competing for food. 930 Intraspecific variability among the Bradysauria. Our study has broadened the taxonomic 931 definition and increased the observed anatomical variability within each species 932 (Supplemental material, Appendix 3). 933 The variability of the number of cusps on the upper and lower teeth rang from seven to nine 934 (Bradysaurus), and seven to ten (Nochelesaurus). Previously Embrithosaurus was considered 935 the only species of the Bradysauria to have more than seven cusps (Boonstra 1934; Lee 936 1994, 1997a; Van den Brandt et al. 2020). Embrithosaurus schwarzi has either seven-or 937 nine-cusped upper teeth, Nochelesaurus alexanderi has seven-or-eight cusped upper teeth 938 (with one outlier, BP/1/7165, having five-and seven-cusped upper teeth) and Bradysaurus 939 baini seven-, eight- or nine-cusped upper teeth. Embrithosaurus may have either 940 dorsoventrally shortened maxillary teeth, lacking a long central trident (nine-cusped teeth) 941 or dorsoventrally elongated maxillary teeth, with a long central trident (seven-cusped 942 teeth). By contrast, Bradysaurus and Nochelesaurus have only dorsoventrally elongated 943 maxillary teeth with a long central trident. 944 Lower jaw teeth show even more variability in all three species. Embrithosaurus schwarzi 945 has the least variability with seven-or nine-cusped, shortened lower teeth which have 946 symmetrical cusp arrangements, no medial cusps, a low lingual bulge, and no lingual 947 cingulum. Nochelesaurus alexanderi is highly variable with seven-to ten-cusped lower, 948 elongated teeth, non-symmetrical cusp arrangements, an incipient cingulum ridge on 949 certain teeth, medial cusps sometimes present, and one to three vertical lingual ridges. The 950 lower teeth of Bradysaurus baini are also highly variable with elongated or shortened teeth, 43 951 seven to nine marginal cusps, symmetrical or non- symmetrical cusp arrangements, medial 952 cusps sometimes present, and a bulged lingual surface. 953 Maxillary teeth are vertically oriented in Embrithosaurus schwarzi and Nochelesaurus 954 alexanderi, but vary from ventromedially to horizontally oriented in Bradysaurus baini. 955 Both Bradysaurus baini and Embrithosaurus schwarzi develop a maxillary boss in ontogeny. 956 Juvenile specimens may lack the boss, sub-adults may have a low or small boss, and large 957 adults have the largest maxillary bosses. Individuals of Nochelesaurus alexanderi never 958 develop a maxillary boss, even in large adults. 959 For Bradysaurus baini the size of the cheek (expressed as degrees below the tooth row that 960 the ventral surface of the quadratojugal creates, Turner et al. 2015, character 44) is highly 961 variable, ranging from small (20°) to large (30°) and two specimens have cheeks larger than 962 30°. Embrithosaurus schwarzi (30°- 40°) and Nochelesaurus alexanderi (~30°) have more 963 consistently larger cheeks. These size differences and patterns are consistent and can 964 reliably be used to identify species. Small cheeks represent Bradysaurus specimens and the 965 largest cheeks belong to Embrithosaurus. 966 The size and shape of the bosses on the posterior cheek of Bradysaurus baini are extremely 967 variable, ranging from small to medium-sized knobs (which are as wide as high), to large, 968 wide and low, indistinct bulbous bosses. The posterior cheek bosses of Nochelesaurus are 969 also variable: either horizontally flat and pointed, plate-like bosses, or medium-sized bosses, 970 as wide as high. Only Embrithosaurus consistently shows wide, low, bulbous bosses. 971 The size and shape of the angular boss of the lower jaw is highly variable in all species and 972 only one pattern emerges: that most large specimens of Bradysaurus baini have relatively 44 973 small angular bosses compared to Embrithosaurus schwarzi and Nochelesaurus alexanderi. 974 Accordingly we tentatively include a relatively small angular boss in the diagnosis of 975 Bradysaurus baini. 976 Most specimens of Bradysaurus baini have a wide, low, irregular frontal central boss, but 977 some specimens have a high, distinct frontal central boss. 978 The three bradysaurian holotypes each have posterolateral and posteromedial nasal tabs 979 (Fig. 2). Bradysaurus and Embrithosaurus also have anterolateral tabs above the external 980 naris (not present in Nochelesaurus). However, when more specimens are studied, it is 981 evident that nasal tabs are highly variable among specimens of the three taxa and no 982 consistent patterns emerge. 983 In the postcranium the development of the olecranon process of the ulna is highly variable 984 (Van den Brandt et al. 2021b) and some authors have used the morphology of the olecranon 985 process to distinguish taxa (Watson 1914; Boonstra 1932b). We concur with Lee (1997b) 986 who considered smaller olecranon processes to be the result of early ontogenetic 987 development and should not be used to distinguish species. 988 The morphology of the internal trochanter of the femur differs among the species. 989 Proximally, it is wide, high and curved in both Bradysaurus and Embrithosaurus, more 990 curved in Bradysaurus into a semi-circular shape. In Nochelesaurus it is straight or very 991 slightly curved proximally. Distally, the internal trochanters of Embrithosaurus and 992 Bradysaurus have an anterior process that is absent in Nochelesaurus. The distal extension 993 of the lateral condyle of the femur shows great variability in Nochelesaurus, being long in 994 most specimens, but not all. 45 995 Pelvic differences between Embrithosaurus and Bradysaurus are consistent and are 996 therefore reliable species separators. Bradysaurus has a typical pareiasaurian pelvis: wide, 997 long and low, with almost parallel iliac blades and upturned anterior ends. In 998 Embrithosaurus the pelvis is narrow, short and high, with three autapomorphies: thick 999 symphysis and flat and straight vertical iliac blades that diverge anteriorly (Van den Brandt 1000 et al. 2021b). 1001 Differences in the torsion of the proximal and distal expansions of the humerus fall within a 1002 narrow range within a species and are consistent. They are therefore reliable species 1003 separators. Both Bradysaurus and Embrithosaurus have high degrees of humeral torsion 1004 within the range 60°-80° and Nochelesaurus has flatter expansions and torsion of 20°-40°. 1005 Are differences due to sexual dimorphism? Very little has been published on sexual 1006 dimorphism in pareiasaurs. Broom (1935) emphasised that the obvious pelvic differences 1007 between the holotypes of Bradysaurus baini and Embrithosaurus schwarzi may be the result 1008 of sexual dimorphism. We disagree with Broom, as these pelvic differences, together with 1009 other characters, consistently separate Bradysaurus and Embrithosaurus. Lee (1994:55-60) 1010 studied several articulated and very complete Scutosaurus specimens and showed that two 1011 morphs exist based on size, a smaller and larger morph, which he attributed to sexual 1012 dimorphism. 1013 Our research shows that Bradysaurus baini exhibits a large variation in cheek flange size. We 1014 suggest that: 1) small cheeks (extending only 20° below the tooth row) may belong to one 1015 sex and, 2) larger cheeks (30°) to the other. Similarly, the bosses of the posterior cheek of 1016 Bradysaurus baini display two morphotypes: 1) small to medium-sized, round, as wide as 1017 high bosses are one condition and, 2) large, low, wider than high, bulbous bosses are the 46 1018 other. However, we are not able to perfectly correlate those two pairs of morphologies with 1019 each other, and so cannot establish these two conditions as definitive sexual dimorphic 1020 characters in Bradysaurus baini. 1021 Effects of distortion. Several authors have recognised that pareiasaur remains from the 1022 Tapinocephalus AZ are often distorted and deformed, complicating taxonomic assessments. 1023 Despite acknowledging this, after the use of dental factors as their primary divisions, 1024 Haughton & Boonstra’s (1929) pareiasaurian classification scheme relied on cranial 1025 proportional differences (interorbital width compared to tabular width) for their sub- 1026 divisions. Lee (1994, 1997a) and Van den Brandt (2016, 2020) later determined that this 1027 particular cranial proportion does not vary among pareiasaur species, and pointed out that 1028 it is more reliable to focus on the absence or presence of discrete dental and cranial 1029 features. Van den Brandt (2016) briefly discussed his observation of two typical cranial 1030 morphotypes: 1) wide, broad, dorsoventrally flattened skulls with splayed out lateral 1031 cheeks, and 2) narrow, high, mediolaterally compressed skulls with tall, thin external nares. 1032 Whilst it is often possible to identify clear signs of distortion (e.g. buckled or displaced 1033 cranial bones), it is also true that in many cases the wider or the narrower skulls appear to 1034 be genuine morphological characters which are not affected by distortion. Interpreting the 1035 true skull shape is therefore often challenging. In general, most pareiasaur skulls from the 1036 Tapinocephalus AZ are dorsoventrally flat and wider than high (Turner et al. 2015, character 1037 33, snout dimensions) apart from Nochelesaurus, which appears to be taller, more square 1038 and almost as high as wide. It is tempting to classify the pareiasaurs from the 1039 Tapinocephalus AZ in two broad categories based on skull shape (wide vs. narrow), but this 47 1040 approach would group many specimens based on deformation and distortion, and not on 1041 cranial features. 1042 Implications of bilateral asymmetry for species identification. Van den Brandt (2016) and 1043 Van den Brandt et al. (2020) were the first to identify two different arrangement patterns of 1044 the posterior cheek bosses on the quadratojugal in basal South African pareiasaurs: a 1045 pattern with one small boss above the large corner boss and a pattern with two small 1046 bosses above the large corner boss. Van den Brandt (2016) found both arrangement 1047 patterns to be present for specimens of Bradysaurus baini, Bradysaurus seeleyi and 1048 Nochelesaurus alexanderi, but not for specimens of Embrithosaurus schwarzi, where the 1049 pattern of one small boss above the corner boss is always present. Our final study now has 1050 revealed specimens of Bradysaurus baini (SAM-PK-5624) and Embrithosaurus schwarzi (CGP 1051 CBT 112) in which both patterns are present on an individual skull, one on either cheek, 1052 indicating bilateral asymmetry. Therefore, posterior cheek boss arrangement patterns are 1053 not diagnostic of any particular species since one individual can have two different 1054 arrangement patterns. 1055 This brings into question the diagnostic validity of using the arrangement patterns of the 1056 posterior cheek bosses in closely related pareiasaurian species, such as the members of the 1057 Bradysauria. Van den Brandt et al. (2021a) discussed the validity of cranial ornamentation in 1058 pareiasaurian species identification, especially the size and shape of the posterior cheek 1059 bosses, and found that size and shape (but not arrangement pattern) to be a reliable species 1060 separator, even though there is overlap in that both Bradysaurus (holotype: wide, low 1061 bosses) and Nochelesaurus (holotype: high, pointed bosses) specimens may present 1062 medium-sized, as wide as high, posterior cheek bosses. 48 1063 There is also large variation and asymmetry within a single individual: the ulnae of the 1064 holotype of Bradysaurus baini, show extreme size differences of the olecranon process 1065 between the left (which is short , at 15 mm dorsoventrally high) and the right (which is three 1066 times as high, at 45 mm dorsoventrally high) (Van den Brandt et al. 2021b). If not found as 1067 part of an articulated skeleton (Seeley 1892) these two ulnae would be sufficiently different 1068 to be assigned to two different individuals and maybe two different species. Similarly, the 1069 femora of Nochelesaurus alexanderi specimen SAM-PK-9137 shows asymmetry with an 1070 internal trochanter that is straight on the left femur, but slightly curved on the right femur. 1071 These examples of bilateral asymmetry in cranial and postcranial features in middle Permian 1072 pareiasaur specimens, which might well identify individual elements as belonging to 1073 different species, should caution against the use of minor morphological differences to 1074 separate species. 1075 Phylogenetic discussion 1076 Lee (1997b) produced the first cladistic phylogenetic analysis of the relationships among 1077 pareiasaur taxa, using knowledge gained from his earlier work on the relationships among 1078 parareptiles (Lee 1993, 1994, 1997a). His first cladistic analysis of pareiasaurs found 1079 numerous most parsimonious trees (MPTs) in which the consensus tree showed 1080 Bradysaurus seeleyi as the most basal pareiasaur, followed by a polytomy including B. baini, 1081 Nochelesaurus alexanderi and a clade including Embrithosaurus schwarzi and all remaining 1082 pareiasaurs (Lee 1997b). Embrithosaurus schwarzi was recovered as the species with the 1083 most derived features of the four taxa from the middle Permian of South Africa and Lee 1084 (1997b) did not recover a monophyletic Bradysaurus. 49 1085 The initial character scores assigned to the four basal South African taxa by Lee (1997b) 1086 have received little revision, especially from direct study of the material. As a result, the 1087 phylogenetic position of the four Tapinocephalus AZ pareiasaurs has been virtually 1088 unchanged across all subsequent analyses (Lee 1997b; Jalil & Janvier 2005; Tsuji & Muller 1089 2008; Tsuji 2013; Tsuji et al. 2013; Turner et al. 2015; Xu et al. 2015; Benton 2016, Liu & 1090 Bever 2018). In the most recent phylogenetic analysis (before our work), Liu & Bever (2018) 1091 recovered B. seeleyi as the most basal pareiasaur, followed by a polytomy including B. baini 1092 and Nochelesaurus and a clade including Embrithosaurus and all remaining pareiasaurs. In 1093 all these previous analyses, Embrithosaurus is recovered as the taxon with the most derived 1094 features among the basal group of the four taxa from the middle Permian of South Africa. 1095 In our first phylogenetic analysis, Van den Brandt et al. (2020) described the cranium of 1096 Embrithosaurus schwarzi and rescored the cranial characters (characters 1-78) using the 1097 matrix of Liu & Bever (2018) excluding postcranial characters (characters 79-139). All 30 taxa 1098 used by Liu & Bever (2018) were retained and the phylogenetic analysis for the first time 1099 recovered Bradysauria (supported by three synapomorphies) as a monophyletic group 1100 comprising the four basal middle Permian South African pareiasaurs (Van den Brandt et al. 1101 2020). Nochelesaurus alexanderi and B. seeleyi formed a polytomy, and Embrithosaurus 1102 schwarzi and Bradysaurus baini formed a monophyletic group. 1103 Here, in our updated phylogenetic analysis, we have recovered a monophyletic Bradysauria 1104 supported by five synapomorphies. Nochelesaurus alexanderi and Bradysaurus baini form a 1105 monophyletic group and, in contrast to all previous analyses, Embrithosaurus is recovered as 1106 the most basal member. Derived phylogenetic features of Embrithosaurus schwarzi (Lee 1107 1997b) included a greater number of marginal cusps on the teeth (seven to nine); straight 50 1108 lateral projections on the caudal vertebrae; a shorter and thicker pubic symphysis; a large 1109 major trochanter on the femur; postaxial flange of the femur that narrows in the middle; a 1110 proximally curved internal trochanter; and a central boss on the osteoderms. Our research 1111 shows that a higher number of marginal cusps are found in all middle Permian South African 1112 pareiasaurian taxa, the postaxial flange of the femur of Embrithosaurus does not narrow in 1113 the middle, and a proximally curved internal trochanter also applies to Bradysaurus baini. In 1114 addition, we have made significant character score updates to all members of Bradysauria. 1115 This new phylogenetic result contradicts the stratigraphic ranges which show 1116 Embrithosaurus to be the last genus of South African middle Permian pareiasaurs to appear 1117 in the stratigraphic succession, suggesting a ghost lineage for the genus. 1118 Taxonomic discussion 1119 How many species: one or many? Figure 1 provides a summary of previous taxonomic 1120 schemes relating to middle Permian South African pareiasaurs, including the changes 1121 proposed in this study (see also Supplemental material, Appendix 5). 1122 A robust taxonomy with reliable distinguishing diagnostic characters to differentiate the 1123 pareiasaur species from the Tapinocephalus AZ of the Karoo remains challenging (Lee 1994, 1124 1997a; Van den Brandt 2016, 2020; Van den Brandt et al. 2020, 2021a, 2021b). 1125 Due to the high degree of intraspecific variability within each species and some overlapping 1126 features among all three species now evident from our study, it is tempting to argue that 1127 only one, extremely variable, pareiasaurian genus existed in the Karoo Basin in the 1128 Guadalupian, and that the differences are the result of extreme variability, ontogenetic 1129 development, sexual dimorphism, and/or distortion. This idea is supported by the fact that 51 1130 23 specimens with intermediate morphology exist, which cannot be confidently assigned to 1131 any taxon and whose identification we narrowed down to one of two species (Supplemental 1132 material, Appendix 1C: numbers 158 to 180). 1133 Here we have synonymised Bradysaurus seeleyi with Bradysaurus baini based on the 1134 observation that the holotype of B. seeleyi possesses no autapomorphies and otherwise 1135 appears to be a sub-adult Bradysaurus baini. The holotype of B. seeleyi is smaller and more 1136 gracile, cranially and postcranially, than any of the other bradysaurian holotypes, and we 1137 interpret the lack of a distinct maxillary boss as the result of incomplete ontogenetic 1138 development. This decreases the number of recognised pareiasaur species from the 1139 Tapinocephalus AZ to three. We chose to recognise these three genera on the basis that the 1140 three holotypes are clearly different from each other, and each shows unambiguous 1141 autapomorphies. The most important morphological differences among the three species 1142 are the orientation of the upper jaw teeth; the shapes of the postfrontal, postparietal, and 1143 the posterior cheek bosses; the shape and dimensions of the pelvis; the morphology of the 1144 internal trochanter of the femur; the degree of twisting of the humerus and the 1145 ornamentation style of the osteoderms. Moreover, the first appearances of all three taxa 1146 show clearly separate stratigraphic ranges (Fig. 5), which further supports the notion that 1147 they should not be synonymised. As such, we believe it is more cautious, given the current 1148 state of knowledge, to maintain the validity of these three genera and species. 1149 Conclusion 1150 This study has reviewed the pareiasaurs from the Tapinocephalus AZ of the Karoo Basin and 1151 differs from previous attempts in that it has considered almost all relevant pareiasaur 52 1152 material from multiple museums around the world (240 specimens). Having done so we 1153 recognise three monospecific pareiasaur genera from the Tapinocephalus AZ. 1154 Embrithosaurus schwarzi has the largest cheek flange of all middle Permian pareiasaurs, the 1155 only genus to reach and marginally exceed a 40° angle below the tooth row; distinctive wide 1156 dentition; and a combination of other cranial features (eg. narrow postparietal and a 1157 mediolaterally elongated postfrontal). In the dentition, there is a wider range in the number 1158 of cusps (seven to nine cusps) than the nine we previously proposed (Van den Brandt et al. 1159 2020). Embrithosaurus schwarzi is most easily separated from the co-occurring species 1160 through features of the pelvis and osteoderms: pelvis high, narrow, and short, with a deeply 1161 concave dorsal puboischiatic plate, and a large and long median pubic process; iliac blades 1162 oriented obliquely from the longitudinal midline, vertical; the thickest midline pelvic 1163 symphysis; anterior dorsal osteoderms with distinct and high, small central bosses, with thin 1164 and flat peripheries. 1165 Nochelesaurus alexanderi is cranially recognised by its autapomorphic wide postparietal and 1166 more distinctive cranial ornamentation, especially the larger and more pointed bosses on 1167 the posterior margins of the cheeks; features of the dentition (non-symmetrical marginal 1168 cusp arrangements, incipient horizontal cingulum with medially (lingually) placed cusp(s)) 1169 and a combination of other cranial features (e.g. square postfrontal, large cheeks, vertical 1170 upper jaw teeth, lack of a maxillary boss, more gracile distal paroccipital process of the 1171 opisthotic). Nochelesaurus alexanderi is most easily separated from the co-occurring species 1172 by: straight internal and small external trochanters of the femur; longer and narrower 1173 scapula blade; less twisting of the humerus; and relatively rugose osteoderms. 53 1174 Bradysaurus baini is diagnosed by maxillary teeth ventromedially or medially (horizontally) 1175 oriented; low cranial ornamentation; large maxillary boss, and autapomorphic distal 1176 paraoccipital process of the opisthotic. Variability in the sizes of the cheeks and in the sizes 1177 of the bosses on the posterior margin of the cheeks is large in this species. Specimens may 1178 have smaller (20°) to larger (30°) cheek flanges, which may have either large and wide, 1179 bulbous bosses on the posterior margins or medium-sized bosses, as high as wide. New 1180 dental features and increased dental variability have been found in B. baini, such as bulged 1181 or ridged lingual surfaces of lower jaw teeth and a higher number of marginal cusps, seven 1182 to nine, on upper and lower teeth. Bradysaurus baini is easily separated from the other 1183 middle Permian South African pareiasaurs through: twisted humerus; wide, low and long 1184 pelvis with posterior iliac blades parasagittal and anterior blades lateral and horizontally 1185 bent; unornamented osteoderms; short and distally flared scapula blade; large femur with a 1186 distinctive, curved and projecting internal trochanter and lack of a large external trochanter, 1187 and second sacral rib the largest. 1188 Bradysaurus baini, Embrithosaurus schwarzi and Nochelesaurus alexanderi are the only valid 1189 species of South African middle Permian pareiasaurs. Bradysaurus seeleyi is synonymised 1190 with Bradysaurus baini. The list of distinguishing features among the three species has been 1191 updated and we have documented intraspecific variability. Despite rigorous re-description 1192 and analysis of both cranial and postcranial characters which lead to the recognition of only 1193 three valid taxa, there are a few specimens that exhibit characters of more than one taxon, 1194 or characters intermediate to the condition present in two taxa. This suggests either greater 1195 intraspecific variability, or a greater diversity of taxa. This problem can be addressed only by 54 1196 the discovery of more well-preserved, well-prepared, and stratigraphically well- 1197 provenanced specimens. 1198 Our phylogenetic analysis recovered the clade Bradysauria (Van den Brandt et al. 2020) and 1199 suggests that it represents the early pulse of pareiasaurian diversification which is restricted 1200 to the middle Permian of the South African Karoo. Nochelesaurus and Bradysaurus are sister 1201 taxa and Embrithosaurus is the most basal member of the group. This new phylogenetic 1202 result contradicts the stratigraphic ranges, which shows Embrithosaurus as the last genus of 1203 South African middle Permian pareiasaurs to appear in the stratigraphic succession, 1204 suggesting a ghost lineage for the genus. 1205 The stratigraphic provenance of 142 of the 240 pareiasaur specimens from the 1206 Tapinocephalus AZ were determined to enable an accurate illustration of the range of each 1207 species (Supplemental material, Appendix 4; Fig. 5: This study). Bradysaurus baini is the 1208 oldest pareiasaurian and has the stratigraphically lowest first occurence in the Koornplaats 1209 Member of the Abrahamskraal Formation. Nochelesaurus has its first occurrence in the 1210 uppermost part of the Koornplaats Member and so, along with Bradysaurus, is a constituent 1211 part of the Eosimops-Glanosuchus Subzone of the Tapinocephalus AZ (Day & Rubidge 2020). 1212 All three genera are found throughout most of the Diictodon-Styracocephalus Subzone and 1213 have their highest occurrence within the Karelskraal Member near the top of the 1214 Tapinocephalus AZ, where they fell victim to the Capitanian mass extinction. 1215 Acknowledgments 1216 Michael Lee, Michael Benton, Nour-Eddine Jalil, Juan Carlos Cisneros, Andrey Sennikov, 1217 Elena Boyarinova, Sean Modesto, Christian Kammerer and Linda Tsuji are thanked for 55 1218 several discussions and their generous sharing of pareiasaurian knowledge and guidance. 1219 For access to the material studied, we thank: Ellen de Kock, Johann Neveling, and 1220 Nonhlanhla Mchunu (CGP); Bernhard Zipfel and Sifelani Jirah (ESI); William Simpson and 1221 Kenneth Angielczyk (FMNH); Sandra Chapman and Paul Barrett (NHMUK); Valeriy Bulanov, 1222 Valeriy Golubev and Andrey Sennikov (PIN); Claire Browning, Zaituna Skosan and Roger 1223 Smith (SAM); and Tom White (UMZC). 1224 For additional preparation of the holotypes of Embrithosaurus schwarzi (SAM-PK-8034) and 1225 Nochelesaurus alexanderi (SAM-PK-6239), we thank Pepson Mukanela, Gilbert Mokgethoa 1226 and Charlton Dube (ESI) and for specimen SAM-PK-6238, Johann van den Berg (SAM). For 1227 the re-assembly of the cranial and postcranial elements of the holotype of Embrithosaurus 1228 schwarzi (SAM-PK-8034), Sifelani Jirah and Charlton Dube (ESI) are thanked. 1229 For accommodation and local transport to museums we thank Naomi and John Van den 1230 Brandt (SAM-Cape Town), Tanya Van den Brandt and Peter van’t Hoogerhuys (NHMUK- 1231 London), Helen and Martin Meyer (UMZC-London), Kenneth Angielczyk and family (FMNH- 1232 Chicago) and Dylan Kammerer, Danielle Krasniqi and Isaac Wolf-Tanner (AMNH-New York). 1233 The research trip to the Paleontological Institute of the Russian Academy of Sciences, 1234 Moscow (PIN) was made possible through the support of the Bi-national Cooperation 1235 project between South Africa (NRF to Fernando Abdala) and Russia (Russian Foundation for 1236 Basic Research to Andrey Sennikov). Christian A Sidor and Mark MacDougall, the Associate 1237 Editor and the Editor-in-Chief of the journal are thanked for their reviews and useful 1238 comments that greatly improved the final manuscript. 1239 We acknowledge the financial support provided by the Department of Science and 1240 Innovation National Research Foundation (DSI/NRF) Centre of Excellence in Palaeosciences 56 1241 (CoE-Pal), South Africa; the National Research Foundation (NRF), South Africa; the 1242 Paleontological Scientific Trust (PAST) and its Scatterlings of Africa program; Johannesburg, 1243 South Africa; the African Origins Program (AOP) of the National Research Foundation (NRF), 1244 South Africa. Fernando Abdala’s research is financed by the NRF of South Africa and 1245 CONICET from Argentina. The support of the DSI/NRF Centre of Excellence in Palaeosciences 1246 (CoE-Pal) towards this research is hereby acknowledged. Opinions expressed and 1247 conclusions arrived at, are those of the author and are not necessarily to be attributed to 1248 the CoE. 1249 References 1250 Benton, M. J. 2016. The Chinese pareiasaurs. Zoological Journal of the Linnean Society, 177, 1251 813–853, doi.org/10.1111/zoj.12389. 1252 Boonstra, L. D. 1932a. The phylogenesis of the Pareiasauridae: a study in evolution. South 1253 African Journal of Science, 29, 480–486. 1254 Boonstra, L. D. 1932b. Pareiasaurian studies. Part 8. The osteology and myology of the 1255 locomotor apparatus. B – fore limb. Annals of the South African Museum, 28, 437–503. 1256 Boonstra, L. D. 1934. Pareiasaurian studies. Part 9. The cranial osteology. Annals of the 1257 South African Museum, 31, 1–38. 1258 Boonstra, L. D. 1969. The fauna of the Tapinocephalus zone (Beaufort beds of the Karoo). 1259 Annals of the South African Museum, 56, 1–73. 1260 Broom, R. 1903. On an almost perfect skeleton of Pareiasaurus serridens Owen. Annals of 1261 the South African Museum, 4, 123–138. 57 1262 Broom, R. 1912. On a new species of Propappus, and on the pose of the pareiasaurian limbs. 1263 Annals of the South African Museum, 7, 323–331. 1264 Broom, R. 1913. On four new fossil reptiles from the Beaufort Series, South Africa. Records 1265 of the Albany Museum, 2, 397–401. 1266 Broom, R. 1914. Further observations on the South African fossil reptiles. The American 1267 Museum Journal, 14, 139–143. 1268 Broom, R. 1916. Pareiasaurian nomenclature. Annals and Magazine of Natural History, 17, 1269 347–348. 1270 Broom, R. 1924. On some points in the structure of the pareiasaurian skull. Proceedings of 1271 the Zoological Society of London, 1924, 499–508. 1272 Broom, R. 1935. Notes on some species of pareiasarian reptiles. Annals of the Transvaal 1273 Museum, 18, 37–51. 1274 Cole, D. I., Johnson, M. R. & Day, M. O. 2016. Lithostratigraphy of the Abrahamskraal 1275 Formation (Karoo Supergroup), South Africa. South African Journal of Geology, 119, 415– 1276 424. 1277 Cope, E. D. 1896. The reptilian order Cotylosauria. – Proceedings of the American 1278 Philosophical Society, 34, 436–456. 1279 Day, M. O. 2013. Middle Permian continental biodiversity changes as reflected in the 1280 Beaufort Group of South Africa: a bio-and lithostratigraphic review of the Eodicynodon, 1281 Tapinocephalus, and Pristerognathus Assemblage Zones. Unpublished Ph.D thesis, 1282 University of the Witwatersrand, Johannesburg, 1–387. 58 1283 Day, M. O. & Rubidge, B. S. 2014. A brief lithostratigraphic review of the Abrahamskraal and 1284 Koonap formations of the Beaufort Group, South Africa: towards a basin-wide stratigraphic 1285 scheme for the Middle Permian Karoo. Journal of African Earth Sciences, 100, 227–242, 1286 doi.org/10.1016/j.jafrearsci.2014.07.001. 1287 Day, M. O., & Rubidge, B. S. 2020. Biostratigraphy of the Tapinocephalus Assemblage Zone 1288 (Beaufort Group, Karoo Supergroup), South Africa. South African Journal of Geology 2020, 1289 123(2), 149–164, doi.org/10.25131/sajg.123.0012. 1290 Day, M. O., Ramezani, J., Bowring, S. A., Sadler, P. M., Erwin, D. H., Abdala, F. & Rubidge, 1291 B.S. 2015. When and how did the terrestrial mid-Permian mass extinction occur? Evidence 1292 from the tetrapod record of the Karoo Basin, South Africa. Proceedings of the Royal Society 1293 of London B, 282, doi.org/10.1098/rspb.2015.0834. 1294 Day, M. O., Benson, R. B., Kammerer, C. F., & Rubidge, B. S. 2018. Evolutionary rates of 1295 mid-Permian tetrapods from South Africa and the role of temporal resolution in turnover 1296 reconstruction. Paleobiology, 44(3), 347–367, doi.org/10.1017/pab.2018.17. 1297 Geological Survey. 1979: 3222 Beaufort West 1:250 000 Geological Map Series. Compiled 1298 by: Marsh, A., Leaman, P.W., and Johnson, M.R., 1978. Government Printer, Pretoria, South 1299 Africa. 1300 Geological Survey. 1983: 3220 Sutherland 1:250 000 Geological Map Series. Compiled by: 1301 Wickens, H. de V., and Siegfried H.P., 1978. Government Printer, Pretoria, South Africa. 1302 Geological Survey. 1989a: 3122 Victoria West 1:250 000 Geological Map Series. Compiled 1303 by: Leroux, F.G. and de Beer, C.H., 1987. Government Printer, Pretoria, South Africa. 59 1304 Geological Survey. 1989b: 3120 Williston 1:250 000 Geological Map Series. Compiled by: 1305 Viljoen, J.H.A., 1985. Government Printer, Pretoria, South Africa. 1306 Goloboff, P., Farris, J. & Nixon, K. 2008. TNT, a free program for phylogenetic analysis, 1307 Cladistics, 24, 774–786. 1308 Goloboff, P. & Catalano, S. A. 2016. TNT version 1.5, including a full implementation of 1309 phylogenetic morphometrics, Cladistics, 32, 221–238, doi:10.1111/cla.12160. 1310 Groenewald, D. P., Day, M. O. & Rubidge, B. S. 2019. Vertebrate assemblages from the 1311 north-central Main Karoo Basin, South Africa, and their implications for mid-Permian 1312 biogeography, Lethaia, 52, 486–501, doi.org/10.1111/let.12326. 1313 Groenewald, D. P., Day, M. O., Penn-Clarke, C. R. & Rubidge, B. S. 2022. Stepping out 1314 across the Karoo retro-foreland basin: Improving constraints on the Ecca-Beaufort shoreline 1315 along the northern margin. Journal of African Earth Sciences, 185, 104389. 1316 Haughton, S. H. & Boonstra, L. D. 1929. Pareiasaurian studies. Part I. An attempt at a 1317 classification of the pareiasauria based on skull features. Annals of the South African 1318 Museum, 28, 79–87. 1319 Haughton, S. H. & Boonstra, L. D. 1930. Pareiasaurian studies. Part 5. On the pareiasaurian 1320 mandible. Annals of the South African Museum, 28, 261–288. 1321 ICZN (International Commission on Zoological Nomenclature). 1999. International Code of 1322 Zoological Nomenclature, fourth edition. International Trust for Zoological Nomenclature, 1323 London, xxix, 1–306. 60 1324 Jalil, N.-E. & Janvier, P. 2005. Les pareiasaures (Amniota, Parareptilia) du Permien superieur 1325 du Bassin d’Argana, Maroc. Geodiversitas, 27(1), 35–132. 1326 Keyser, A. W. & Smith, R. H. M. 1979. Vertebrate biozonation of the Beaufort Group with 1327 special reference to the Western Karoo Basin. Annals Geological Survey South Africa, 12, 1– 1328 36. 1329 Kitching, J.W. 1995. Biostratigraphy of the Dicynodon Assemblage Zone. In: Rubidge, B.S. 1330 (ed.). Biostratigraphy of the Beaufort Group (Karoo Supergroup), Biostratigraphic Series 1. 1331 South African Committee for Stratigraphy, Pretoria, 29–34. 1332 Kuhn, O. 1969. Cotylosaria. Handbuch der Palaoherpetologie, Teil 6. Stuttgart: Gustav 1333 Fischer Verlag. 1334 Laurenti, J. N. 1768. Classis Reptilium. Specimen medicum, exhibens synopsis Reptilium 1335 emendatum, cum experimentis circa venena et antidote Reptilium Austriacorum. J. Thom., 1336 Nob. et Trattnern, Vienna. 1337 Lee, M. S. Y. 1993. The origin of the turtle body plan: bridging a famous morphological gap, 1338 Science, 261, 1716–1720. 1339 Lee, M. S. Y. 1994. Evolutionary morphology of pareisaurs. Unpublished Ph.D thesis, 1340 University of Cambridge, Cambridge, 1–392. 1341 Lee, M. S. Y. 1995. Historical burden in systematics and the interrelationships of 1342 ‘parareptiles’. Biological Reviews, 70, 459–547. 1343 Lee, M. S. Y. 1996. Correlated progression and the origin of turtles. Nature, 379, 812–815. 61 1344 Lee, M. S. Y. 1997a. A taxonomic revision of pareiasaurian reptiles: implications for Permian 1345 terrestrial palaeoecology. Modern Geology, 21, 231–298. 1346 Lee, M. S. Y. 1997b. Pareiasaur phylogeny and the origin of turtles. Zoological Journal of the 1347 Linnean Society, 120, 197–280. 1348 Lee, M. S. Y., Gow, C. E. & Kitching, J. W. 1997. Anatomy and the relationships of the 1349 pareiasaur Pareiasuchus nasicornis from the upper Permian of Zambia. Palaeontology, 1350 40(2), 307–335. 1351 Liu, J & Bever, G. S. 2018. The Tetrapod Fauna of the Upper Permian Naobaogou Formation 1352 of China: a new species of Elginia (Parareptilia, Pareiasauria). Papers in Palaeontology, 4(2), 1353 197209. doi.org/10.1002/spp2.1105. 1354 Olson, E. C. 1947. The Family Diadectidae and its bearing on the classification of reptiles. 1355 Fieldiana Geology, 11, 1–53. 1356 Owen, R. 1856. Report on the reptilian fossils of South Africa. Part 3. On parts of the 1357 skeleton of the trunk of the Dicynodon tigriceps. Transactions of the Geological Society of 1358 London, Series 2, 7, 241–247. 1359 Owen, R. 1862. On the dicynodont Reptilia, with a description of some fossil remains 1360 brought by H.R.H. Prince Alfred from South Africa, November 1860. Philosophical 1361 Transactions of the Royal Society of London B, 152, 455–467. 1362 Owen, R. 1876. Descriptive and Illustrated Catalogue of the Fossil Reptilia of South Africa in 1363 the Collection of the British Museum. London, British Museum (Natural History). 62 1364 Rubidge, B. S. (ed.), 1995. Biostratigraphy of the Beaufort Group (Karoo Supergroup). 1365 Pretoria, South African Committee for Stratigraphy, Biostratigraphic Series 1, 46p. 1366 Seeley, H.G. 1888. Researches on the structure, organisation, and classification of the fossil 1367 Reptilia. II. On Pareiasaurus bombidens (Owen), and the significance of its affinities to 1368 amphibians, reptiles and mammals. Philosophical Transactions of the Royal Society of 1369 London B, 179, 59–109. 1370 Seeley, H.G. 1892. Researches on the structure, organisation, and classification of the fossil 1371 Reptilia. VII. Further observations on Pareiasaurus. Philosophical Transactions of the Royal 1372 Society of London B, 183, 311–370. 1373 Smith, R. M. H. & Botha-Brink, J. 2014. Anatomy of a mass extinction: Sedimentological and 1374 taphonomic: evidence for drought-induced die-offs at the Permo-Triassic boundary in the 1375 main Karoo Basin, South Africa. Palaeogeography, Palaeoclimatology, Palaeoecology, 396, 1376 99–118, doi.org/10.1016/j.palaeo.2014.01.002. 1377 Smith, R. M. H. & Keyser, A. W. 1995a. Biostratigraphy of the Tapinocephalus Assemblage 1378 Zone. In: Rubidge, B.S. (ed.). Biostratigraphy of the Beaufort Group (Karoo Supergroup), 1379 Biostratigraphic Series 1. South African Committee for Stratigraphy, Pretoria, 8–12. 1380 Smith, R. M. H. & Keyser, A. W. 1995b. Biostratigraphy of the Pristerognathus Assemblage 1381 Zone. In: Rubidge, B.S. (ed.). Biostratigraphy of the Beaufort Group (Karoo Supergroup), 1382 Biostratigraphic Series 1. South African Committee for Stratigraphy, Pretoria, 13–17. 1383 Smith, R. M. H. & Keyser, A. W. 1995c. Biostratigraphy of the Tropidostoma Assemblage 1384 Zone. In: Rubidge, B.S. (ed.). Biostratigraphy of the Beaufort Group (Karoo Supergroup), 1385 Biostratigraphic Series 1. South African Committee for Stratigraphy, Pretoria, 18–22. 63 1386 Smith, R. M. H. & Keyser, A. W. 1995d. Biostratigraphy of the Cistecephalus Assemblage 1387 Zone. In: Rubidge, B.S. (ed.). Biostratigraphy of the Beaufort Group (Karoo Supergroup), 1388 Biostratigraphic Series 1. South African Committee for Stratigraphy, Pretoria, 23–28. 1389 Smith, R.M.H., Rubidge, B.S., Day, M.O. & Botha, J. 2020. Introduction to the tetrapod 1390 biozonation of the Karoo Supergroup. South African Journal of Geology, 123, 131–140. 1391 https://doi.org/10.25131/sajg.123.0009 1392 Tsuji, L. A. & Muller, J. 2008. A re-evaluation of Parasaurus geinitzi, the first named 1393 pareiasaur (Amniota, Parareptilia). Canadian Journal of Earth Sciences. 45(10), 1111–1121, 1394 doi.org/10.1139/E08-060. 1395 Tsuji, L. A. 2010. Evolution, Morphology and Paleobiology of the Pareiasauria and their 1396 Relatives (Amniota: Parareptilia). Unpublished Ph.D thesis, Humboldt-University of Berlin, 1397 Berlin, 1–188. 1398 Tsuji, L. A. 2013. Anatomy, cranial ontogeny and phylogenetic relationships of the 1399 pareiasaur Deltavjatia rossicus from the Late Permian of central Russia. Earth and 1400 Environmental Science Transactions of the Royal Society of Edinburgh, 104, 1–42. 1401 Tsuji, L. A., Sidor, C. A., Steyer, S. J., Smith, R. M. H., Tabor, N. J. & Ide. O. 2013. The 1402 vertebrate fauna of the Upper Permian of Niger –VII. Cranial anatomy and relationships of 1403 Bunostegos akokanenis (Pareiasauria). Journal of Vertebrate Palaeontology, 33(4), 747–763, 1404 doi.org/10.1080/02724634.2013.739537. 1405 Turner, M. L., Tsuji, L. A., Ide, O. & Sidor, C. A. 2015. The vertebrate fauna of the Upper 1406 Permian of Niger –IX. The appendicular skeleton of Bunostegos akokanenis (Parareptilia: 64 1407 Pareiasauria). Journal of Vertebrate Palaeontology, 35(6), 1408 doi.org/10.1080/02724634.2014.994746. 1409 Van den Brandt, M. J. 2016. Cranial morphology of Embrithosuarus schwarzi (Parareptilia, 1410 Pareiasauria) and a taxonomic and stratigraphic reassessment of the South African middle 1411 Permian pareiasaurs. Unpublished MSc thesis, University of the Witwatersrand, 1412 Johannesburg, 1–214. 1413 Van den Brandt, M. J. 2020. Morphology, taxonomy, phylogeny and stratigraphic ranges of 1414 South African middle Permian pareiasaurs. Unpublished Ph.D thesis, University of the 1415 Witwatersrand, Johannesburg, 1–565. 1416 Van den Brandt, M. J., Abdala, F. & Rubidge, B. S. 2020. Cranial morphology and 1417 phylogenetic relationships of the Middle Permian pareiasaur Embrithosaurus schwarzi from 1418 the Karoo Basin of South Africa. Zoological Journal of the Linnean society. 188, 202–241, 1419 doi.org/10.1093/zoolinnean/zlz064. 1420 Van den Brandt, M. J., Rubidge, B. S., Benoit, J. & Abdala, F. 2021a. Cranial morphology of 1421 the middle Permian pareiasaur Nochelesaurus alexanderi from the Karoo Basin of South 1422 Africa. Earth and Environmental Science Transactions of the Royal Society of Edinburg, 112, 1423 29-49, doi:10.1017/S1755691021000049. 1424 Van den Brandt, M. J., Benoit, J., Abdala, F. & Rubidge, B. S. 2021b. Postcranial morphology 1425 of the South African middle Permian pareiasaurs from the Karoo Basin of South Africa. 1426 Palaeontologia africana, 55, 1–91, https://hdl.handle.net/10539/31290. 1427 Watson, D. M. S. 1914. On the nomenclature of pareiasaurs. Annals of the Magazine of 1428 Natural History, 14, 98–102. 65 1429 Xu, L., Li, X. W., Jia, S. H. & Liu, J. 2015. The Jiyuan Tetrapod Fauna of the Upper Permian of 1430 China: New Pareiasaur Material and the Reestablishment of Honania complicidentata. Acta 1431 Palaeontologica Polonica, 60(3), 689–700, https://doi.org/10.4202/app.00035.2013. 1432 Figure Captions 1433 Figure 1: Species flow chart, showing a summary of previous taxonomic revisions of 1434 Guadalupian South African pareiasaurs and the changes proposed in this study. Rows 1435 represent species. Columns indicate author recognition of species: dotted lines indicate 1436 where a species was not recognized by an author, solid lines indicate the species assigned, 1437 including synonymising with other species. All authors apart from Haughton and Boonstra 1438 (1929) and Kuhn (1969) consider Platyoropha broomi (SAM-PK-5002) a synonym of 1439 Bradysaurus baini. Lee (1997a) lumped several taxa into only four species, which were 1440 retained by Van den Brandt (2016, et al. 2020). This study proposes synonymising 1441 Bradysaurus seeleyi, Bradysaurus vanderbyli and Bradysaurus acutirostris with Bradysaurus 1442 baini, Bradysaurus whaitsi with Embrithosaurus schwarzi, and Dolichopareia angusta, 1443 Brachypareia rogersi and Brachypareia watsoni with Nochelesaurus alexanderi. 1444 Figure 2: Interpretive drawings of the skull of the holotype of Bradysaurus baini (NHMUK PV 1445 R 1971), in A, dorsal, and B, left lateral views; the skull of the holotype of Embrithosaurus 1446 schwarzi (SAM-PK-8034) in C, dorsal, and D, left lateral views; the skull of the holotype of 1447 Nochelesaurus alexanderi (SAM-PK-6239), in E, dorsal, and F, left lateral views. A, B from Lee 1448 (1994, 1997a) reproduced with permission. Abbreviations: ang, angular; den, dentary; fr, 1449 frontal; ju, jugal; la, lacrimal; mf, maxillary foramen; max, maxilla; na, nasal; par, parietal; 1450 pob, postorbital; pm, premaxilla; po, postorbital; pof, postfrontal; pp, postparietal; prf, 1451 prefrontal; qj, quadratojugal; spl, splenial; sq, squamosal; st, supratemporal; sur, 66 1452 surangular. Numbers indicate marginal cheek bosses. Cranial bosses are in bold, rugose 1453 ridges are in regular line thickness. Sutures drawn in solid lines have been positively 1454 identified; those in dotted lines are perpendicular extrapolations from internal surface 1455 sutures; light grey indicates sides of bones or bone breaks, dark grey indicates foramen or 1456 fenestra, black indicates deep cavities such as orbits. 1457 Figure 3: Interpretive drawings of the pelvis of the holotype of Embrithosaurus schwarzi 1458 (SAM-PK-8034) in A, anterior, B, dorsal, and C, right lateral views; the pelvis of the holotype 1459 of Bradysaurus baini (NHMUK PV R 1971), in D, anterior, E, dorsal, and F, right lateral views. 1460 Dorsal to the top of the page in A, D, C and F. Anterior to the top of the page in B, E. 1461 Abbreviations: ace, acetabulum; ib, ischial buttress; lpp, lateral pubic process; mpp, median 1462 pubic process; pb, pubic buttress; pf, pubic/obturator foramen; sb, supra-acetabular 1463 buttress. 1464 Figure 4: Cladistic relationships of the Bradysauria within Pareiasauria, A, strict consensus 1465 tree, B, 50% majority-rule consensus tree of 1065 most parsimonious trees. Numbers at 1466 each node indicate the frequency of clades in the fundamental trees. 1467 Figure 5: Composite minimum stratigraphic ranges of pareiasaur genera in the Guadalupian 1468 Beaufort Group based on positively identified specimens of each species, Day (2013), Van 1469 den Brandt (2016) and this study. Biostratigraphy after Day & Rubidge (2020). WUC = 1470 Wuchiapingian. 1471 Figure 6: Geographic distribution of localities for 142 identified pareiasaur specimens in the 1472 middle Permian Beaufort Group, superimposed on geological data. Almost all specimens 1473 occur between 33° and 32° South latitude. Figure created using the software Global Mapper 67 1474 version 15. Geology is based on the 1:250 000 Geological map sheets 3122 Victoria West 1475 (Geological Survey 1989a); 3120 Williston (Geological Survey 1989b); 3220 Sutherland 1476 (Geological Survey 1983); 3222 Beaufort West (Geological Survey 1979). M = Member, Fm = 1477 Formation. 1478 68 Figure 1 Species flow chart, showing a summary of previous taxonomic revisions of Guadalupian South African pareiasaurs and the changes proposed in this study. Rows represent species. Columns indicate author recognition of species: dotted lines indicate where a species was not recognized by an author, solid lines indicate the species assigned, including synonymising with other species. All authors apart from Haughton and Boonstra (1929) and Kuhn (1969) consider Platyoropha broomi (SAM-PK-5002) a synonym of Bradysaurus baini. Lee (1997a) lumped several taxa into only four species, which were retained by Van den Brandt (2016, et al. 2020). This study proposes synonymising Bradysaurus seeleyi, Bradysaurus vanderbyli and Bradysaurus acutirostris with Bradysaurus baini, Bradysaurus whaitsi with Embrithosaurus schwarzi, and Dolichopareia angusta, Brachypareia rogersi and Brachypareia watsoni with Nochelesaurus alexanderi. 69 Figure 2 Interpretive drawings of the skull of the holotype of Bradysaurus baini (NHMUK PV R 1971), in A, dorsal, and B, left lateral views; the skull of the holotype of Embrithosaurus schwarzi (SAM-PK-8034) in C, dorsal, and D, left lateral views; the skull of the holotype of Nochelesaurus alexanderi (SAM-PK-6239), in E, dorsal, and F, left lateral views. A, B from Lee (1994, 1997a) reproduced with permission. Abbreviations: ang, angular; den, dentary; fr, frontal; ju, jugal; la, lacrimal; mf, maxillary foramen; max, maxilla; na, nasal; par, parietal; pob, postorbital; pm, premaxilla; po, postorbital; pof, postfrontal; pp, postparietal; prf, prefrontal; qj, quadratojugal; spl, splenial; sq, squamosal; st, supratemporal; sur, surangular. Numbers indicate marginal cheek bosses. Cranial bosses are in bold, rugose ridges are in regular line thickness. Sutures drawn in solid lines have been positively identified; those in dotted lines are perpendicular extrapolations from internal surface sutures; light grey indicates sides of bones or bone breaks, dark grey indicates foramen or fenestra, black indicates deep cavities such as orbits. 70 Figure 3 Interpretive drawings of the pelvis of the holotype of Embrithosaurus schwarzi (SAM-PK-8034) in A, anterior, B, dorsal, and C, right lateral views; the pelvis of the holotype of Bradysaurus baini (NHMUK PV R 1971), in D, anterior, E, dorsal, and F, right lateral views. Dorsal to the top of the page in A, D, C and F. Anterior to the top of the page in B, E. Abbreviations: ace, acetabulum; ib, ischial buttress; lpp, lateral pubic process; mpp, median pubic process; pb, pubic buttress; pf, pubic/obturator foramen; sb, supra-acetabular buttress. 71 Figure 5 Cladistic relationships of the Bradysauria within Pareiasauria, A, strict consensus tree, B, 50% majority-rule consensus tree of 1065 most parsimonious trees. Numbers at each node indicate the frequency of clades in the fundamental trees. Figure 4 Composite minimum stratigraphic ranges of pareiasaur genera in the Guadalupian Beaufort Group based on positively identified specimens of each species, Day (2013), Van den Brandt (2016) and this study. Biostratigraphy after Day & Rubidge (2020). WUC = Wuchiapingian. 72 Figure 6 Geographic distribution of localities for 142 identified pareiasaur specimens in the middle Permian Beaufort Group, superimposed on geological data. Almost all specimens occur between 33° and 32° South latitude. Figure created using the software Global Mapper version 15. Geology is based on the 1:250 000 Geological map sheets 3122 Victoria West (Geological Survey 1989a); 3120 Williston (Geological Survey 1989b); 3220 Sutherland (Geological Survey 1983); 3222 Beaufort West (Geological Survey 1979). M = Member, Fm = Formation. 73 Van den Brandt et Resampling al. (2020) This study Standard Bootstrapping Par 69 Bra 58 Par 85 Bra 62 Poisson Bootstrapping Par 68 Bra 56 Par 84 Bra 54 Jackknife Par 62 Bra 60 Par 74 Bra 76 Symmetric Resampling Par 76 Bra 69 Par 91 Bra 77 Table 1: Values of resampling obtained using different methods, after data of Van den Brandt et al. (2020) and this study. Jackknife and symmetric resampling used default values in TNT. Par, Pareiasauria; Bra, Bradysauria. Unidentified/ Identifications to Institution Identified one of two species Total GPIT 1 0 1 CGP 16 10 26 SAM 84 25 109 BPI/ESI 24 27 51 NHMUK 5 1 6 AM 1 2 3 AMNH 7 13 20 FMNH 19 5 24 Total 157 83 240 Table 2: Numbers and identification of pareiasaur specimens from the Tapinocephalus Assemblage Zone considered in this study. Taxa Specimens Bradysaurus baini 74 Embrithosaurus schwarzi 40 Nochelesaurus alexanderi 43 Bradysaurus or Embrithosaurus 6 Bradysaurus or Nochelesaurus 7 Embrithosaurus or Nochelesaurus 10 Unidentified 60 TOTAL 240 Table 3: Numbers of pareiasaur specimens from the Tapinocephalus Assemblage Zone identified by taxa. 74 Genus Species Lowest Level Lowest biostrat specimen Bradysaurus baini BP/1/7886 Koornplaats Eosimops- Member Glanosuchus subzone Nochelesaurus alexanderi BP/1/8011 upper Eosimops- Koornplaats Glanosuchus Member subzone Embrithosaurus schwarzi SAM-PK- Swaerskraal/ Boundary of Eosimops- 9165 Moordenaars Glanosuchus & Diictodon- Member Styracocephalus subzones boundary Genus Species Highest Level Lowest biostrat specimen Bradysaurus baini NHMUK PV Karelskraal Diictodon-Styracocephalus OR 49426 or Member subzone BP/1/7213 or several from north of the Nuweveldber ge. Nochelesaurus alexanderi BP/1/7252 Karelskraal Diictodon-Styracocephalus Member subzone Embrithosaurus schwarzi BP/1/7241 Karelskraal Diictodon-Styracocephalus or SAM-PK- Member subzone 12252 Table 4: Stratigraphically lowermost and uppermost occurring specimens of each valid genus Bradysaurus, Embrithosaurus and Nochelesaurus, from which their stratigraphic range is determined. Bradysauria: ‘leaf-shaped’ tooth Lopingian pareiasaurs: ‘fan-shaped’ morphology tooth morphology Presence of elongated, central, three- Absence of central, three-cusped tridents cusped tridents on upper and/or lower teeth Tooth crowns elongated Tooth crowns shorter Teeth narrower, less mesio-distally Teeth wider, more mesio-distally expanded expanded Fewer marginal cusps per tooth (7-9) More marginal cusps per tooth (9+) Cusps irregularly arranged around the Cusps regularly arranged around the crown (spaces between cusps variable) crown (consistent spaces between cusps) Larger spaces between the cusps Smaller spaces between cusps Symmetrical or non-symmetrical mesial Symmetrical mesial and distally cusp and distal cusp arrangements arrangements Longer individual marginal cusps that Shorter individual marginal cusps that do project far out from the crown not project far out from the crown Table 5: Difference in tooth morphology between South African Guadalupian pareiasaurs (Bradysauria) and Lopingian pareiasaurs. 75