This is the accepted version of the article: Sorbelli, Leonardo; Alba, David M.; Cherin, Marco; [et al.]. «A review on Bison schoetensacki and its closest relatives through the early-Middle Pleistocene transition: Insights from the Vallparadís Section (NE Iberian Peninsula) and other European localities». Quaternary Science Reviews, Vol. 261 (June 2021), art. 106933. DOI 10.1016/j.quascirev.2021.106933 This version is available at https://ddd.uab.cat/record/239759 under the terms of the license 1 A review on Bison schoetensacki and its closest relatives through the Early-Middle Pleistocene Transition: 2 insights from the Vallparadís Section (NE Iberian Peninsula) and other European localities 3 4 Leonardo Sorbellia, David M. Albaa, Marco Cherinb, Pierre-Élie Moulléc, Jean-Philip Brugald, Joan Madurell- 5 Malapeiraa,* 6 a Institut Català de Paleontologia Miquel Crusafont, Universitat Autònoma de Barcelona, Edifici ICTA-ICP, c/ Columnes 7 s/n, Campus de la UAB, 08193 Cerdanyola del Vallès, Barcelona, Spain 8 b Dipartimento di Fisica e Geologia, Università degli Studi di Perugia, Via A. Pascoli, 06123 Perugia, Italy 9 c Musée de Préhistoire Régionale de Menton, 06500 Menton, France 10 d Aix-Marseille Université, CNRS, Minist. Culture, UMR 7269 Lampea, MMSH, 13094 Aix-en-Provence cedex 2, France 11 12 * Corresponding author. 13 E-mail address: joan.madurell@icp.cat (J. Madurell-Malapeira). 14 15 ABSTRACT 16 The evolutionary history of Bison is a matter of debate due to the scarcity of fossil remains from the 17 earliest members of this clade and the close morphological similarities among species. To clarify the 18 taxonomic status of the earliest stouter bison and their relationships to their putative ancestor, Leptobos, as 19 well as other primitive forms traditionally referred to subgenus Bison (Eobison), we carry out a complete 20 revision of the available European fossil record, with a focus on the forms occurring during the Early-Middle 21 Pleistocene Transition. Emphasis is put on the description of the unpublished Bison remains from the 22 Vallparadís Composite Section (VCS), including the sites of Cal Guardiola and Vallparadís Estació (Terrassa, NE 23 Iberian Peninsula). VCS fossiliferous layers yielded one of the richest faunal assemblages from the European 24 latest Early Pleistocene and one of the few European fossil sites covering almost entirely the Early-Middle 25 Pleistocene Transition (1.2–0.6 Ma). The collection comprises thousands of ungulates remains, especially 26 abundant fossils of a large Bison species. The morphology of the postcranial sample from VCS fits that of 27 Bison (Bison) schoetensacki, i.e., the earliest stout bisons (Bison s.s.) recorded in Europe. We studied more 1 28 than 200 cranial and postcranial elements with a focus on the metapodial remains. Comparisons were 29 performed with all the available fossil record of Pleistocene Eurasian fossil Bison species. We confirm the 30 taxonomic validity of B. schoetensacki and recognize distinct eco-morphotypes of European bison between 31 the late Early Pleistocene and the beginning of the Holocene based on the size and proportions of the 32 metapodials. Although the appendicular skeleton shows reliable characters for the diagnosis of different 33 species, the great morphological homogeneity recognized within the genus requires a cautious approach in 34 systematic studies based on postcranial material. 35 36 Keywords: Bison; Bovidae; Early-Middle Pleistocene Transition; Epivillafranchian; Europe; Quaternary. 37 38 1. Introduction 39 The genus Bison Hamilton Smith, 1827 includes two extant species, the North American bison (Bison bison) 40 and the European bison or wisent (Bison bonasus). Additional extinct species are distinguished, but the 41 evolutionary history of this group is still unclear, mainly due to the difficulty in distinguishing early Bison 42 remains from those of its putative ancestor. Nowadays, it is commonly accepted that Bison originated during 43 the Early Pleistocene from a species of mid-sized bovid Leptobos (e.g., Pilgrim, 1947; Brugal, 1985; Masini, 44 1989; Bukhsianidze, 2005; Martínez-Navarro et al., 2007). During the last century, a large amount of Bison 45 remains has been discovered from many European Quaternary sites and several species, or sub-species, of 46 this genus, or sub-genus, have been described fueling the debate regarding the evolutionary history of this 47 group of Bovini (Masini, 1989; Sher, 1997; Bukhsianidze, 2005; Maniakas and Kostopoulos, 2018; Kostopoulos 48 et al., 2018) (Fig. 1). 49 50 1.1. The European record of Leptobos 51 The genus Leptobos includes medium-sized slender bovids from the whole Villafranchian Land Mammal 52 Age (sensu Rook and Martínez Navarro, 2010), being constantly present in the faunal assemblages from the 53 Late Pliocene to most of the Early Pleistocene across Eurasia (from Iberia to northern China). According to 2 54 the most recent review (Cherin et al., 2019), this genus includes the following species: Leptobos brevicornis 55 and Leptobos crassus (China), Leptobos falconeri (Pakistan), Leptobos stenometopon and Leptobos merlai 56 (France and Italy), Leptobos furtivus (France and, possibly, Italy), Leptobos etruscus (France, Italy, and Spain), 57 and Leptobos vallisarni (Italy and China). According to Masini et al. (2013), Leptobos species from Europe can 58 be divided into two groups or lineages: one includes L. stenometopon, L. merlai, and L. furtivus; and the 59 other includes L. etruscus and L. vallisarni. The latter group comprises the younger, larger and more derived 60 Leptobos species, which have been considered—not without controversy (Bukshianidze, 2005)—as possible 61 ancestors of bison (Masini, 1989). The dispersal of bison from Asia marks the last occurrences of Leptobos in 62 Europe, even if the coexistence between early Bison and Leptobos seems documented in some late 63 Villafranchian Chinese, Greek-Balkan and Eastern Europe localities (Duvernois, 1990; Tong et al., 2016; 64 Agadzhanyan, 2017; Kostopoulos et al., 2018; Lopatin et al., 2019). 65 66 1.2. Earliest Asian and European bison 67 The earliest occurrences of Bison date to the Late Pliocene-Early Pleistocene of Southeast Asia (Tong et 68 al., 2016). The poorly-known Bison (Eobison) sivalensis from India (3.4–2.6 Ma) and Bison (Eobison) 69 palaeosinensis from China (ca. 2.6 Ma) share some features with the most derived Leptobos species and 70 differ in size and cranial morphology from “true” bison of subgenus Bison s.s. (Flerov, 1972; Tong et al., 71 2016). In some areas of Asia, these primitive bison might have been partially coeval and even co-occurring 72 for some time with Leptobos, which eventually went extinct by the end of the Villafranchian (Tong et al., 73 2016). 74 Flerov (1972) erected the subgenus Eobison to include the two aforementioned primitive Asian species as 75 well as the poorly-known Bison (Eobison) tamanensis from Europe, first described by Verestchagin (1959). 76 Later on, Eobison has been widely used, either as a subgenus (Masini et al., 2013) or as a separate genus 77 (Geraads, 1992), to allocate all the Early Pleistocene small-sized bison remains from Europe. However, this 78 taxonomic arrangement has been recently questioned (Bukshianidze, 2005; Kostopoulos et al., 2018), by 79 showing that Eobison is a weakly defined taxon and that some forms previously included in this subgenus 80 should be transferred to Bison (Bison) (e.g., Bison degiulii). 3 81 The earliest record of Bison in Europe is represented by B. (Eobison) georgicus from Dmanisi (Georgia; 82 1.76 Ma; Burchak-Abramovich and Vekua, 1994; Bukhsianidze, 2005). Other species from Eastern Europe 83 include the above-mentioned B. tamanensis, from the Epivillafranchian (ca. 1 Ma) of the Azov Sea area, and 84 whose taxonomic status is uncertain (being considered a species inquirenda by Kostopoulos et al., 2018), as 85 well as Bison suchovi from the late Villafranchian of Dolinskoye (Ukraine; Alekseeva, 1967, 1977), whose 86 taxonomic validity has similarly been questioned (Sher, 1997). 87 The earliest records of the genus Bison from Europe include: Bison sp. from Venta Micena (Spain; ca. 1.5 88 Ma; Moyà-Solà, 1987; Martínez-Navarro et al., 2011); B. (Eobison) sp. from Le Riège (France; ca. 1.4 Ma; 89 Ambert et al., 1996); B. (Eobison) degiulii from Pirro Nord and Capena (Italy; ca. 1.6–1.2 Ma; Masini, 1989; 90 Masini et al., 2013); and Bison cf. degiulii from the Mygdonia Basin (Greece; ca. 1.7–1.2 Ma; Kostopoulos et 91 al., 2018). By the end of the Early Pleistocene, the first “true” large bison appear, being represented by the 92 long-legged Bison (B.) menneri from Untermassfeld (Germany; 1.05 Ma; MIS31; Sher, 1997; Bukhsianidze, 93 2020)—recently included in the subgenus Bison (Poephagus) by Bukhsianidze (2020)—and the relatively 94 stouter woodland wisent B. (B.) schoetensacki from Le Vallonnet (France; ca. 1.2 Ma; Moullé, 1992), Durfort 95 (France; ca. 1.0; Brugal, 1995), Mauer (Germany; ca. 0.6 Ma; Freudenberg, 1914), Süssenborn (Germany; ca. 96 0.6 Ma; Flerov 1969)and Isernia La Pineta (Italy; ca. 0.58 Ma; Sala, 1986; Peretto et al., 2015). Bison menneri 97 and Bison schoetenacki are characterized by derived cranial features, large body size, relatively long limbs, 98 and short and swollen horn-cores. During this same period, Bison cf. menneri and Bison (Bison) 99 voigtstedtensis have been respectively recorded from Gran Dolina TD8 and Sima del Elefante TE9c (Spain; ca. 100 1.1–0.75 Ma; Huguet et al., 2017; Van der Made et al., 2017). The taxonomic status of B. voigtstedtensis, 101 originally described on the basis of cranial material from Voigtstedt (Germany; ca. 0.7 Ma), has been 102 debated, being considered a subspecies of Bison (Bison) schoetensacki by some authors (e.g., Fischer, 1965; 103 Sala, 1986; Brugal, 1995; van Asperen and Kahlke, 2017), and a distinct species by others (e.g., Flerov, 1979; 104 Van der Made et al., 2017). 105 Only by the mid-Middle Pleistocene, the well-known steppe bison, Bison (Bison) priscus, with the two 106 subspecies B. priscus priscus and B. priscus mediator, appears in eastern Eurasia (Kahlke, 1999). This massive 107 species exhibits stouter limbs, larger head, and longer horns than earlier bison, but is overall very 4 108 polymorphic throughout its chronostratigraphic and geographic range (Kahlke, 1999). The steppe bison 109 dispersed across the whole Holarctic, reaching North America through Beringia and giving rise to the 110 American bison lineages in the Late Pleistocene, as demonstrated by fossil and molecular evidence (Shapiro 111 et al., 2004; Froese et al., 2017). In turn, the extant European wisent Bison (B.) bonasus appears during the 112 Late Pleistocene as a possible relative of the steppe bison (Soubrier et al., 2016) or, with more uncertainty, 113 to the woodland wisent, with probable introgression from the auroch (Bos primigenius), as testified by 114 recent molecular studies studies (e.g., Palacio et al., 2017; Grange et al., 2018). 115 116 2. The Vallparadís Composite Section 117 The Vallparadís Composite Section (VCS) includes the paleontological open-air sites of Cal Guardiola (CGR) 118 and Vallparadís Estació (EVT), located in the Vallès-Penedès Basin (NE Iberian Peninsula; Madurell-Malapeira 119 et al., 2010, 2017; Fig.1). During the excavations carried out between 1997 and 2008, more than 30,000 120 remains of vertebrates from the late Early to Middle Pleistocene sequences were recovered from CGR and 121 EVT. The two sites are characterized by a depositional setting influenced by the dynamics of an alluvial fan 122 system and the geometry of the Miocene paleorelief. The excavated sediments consist of debris-flows and 123 mud-flows resulting from alluvial fan system dynamics with influence of close colluvial processes. 124 Biochronological, magnetostratigraphic, and U-series-ESR data agree indicate that the VCS ranges from 125 before the Jaramillo paleomagnetic subchron (ca. 1.1–1.0 Ma) to the early Middle Pleistocene (ca. 0.6 Ma; 126 Madurell-Malapeira et al., 2010, 2012, 2014, 2017; Minwer-Barakat et al., 2011). The timespan embraced by 127 the VCS can be divided into four different time intervals (Madurell-Malapeira et al., 2010, 2014, 2017; 128 Minwer-Barakat et al., 2011): (1) pre-Jaramillo (layers CGRD1 to CGRD3); (2) Jaramillo subchron interval, 129 (layers EVT9 to EVT12); (3) post-Jaramillo, Matuyama (layers EVT4 to EVT7 and CGRD4 to CGRD8); and (4) 130 early Middle Pleistocene interval (layers EVT2 and EVT3). 131 The timespan comprised between 1.25 and 0.6 Ma was marked by the onset of new asymmetric 132 glacial/interglacial cycles that affected climate (decrease in temperatures and humidity couplet with 133 seasonality with longer and harsher winters) as well as vegetation structure (long alternations between 134 steppe and deciduous forests) in Europe—the so-called ‘Early-Middle Pleistocene Transition’ (EMPT; Head 5 135 and Gibbard, 2005; Clark et al., 2006). In the older part of the VCS (ca. 1.1 Ma) the pollen and wood analyzed 136 from CGRD2 suggest a warm-temperate and humid paleoenvironment, indicating the presence of a river or 137 river-marsh ecosystem with a variety of plant groups, from aquatic macrophytes to deciduous trees and 138 grasses. The abundance of hippo remains in this layer is consistent with the inferred fluvial/lacustrine main 139 depositional environment, and the high diversity of large-sized ungulates such as deer, horses, and bison 140 suggest the existence of a wide spectrum of different environments in the surroundings, including 141 woodlands and more open and dry areas (Mijarra et al., 2007). Meso- and microwear analyses performed on 142 a large sample of ungulate teeth from VCS indicate that, since 0.9 Ma (MIS22), the paleoenvironments 143 experienced a substantial change, from the predominance of open dry grasslands with a certain seasonality 144 (Layer EVT12, ca. 1.0 Ma; MIS31) to more humid woodlands with, possibly, an even more marked seasonality 145 (Layers EVT7 and CGRD7; ca. 0.86 Ma; MIS21), in agreement with data from other Southern European coeval 146 sites (Strani et al., 2019). 147 148 3. Materials and methods 149 The VCS bovid sample analyzed herein is housed in the Institut Català de Paleontologia Miquel Crusafont 150 (ICP), Sabadell, Spain. The complete list of specimens is reported Table S1. The stratigraphic provenance of 151 the bovid material is unbalanced: 135 specimens come from CGRD7 and EVT7 (0.86–0.78 Ma); 70 from 152 EVT10 and EVT12 (1.07–0.99 Ma); 14 from CGRD2, CGRD3, and CGRD4 (1.1–1.0 Ma), and 4 from EVT3 (<0.6 153 Ma). The descriptions of dental and postcranial features follow the nomenclature used by Masini (1989), Sala 154 (1986), Sher (1997), and Maniakas and Kostopoulos (2017a). Measurement abbreviations are explained in 155 Table 1 and shown in Fig. 2. Measurements are partially modified from Brugal (1985), Masini (1989) and 156 Maniakas and Kostopoulos (2017a). All measurements were taken with a digital caliper to the nearest 0.1 157 mm. Juvenile specimens are not included in the analyses. 158 Throughout this work we use Bison s.l. for all members of the genus Bison, and Bison s.s. for species 159 referred to the subgenus Bison (Bison), i.e., excluding those commonly attributed to Bison (Eobison). 160 Comparative material of Leptobos spp. from Upper Valdarno and Olivola (Tuscany, Italy), Leptobos aff. 161 vallisarni from Pietrafitta (Umbria, Italy) and Bison schoetensacki (Isernia La Pineta, Italy) studied by us is 6 162 housed, respectively, in the IGF, MPLB, and MPPPL—see institutional abbreviations below. Other 163 comparative data were taken from the literature. 164 Z-scores were used to compare dental measurements of the described specimens with those from other 165 samples, box plots of the ratio (%) between tooth width and length (W/L%) and bivariate plots of L vs W 166 were employed to assess the size and proportions of the molars among the different Bison s.l. populations. 167 Two shape indices were computed for the humerus to distinguish Bison from Bos and Leptobos: Stampfli’s 168 trochlea index (Stampfli, 1963), computed as the ratio (%) between DEAW and TWl; and Lehmann trochlear 169 index (Martin, 1987) computed as the ratio (%) between THl and THm (all measurement abbreviations are 170 explained in Table 1). To distinguish Bison from Bos metapodials we used an index expressed as the ratio (%) 171 between DEW and DDW (Delpech, 1972). A shape index for the metacarpals was computed as the ratio (%) 172 between PFWl and PFWlm. To quantify the magnitude of sexual dimorphism in metacarpals we applied, to 173 the samples in which putative males and females were recognized, the equation given by Schertz (1936b): 174 (x1 – x2) / x1*100, where x1 and x2 are the average male and female value, respectively, for the selected 175 variable. To identify sources of significant biometric differences within the VCS sample, we performed (1) 176 univariate ANOVA on eigth variables (raw values of Lmax, PEW, PET, DW, DT, DEW, DT) for metacarpals and 177 metatarsals from the two main chronologies of VCS (1.07–0.99 Ma and 0.86–0.78 Ma) and (2) a MANOVA on 178 the complete metapodials (metacarpals and metatarsals analyzed separatedly) based on five variables (raw 179 values of Lmax, PEW, DW, DEW, DEW/Lmax%). Moreover, biometric differences in teeth were assessed by 180 means of an ANOVA based on the W/L% ratio of M2 and M3 (the scarcity of specimens from different layers 181 prevented us to perform the same analysis on other teeth). Significance level set at 0.05. To assess the 182 stoutness of metapodials, bivariate plots of Lmax vs DEW/Lmax % were employed. Log10 ratio diagrams 183 (Simpson, 1941) were constructed based on the average values of seven selected variables of Bison samples 184 from various Eurasian sites. The extant Bison bonasus was used as a standard of comparison (y=0; data taken 185 from Reshetov and Sukhanov, 1979). Principal component analyses (PCAs) were performed based on 186 metacarpal and metatarsal bones separately to explore the main morphological differences among different 187 extinct Bison forms and series. Two sets of variables were used. The first approach relies on seven Mosimann 188 shape variables (Table 2) obtained by log-transforming the ratio between each measurement and the 7 189 geometric mean of the seven measurements for each specimen (Jungers et al., 1995). The second approach 190 considers eight variables calculated following Scott and Barr (2014), i.e., adjusting each measurement as the 191 log transformed ratio between the measurement and Scott’s (2004) metapodial global size variable (MGSV). 192 MGSW = ((PEW * PET * DW * DT * ABETI * AETm * DEW * (DETm * DETl)1/2)1/9. For consistency, a second 193 multivariate approach was used with the computation of a MANOVA on seven raw variables (Lmax, PEW, 194 PET, DW, DT, DEW, DET). Significance level set at 0.05. 195 Statistical computations were made with PAST v. 3 (Hammer et al., 2001). 196 Institutional abbreviations: AUTH, Aristotle University of Thessaloniki, Greece; GNM, Georgina National Museum, Tblisi (Georgia); 197 HLMD Hessisches Landesmuseum, Darmstadt (Germany); ICP, Institut Català de Paleontologia Miquel Crusafont, Sabadell (Spain); 198 IGF, Museo di Storia Naturale, Sezione di Geologia e Paleontologia, Università di Firenze (Italy); IPHES, Institut Català de 199 Paleoecologia Humana i Evolució Social, Tarragona (Spain); IQW, Senckenberg Research Station of Quaternary Palaeontology, 200 Weimar (Germany); IVPP, Institute of Vertebrate Paleontology and Paleoanthropology, Beijing (China); MCM, Musée Cuvier de 201 Montbéliard (France); MNHN, Muséum national d'Histoire naturelle, Paris (France); MPLB: Museo Paleontologico Luigi Boldrini” di 202 Pietrafitta (Italy); MPPPL, Museo di Paleontologia e Preistoria Piero Leonardi, University of Ferrara (Italy); MPRM, Musée de 203 Préhistoire Régionale, Menton (France); MuPA, Museo Archeologico e Paleontologico, Serravalle del Chienti (Italy); NAS, National 204 Alliance of Shidlovskiy “Ice Age Period” “Ice Age Museum”, Moscow (Russia); UNIFE, Prehistoric Sciences section of the Department 205 of Humanistic Studies, University of Ferrara (Italy). 206 207 4. Systematic paleontology 208 209 Order Artiodactyla Owen, 1841 210 Family Bovidae Gray, 1821 211 Subfamily Bovinae Gray, 1821 212 Genus Bison Hamilton Smith, 1827 213 Subgenus Bison Hamilton Smith, 1827 214 Bison (Bison) schoetensacki Freudenberg, 1914 215 Figs. 3–8; Figs. S1–3 216 8 217 4.1. Referred specimens 218 The VCS sample includes a total of 90 cranial (Figs. 3–4) and 130 postcranial (Figs. 4–8; Figs. S1–3) 219 remains. See Table S1 for details. 220 221 4.2. Description 222 4.2.1. Horn cores 223 DescriptionOnly a well-preserved apical portion of a horn core (IPS92970; Fig. 4c) and several 224 fragments of a basal part of a second horn (IPS92971) were recovered from EVT7 and EVT10, respectively. In 225 IPS92970 the tip curves markedly upward, the section is subcircular, slightly dorso-ventrally compressed in the 226 basal portion. In dorsal view, the horn core is slightly curved backward along the longitudinal axis. The 227 longitudinal grooves along the surface are quite shallow and more pronounced on the ventral side in the 228 proximal portion. Narrower, shorter, and shallower furrows are present on the dorsal portion of the tip. 229 IPS92971 is severely fragmented. The estimated large diameter and the presence of sinuses on the internal 230 side of some fragments suggest that the remains represent the basal portion of the horn core. Relatively deep 231 furrows are present on the external ventral surface. No further characters are recognizable. 232 Remarks and comparisonsBovid horn cores are very variable, also within Bison.Early bison species 233 (e.g., B. menneri) are characterized by short horns that emerge backward (i.e., inserted caudally) and 234 commonly taper abruptly, whereas the more derived species (e.g., B. priscus) have longer, laterally inserted, 235 and more gently tapering horn cores (Sala, 1986). Bison schoetensacki shows an intermediate morphology. 236 Even if these features are relatively stable, it must be taken into account that the size, shape, and position of 237 horn cores in Bison spp. are strongly influenced by sexual dimorphism and ontogenetic stage. This may cause 238 erroneous taxonomic attributions of fossils, especially based on small samples. The horn cores from VCS are 239 too fragmentary to substantiate a determination to species rank, mainly because their basal portion is not 240 preserved. However, the two available specimens (especially IPS92970) exhibit some features (ventral furrows 241 in the middle portion of the horn core, smaller dorsal furrows on the tip, dorsoventral compression of the horn 242 core, and backward bending of the tip) that have previously been described by Sala (1986) and Brugal (1995) 9 243 for B. schoetensacki. However, at the state of the art, we cannot exclude that similar characters were also 244 present in some species of Bison (Eobison), for which the morphology of the horn cores is still largely unknown 245 246 4.2.2. Dentognathic remains 247 The studied specimens include a large number of isolated teeth, a few fragmentary mandibles, and a 248 maxillary fragment (Figs. 3–4; Table 3). Almost all the specimens come from layer EVT7. The cheek teeth are 249 typically bovine in overall morphology, mesiodistally elongate, moderately hypsodont, and have well- 250 developed styles/stylids. 251 Description of the upper dentitionThe P2 is mesially tapering and narrow relatively to length. The 252 parastyle is pointed, the paracone rib is relatively marked, and the metastyle is faint (Fig. 3f). The P3 is 253 mesiodistally elongate (Fig. 3g) whereas the P4 has squarerocclusal contour (Fig. 3h). In the P3 and P4, the 254 parastyle and metastyle are similar in size and somewhat protruding on the labial side (more markedly in the 255 P4). The paracone rib is not particularly protruding. A small fold can be observed in the inner recess on the 256 lingual margin of the P3. In the M1 the mesial lobe is slightly narrower buccolingually than the distal one, 257 whereas in the M2 the two lobes are similar in width, and in the M3 the distal lobe is markedly narrower. The 258 cement is visible in most of the molars, concentrated on the lingual walls and less abundant on the buccal side 259 (e.g., IPS93005). The styles are strong and prominent, with the parastyle and metastyle being the equally 260 developed. The mesostyle, which is somewhat distally oriented, is generally the most developed. In the M3, 261 the metastyle protrudes distally. The entostyle is relatively high and cylindrical in lingual view, and in most 262 cases shows an ovoid occlusal section (only in a few specimens it has an irregular occlusal outline). In most 263 specimens, the cement penetrates deeply between the entostyle and the two lobes, isolating the entostyle 264 from the protocone and the metaconule. In some specimens, the central cavities have a simple crescent- 265 shaped enamel outline, while in other cases the enamel shows a fold in the middle of the distal wall of the 266 central cavities. This “bubaline fold” (after Masini, 1989), if present, is deeper in the distal central cavity. A 267 small enamel islet is present between the two lobes in several specimens (e.g., IPS93032, IPS93018). 268 Description of the lower dentitionOnly two p4 are preserved in the entire sample. The parastylid is 269 large and relatively sharp, curved toward the lingual margin. The lingual wall has three vertical grooves of 10 270 varying depth. The entoconid is well developed and slightly curved distally. The distal margin of the tooth is 271 flat. The molars have cement both lingually and buccally. Like in the upper molars, the cement is concentrated 272 on the lingual inner walls between the protoconid and hypoconid. On the buccal wall, cement is absent in most 273 specimens. The stylids are very prominent; the parastylid and entostylid (especially in the m1 and m2) are 274 similar in size; in the m3, the parastylid is prominent (mesially projecting). The m3 ectostylid, located between 275 the protoconid and the hypoconid, is slightly mesiodistally flattened, with an ovoid occlusal outline; in some 276 teeth, the ectostylid is located on the distal portion of the protoconid, whereas in others it is more shifted 277 distally, between the two lobes. The recess filled by the enamel between the entostylid and protoconid is 278 deeper and generally narrower than that of the hypoconid internal flange. In the less worn m1 and m2, the 279 entostylid and the parastylid are slightly bent distally. In the m3, the labial recess between the hypoconulid 280 and the hypoconid is deep and narrow. 281 Remarks and comparisons82 out of 86 dental remains from the VCS were unearthed from layer 282 EVT7, dated to ca. 0.86 Ma. The teeth are relatively homogeneous both in size and shape. This is also confirmed 283 by the one-way ANOVA results performed on the W/L% ratio of M2 and M3 from the two main VCS 284 chronologies, that is, sin-Jaramillo (EVT10 and EVT12) and post-Jaramillo (CGRD7 and EVT7) (Table 4). 285 Generally, their morphology is typically “bisontine”, with a square and buccolingually wide occlusal contour, 286 especially the upper teeth. Decades ago, bison tooth size and shape were considered taxonomically relevant 287 (e.g., Merla, 1949; Flerov, 1969; Sala, 1986). However, based on our experience, we concur with Sher (1997) 288 that tooth morphology is strongly influenced by the degree of wear and, hence, is not particularly diagnostic, 289 especially when working with small samples. Nevertheless, some characters such as enamel penetration in the 290 inner lingual wall of the upper molars, entostyle development, and tooth size overall can give some taxonomic 291 hints. The presence of a “bubaline fold” in the upper molars from the VCS is variable, as in other bovines (e.g., 292 it is absent in Proamphibos, but variably developed in Leptobos, and subject to individual variation in Bos and 293 Syncerus; Merla, 1949), and thus of little taxonomic value. The VCS upper molars are distinctly swollen just 294 above the cervix, as often in Bison, but unlike in Bos (where this feature is almost entirely absent; Sala, 1986). 295 However, a less-developed swelling is present in some Leptobos specimens from Upper Valldarno and Olivola, 296 thus only unequivocally distinguishing Bos from Leptobos and Bison. In contrast, cement is commonly present 11 297 on the lingual side of the teeth from the VCS (especially in old individuals), and even on the buccal side in some 298 specimens, whereas in Leptobos cement is almost entirely absent (Masini, 1989; Demirel and Mayda, 2014). 299 The abundance of cement is very variable among European Bison populations (Sher, 1997)—e.g., it is lacking 300 in B. priscus from Taubach but present in B. menneri from Untermassfeld. The presence of an enamel islet in 301 the upper molars between the protocone and the hypocone distinguishes B. (Eobison) spp., B. menneri and B. 302 schoetensacki from B. priscus and Bos primigenius, where it is rare (Prat, 1968; Sala, 1986) although these 303 islets are relatively common in recent B. bonasus populations. The VCS sample displays this character in almost 304 all the teeth with a medium-advanced wear stage, fitting the morphology of B. schoetensacki from Isernia and 305 other European sites (Sala, 1986). 306 Average values of tooth length and width for the analyzed Bison samples (Table S2) show that there is 307 a general overlap between all the considered species. The L vs W diagram and the box-plots of the ratio W/L% 308 of m3, M1, M2, and M3 (Fig. S1) help us to compare the size and proportions of Bison s.l. teeth. The extremely 309 massive B. priscus from Taubach features large teeth (especially m3) (Fig. S1a–d) but with low values of W/L%, 310 i.e., teeth are narrow as compared with total length (Fig. S1e–h). Bison cf. priscus from Westbury shows long 311 and very wide teeth (Fig. S1a–d). The B. schoetensacki specimens from Le Vallonnet, Cromer Forest-bed and 312 the few remains from Durfort are characterized by general smaller size (especially the M1s from Durfort; Fig. 313 S1b), but higher values of W/L%, i.e., relatively more squared teeth (Fig. S1e–h). The Isernia m3s and M3s are 314 quite stout and relatively large (Fig. S1a, d, e, h), whereas the Süssenborn specimens are elongated with very 315 low values of W/L%, similar to those of B. menneri from Untermassfeld and B. priscus from Taubach (Fig. S1a- 316 h). The B. menneri remains display a high degree of variation but, generally, are among the narrowest teeth 317 analyzed (Fig. S1). The bovid from Mygdonia basin also shows high variation (see, for instance, the M1 biplot 318 and box-plot; Fig. S1b, f), overlapping with almost every other sample. The B. (Eobison) teeth from Pirro are 319 among the shortest and stoutest (Fig. S1). The teeth from VCS appear short and relatively wide, compared 320 with the othersamples examined, showing proportions similar to those of B. schoetensacki from Le Vallonnet, 321 Cromer forest-bed, and Durfort (Fig. S1). On the other hand, the VCS teeth differ from those from Isernia and 322 Süssenborn, which feature larger size and more elongated proportions. Some specimens from VCS (e.g., the 323 single M3 from EVT3; IPS93023) display a quite short and stout morphology, similar to that shown by the older 12 324 remains of B. (Eobison) degiulii from the Italian Peninsula. The z-scores computed for the VCS sample (Table 325 S3) show negative values in most instances for the length of the upper molars and the m3 compared with the 326 other Bison populations, being similar to the specimens from Venta Micena, Mygdonia basin, and Le Vallonnet. 327 On the contrary, for the width, z-scores are positive in most cases, showing that the VCS sample has relatively 328 wide molars, also compared with the large form of B. priscus from Taubach (Table S3, Fig. S1). 329 330 4.2.3. Vertebrae 331 DescriptionAn axis (IPS114551), another cervical vertebra (IPS107615), and four fragmentary 332 thoracic vertebrae (IPS92954, IPS92955, IPS92956, IPS118117) were recovered from VCS (Fig. S2; Table S4). 333 The axis is quite high and elongate, and displays a low spinous process directed dorsally. The transverse 334 foramina are small and located on the posterior portion of the lateral expansions. The neural canal is teardrop- 335 shaped in anterior view. The lateral expansions of the anterior articulation are quite prominent and have a 336 circular anterior outline. The anterior part of the neural process and the transverse processes are broken. 337 The cervical vertebra IPS107615 is short and massive, with subcircular neural canal, and robust pre- 338 and postzygapophyses. Most of the spinous and transverse process and part of the vertebral body are 339 missing. Its general morphology indicates that it could be one of the last cervical vertebrae due to the 340 anteriorly prominent prezygapophysis, the enlarged postzygapophysis, the position of the transverse process 341 (directed anter-oposteriorly to dorso-ventrally) and the reduced posterior tuberculum of the ventral crest. 342 The thoracic vertebrae display smaller bodies with subcircular anterior and posterior rib facets. The 343 neural canal is large and slightly compressed dorso-ventrally. The left transverse process is broken in IP92954, 344 and both tranverse processes are missing in IPS92956 and IPS92954. In IPS92956, the foramina on the left side 345 are absent, but there is a narrow, deep groove between the transverse process and the posterior rib facet. 346 The spinous process is inclined posteriorly; in IPS92954, it is almost complete (about 400 mm high). 347 Remarks and comparisonsThe five vertebrae from VCS have the typical morphology of large bovids. 348 It is noteworthy that the extremely long spinous process of the thoracic vertebra IPS92954 exceeds the height 349 recorded for any species of Leptobos and Bos. According to Sher (1997), the tallest spinous process recorded 13 350 for B. menneri is ca. 360 mm, whereas in some specimens of B. priscus from Siberia it exceeds 600 mm. Apart 351 from this character, no other vertebral features are helpful in distinguishing Bison from either Leptobos or Bos. 352 353 4.2.4. Scapula 354 DescriptionSee Fig. 4d and Table S5. IPS107637 is almost complete, although slightly fragmentary 355 and lacking the distal margin, while IPS92922 only preserves the glenoid cavity and a proximal portion of the 356 blade. The glenoid cavity is subovoid, with its major axis oriented antero-posteriorly. The lateral margin of the 357 cavity is sligthy concave in distal view. The acromion, partially preserved in both specimens, is not particularly 358 pointed. The spine, almost complete in IPS107637, is relatively sharp in the proximal half and constituted by a 359 flat surface in the distal half (about 80 mm of width), which displays a small, shallow groove that runs along 360 the lateral surface up to the distal margin. The incisura scapulae is straight and relatively deep. The 361 supraglenoidal tubercle (only preserved in IPS92922) is large, displays a rough surface, and slopes dorsally 362 from the edge of the glenoid. There is a narrow groove on the medial surface of the collum scapulae between 363 the glenoid cavity and the supraglenoidal tubercle. The infraglenoidal tubercle is developed postero-ventrally 364 and located on the medial side of the bone, just above the glenoidal margin. 365 Remarks and comparisonsThe two scapulae from VCS closely resemble each other in size and shape, 366 and fit well with the morphology of large Bovids, especially Bison (eg., glenoid cavity more rounded in Bos). 367 The dimensions of the glenoid cavity in the VCS specimens fall within the variation ranges of the smallest 368 species of Bison (i.e., B. menneri and B. bonasus; Table S6). 369 370 4.2.5. Humerus 371 DescriptionSix humeri from the VCS are available (Fig. 5a–b; Table S7). They only preserve the distal 372 epiphisys and, in some cases, part of the diaphysis. The distal epiphysis is massive and has the typical bovid 373 morphology, with a large trochlea divided by a crest shifted toward the lateral margin. The medial margin of 374 the trochlea is proximodistally higher than the lateral, thus the trochlea in anterior view tapers lateralward. 375 The large groove and the crest that separate the two epitrochleae are located slightly lateral to the distal 376 articulation midline. On the lateral margin of the trochlea, a large and bulging trochlear crest is present. The 14 377 coronoid fossa is a mediolaterally wide. The olecranon fossa is deep, relatively narrow, and medially and 378 laterally delimited by strong crests that slightly converge distally. On the lateral surface of the shaft, at the 379 contact between the diaphysis and distal epiphysis there is a rough half-moon-shaped ridge. The lateral 380 trochlear pit (which is absent on the medial side) is large and deep. 381 Remarks and comparisonsThe VCS humeri are morphologically homogeneous, except for the 382 differences in overall size and robusticity, which are attributable to sexual dimorphism. The presumably male 383 specimens IPS50672 and IPS114549 display a larger distal epiphysis than the rest of the sample, which are 384 likely female. The humerus of Bison s.s. generally resembles that of Leptobos and B. (Eobison), except for the 385 larger distal ends of the former and some small differences in the trochlear elements (see below). The distal 386 epiphysis of the VCS humeri falls within the size variation of Bison, in particular B. schoetensacki and B. 387 menneri, being slightly smaller compared to B. priscus (Table S8). Several studies have attempted to find 388 diagnostic characters in the humerus, focusing on the distal epiphysis, to distinguish Bison from other large 389 bovids such as Bos. The humeri of Bos and Bison are markedly polymorphic (Martin, 1987; Sher, 1997), but 390 according to Martin (1987) the two genera differ in the proportions between the medial and lateral 391 epitrochleae (affected by the position of the trochlear crest), the shape of the trochlear crest, and the shape 392 of the olecranon fossa. In particular, Bison has parallel trochlear margins and a smoother outline of the 393 trochlear crest, whereas Bos has a higher crest, a deeper trochlear groove, and generally irregular dorsal and 394 ventral edges of the trochlea (Stampfli, 1963; Sala, 1986). However, we concur with Sher (1997) that this last 395 difference is not reliable. The VCS specimens show a wide range of variation in this character (e.g., sharp in 396 IPS50672 and smooth in IPS107620), indicating it has no diagnostic value. The proportions between the lateral 397 and medial epitrochleae, as measured by Stampfli’s trochlea-index (Stampfli, 1963), is extremely variable in 398 the VCS sample and the analyzed comparative sample (including Leptobos, Eobison, Bison, and Bos), indicates 399 great overlapping (Table S9). The lateral tapering of the trochlea has been proposed as a diagnostic character, 400 as measured by the Lehmann trochlear index (Table S10), although it is proved similarly undiagnostic (Sher, 401 1997). In the VCS sample, this index has the widest range of variation among the analyzed samples (ranging 402 between 62.9 and 76.5), although on average it most closely resembles the values displayed by the smallest 403 taxa (Eobison and B. menneri; Table S10). 15 404 4.2.6. Radius 405 DescriptionRadii are among the most common and best-preserved long bones found in VCS (Fig. 406 5c–d and Table S11). The ulna is fused with the radius in all studied specimens except IPS39893 and IPS92916. 407 The lateral tuberosity of the radius, which has an antero-posterior diameter roughly similar to the proximo- 408 distal one, forms an obtuse angle with the distal epiphysis. The dorso-medial and medial portions of the 409 proximal epiphysis form a clearly protruding shelf with a square proximal outline. The posterior contour of the 410 proximal articulation is slightly undulated, forming an obtuse angle with the lateral edge. The outline of the 411 lateral glenoid cavity is dorsally convex, being more or less deep depending on the specimen (e.g., gentle and 412 shallow dorsal convexity in IPS13570, deep and almost angular in IPS48775). Distal to the lateral glenoid cavity, 413 on the lateral part of the posterior portion, there is a wide depression of variable depth and, medially, a rough 414 area. The notch between the two proximal glenoid cavities is wide and shallow (V-shaped in IPS48775). When 415 preserved, the articular surface for the ulna is well visible in the middle of the proximal diaphysis of the radius, 416 being wide and rough. On the anterior surface of the distal epiphysis there is a grooved area for the tensor 417 tendon, parallel to the diaphysis and delimited medially and laterally by blunt crests. The distal epiphysis is 418 inclined posteriorly and shows a circular pit at the medio-distal end, surrounded by radial rugosities. The 419 fissure marking the contact between radius and ulna in the distal diaphysis is well visible (e.g., in IPS107628 420 and IPS92919). The articular surfaces for the scaphoid and semilunar are particularly deep in the anterior 421 portion, edged posteriorly by two pits. Between the ulnar articular surface and and the pointed styloid process 422 there is a wide and deep notch. On the medial side of the distal end there is a small bulge for the metacarpal 423 carpal ligament, more developed in the largest specimens (IPS92916, IPS39893, and IPS92918) than in the rest 424 of the sample. 425 Remarks and comparisonsThe radii from the VCS show the typical morphology of Bison. They can 426 be distinguished from the ones of Bos based on several characters, including: the poorly developed lateral 427 tuberosity of the proximal epiphysis, the wide and shallow notch between the two proximal glenoid facets 428 with a smooth posterior outline, and a fissure marking the contact between ulna and radius (Brugal, 1983; 429 Sala, 1986; Gee, 1993; Sher, 1997). These characters are also present in Leptobos spp. (Masini, 1989) which 430 nevertheless show smaller and slenderer radii. The VCS radii display stout proportions as showed by the 16 431 diagram of stoutness (built as the ratio between proximal epiphysis width and total bone length VS the total 432 length; Fig. 9c), especially when compared with Leptobos and Eobison. The VCS radii are similar in size to those 433 of B. menneri and B. schoetensacki. The aforementioned index of stoutness indicates that VCS specimens 434 overlap with those of B. schoetensacki, being slenderer than those of B. priscus (which is characterized by a 435 very stout radius, i.e., wide diaphysis and epiphyses) but slightly more robust than those of B. menneri (which 436 displays similarly long radii but with lower width values; Table S12). 437 438 4.2.7. Ulna 439 DescriptionThree almost complete ulnae are available from EVT and CGR (IPS39893, IPS107616, 440 IPS107617; Fig. 5c–d; Table S13), all of them paired with the corresponding radii. The olecranon process is 441 large and square, gently inclined posteriorly with respect to the long axis of the bone. The medial and lateral 442 surfaces of the oleocran are slightly concave. The ulnar shaft is posteriorly concave. The medial articular facet 443 for the radius is subrectangular and concave. The radius lateral articulation is latero-medally wide, divided into 444 two facets by a small crest, one distal to the other. The proximal facet is subtriangular and almost flat, while 445 the distal facet is subcircular and concave. The semilunar notch is large and has a small pointed tip at the distal 446 margin which reaches the medial articulation facet for the radius. 447 Remarks and comparisonsThere are no substantial differences between the VCS ulnae and those of 448 Bos and Bison (Table S14). Even if some authors noted some minor differences between these genera 449 (Bibikova, 1958; Gee, 1993; Brown and Gustafson, 2000), the only clear difference relates to the articulation 450 between the radius and the ulna (Gee, 1993; Sher, 1997): in Bison, the lateral articulation with the radius is 451 wide and does not intrude into the postero-proximal part of the radius, whereas in Bos this facet is shaped as 452 an elongated triangle that deeply intrudes into the radius. The VCS sample displays the former condition, 453 although with some degree of variation, as already noticed in Bison spp. by Sher (1997). 454 455 4.2.8. Carpal bones 456 Only very few carpal bones are available from the VCS collection (Fig. S3; Table S15), mostly from layer 457 EVT10 (IPS92960, IPS92961, IPS92962, IPS92963). Morphological description and comparisons of large bovid 17 458 carpal bones are often neglected in literature (Sher, 1997); moreover, the few studies performed are 459 sometimes contradictory and do not agree on the measurements to be taken (Bibikova, 1958; Stampfli, 1963; 460 Sala, 1986; Sher, 1997). For these reasons, only a short description and comparison are presented for these 461 anatomical elements. The lack of Leptobos material precludes comparisons with this genus, while no 462 differences in size or shape were found among Bison species. 463 Pyramidal (carpi ulnare)The laterally S-shaped morphology of the pyramidal, with an elongated 464 distal portion, is typical of Bos and Bison. According to Sala (1986), the proximal margin of the facet for the 465 unciform in Bison has a smaller radius of curvature as compared with Bos (Stampfli, 1963; Sala, 1986). This is 466 disproved by the two pyramidal bones from the VCS, which display a quite wide radius of curvature, as well as 467 by the pyramidal of B. menneri figured by Sher (1997), who emphasized the high degree of variation in the 468 population from Untermassfeld. 469 Semilunar (carpi intermedium)The only specimen recovered from VCS is partially broken. Its 470 morphology does not differ from that of Bison and Bos. Sala (1986) evidenced some differences regarding the 471 proximal and distal facets, which are hard to recognize in our sample due to the fragmentary state of the fossil. 472 The comparison of IPS92961 with the Bos and Bison spp. semilunars figured by Sala (1986: Figs. 11–12) does 473 not highlight appreciable differences, although the VCS single specimen more closely resembles the B. 474 schoetensacki sample from Isernia. 475 Unciform (os carpale IV) According to Sher (1997), this bone is extremely variable in size within a 476 single population of Bison. This is confirmed by the two specimens from VCS, which are similar in shape, but 477 differ in size (IPS92962 is about 25% larger than IPS92964). This variation might be due to sexual dimorphism. 478 Their morphology fits well with that described for Bison (Sala, 1986), being quite different from the one of Bos 479 (i.e., the more rounded shape of the axial edge in the proximal articular facet of Bison). 480 Capitatotrapezoid (os carpale II+III)The poor state of preservation of the single available 481 capitatotrapezoid does not allow a detailed description or comparisons. 482 483 4.2.9. Metacarpal 18 484 DescriptionAll the VCS metacarpals are stout (Fig. 6a–c; Table 5). The proximal and distal epiphyses 485 have almost the same width. Four specimens (of which two complete: IPS107626, IPS92910 from EVT12) are 486 slenderer than the remaining ones, most probably due to sexual dimorphism (see Section 6.3.). The proximal 487 articular surface displays a D-shaped outline in proximal view. The lateral articular facet is triangular and 488 located on a lower plane than the medial articular facet, being separated from it by a high crest oriented 489 antero-posteriorly. In some specimens the lateral facet is characterized by a small depression on the anterior 490 margin (e.g., IPS92907, IPS92909, IPS92908). The medial facet is larger and subsquare. Its medial outline varies 491 from subcircular to square. The synovial fossette is a variably developed irregular pit located in the middle of 492 the medial facet (not visible or very small in IPS92912, IP92911, and IPS14985). The proximal epiphysis is 493 marked by a bulging ridge that follow the articular surface, interrupted in the posterior margin by a U-shaped 494 small notch that corresponds to the proximal nutrient foramen, which is located at the posterior margin of the 495 crest that separates the two proximal facets. On the lateral margin of the ridge cited above, there is a 496 subrectangular facet for the fifth metacarpal. The postero-proximal nutrient foramen is teardrop-shaped and 497 located in the middle of a deep rough depression. This depression does not extend onto the posterior portion 498 of the diaphysis, which is almost flat. In anterior view, the diaphysis is generally wide and hourglass-shaped, 499 with the narrowest mediolateral diameter located in the distal half of the diaphysis. The vascular groove on 500 the distal portion of the diaphysis shows a great variation, from narrow and shallow to wide and deep; its 501 deepest part is located just proximally to the distal epiphysis, and the groove always stops distally to the 502 midshaft. The distal anterior foramen is always present, although variable in size, located within the vascular 503 groove just above the distal end of the groove itself. The distal portion of the diaphysis displays marked 504 epitrochlear tubercles (sensu Sher, 1997), so that the maximum width between them almost equals that 505 measured across the trochleae. There is a constriction between each tubercle and the distal end of the 506 corresponding trochlea. The intertrochlear margins tend to converge distally. The distal trochlear crests 507 (intercondylar crests) are subparallel to distally convergent. Two deep depressions are present proximally to 508 either side of each trochlear crest. The two outer depressions are larger and deeper than the inner ones. The 509 lateral and medial trochlear pits are deep and marked by radial rugosities. 19 510 Remarks and comparisonsThe metacarpals from VCS are characterized by similar morphologies with 511 some exceptions. Two specimens (IPS107626 and IPS92910) from EVT12 are particularly slender, with 512 narrower diaphysis and epiphyses than most of the sample, although two proximal epiphyses (IPS14102 and 513 IPS14702) from CGR7 show similar dimensions. This is most likely attributable to sexual dimorphism, which is 514 quite common in bovids and normally implies that females have slenderer metapodials than males (Sher, 1997; 515 Brugal and Fosse, 2005; Kostopoulos et al., 2018). The one-way ANOVA test, performed on 8 eight selected 516 varaibles of the specimens coming from the two main VCS chronologies of VCS (EVT10-EVT12 and CGRD7- 517 EVT7) shows that, except for the maximum length, no significant differencies were found (Table 4). It is 518 noteworthy that one of the specimens (IPS14917) from layer CGRD2 displays a pathological bone outgrowth 519 (of unknown etiology) along the medial margin of the distal shaft, which reaches the epiphysis and expands in 520 a large and thick bone outgrowth on the posterior portion of the diaphysis. A similar malformation is also 521 present in one intermediate phalanx (IPS16778) from the same layer and given their corresponding size it 522 might belong to the same individual. A MANOVA between the two main subsamples from VCS (i.e., specimens 523 divided by stratigraphic provenance), performed on five variables, does not evidence significant differences 524 between the two groups (Table 4). The aforementioned metacarpal IPS14917 was not included in the ANOVA 525 and MANOVA because it is the only metacarpal from the pre-Jaramillo layers. Nonetheless, the “stoutness” 526 diagram and the PCA clearly show the overall similarity between this specimen and the rest of the sample 527 (particularly the specimens from EVT7-CGRD7; see below). 528 Given their abundance in the fossil record and distinctive features, metacarpals are the most 529 diagnostic postcranial bones for bovines and thus those most frequently used for taxonomic purposes in the 530 literature (Table S16). Bison s.l. is readily distinguished from Bos based on the diaphysis distal end (contact 531 between the diaphysis and distal epiphysis; Schertz, 1936a, 1936b; Lehman, 1949; Bibikova, 1958; Stampfli, 532 1963; Ayrolles, 1973; Brugal, 1983, 1985, 1995; Sala, 1986; Gee, 1993; Sher, 1997). Our analysis performed on 533 several samples of Pleistocene large bovids shows that the ratio between DDW and DEW (Delpech, 1972) is a 534 very useful tool to distinguish Bos and Bison/Leptobos, with the former showing significantly lower values 535 (Table S17). In Bison, this area has marked tubercles, so that the width of the bone is roughly equal to, and 536 sometimes even greater, than maximum distal width; on the contrary, in Bos the epitrochlear inflation is less 20 537 pronounced and the distal end gently curves, following the distal margins of the hourglass-shaped diaphysis 538 being quite medio-laterally thinner than the distal epiphysis width. According to our observations, the Bison- 539 like distal inflation is also present in the Leptobos sample. Despite some minor individual variation, the 540 specimens from VCS resemble the condition of the Bison/Leptobos group. According to many authors (e.g., 541 Schertz 1936a, b; Bibikova, 1958), the medial facet of the proximal articular surface would be quadrangular in 542 Bos, and more rounded and medio-laterally developed in Bison. However, this feature is very variable (Sher, 543 1997), as shown by the VCS sample, where it ranges from rounded (e.g., IPS13928, IPS13547) to square (e.g., 544 IPS14072, IPS92913). Nevertheless, the ratio between the diameters of the lateral and medial facets 545 distinguishes Bison from Leptobos, with the latter displaying a relatively wider lateral facet. The Bison sample 546 from VCS overlaps the mean values of the Bison s.l. group (Table S18). 547 The VCS metacarpals, despite some variation, are quite heavily built, especially the male specimens. 548 They are more robust than those of B. menneri, and longer and stouter than those of Leptobos or B. (Eobison) 549 spp., but less massive than those of B. priscus, and most similar to those of B. schoetensacki (Table S16). Our 550 “stoutness” diagram (Fig. 9a) indicates that, despite some overlap, the VCS metacarpals (especially the 551 specimens coming from EVT7, CGRD2, and CGRD7 from EVT12) fall within the variation range of B. 552 schoetensacki from various localities (Süssenborn, Mauer, Mosbach, Le Vallonnet, Durfort, and Cromer Forest- 553 bed) although some large specimens (IPS107636, IPS107635, and IPS92907) also resemble those of female B. 554 priscus. The curves established by the Log10 ratio diagrams (Fig. 9e–g) confirm these results, as the VCS bison 555 is characterized by a trend that closely resembles B. schoetensacki samples. 556 The PCAs highlight the considerable overlap between different Bison samples. In the first PCA (Fig. 557 10a, Table S19), PC1 (60% variance) is mostly driven by Lmax (negative scores) and, to a lesser extent, mvDW, 558 mvDEW and mvPEW (negative scores), separating slender with relatively elongated metacarpals, toward the 559 negative values, from stout and short metacarpals at the opposite portion of the diagram, characterized by 560 robust structures. Bison priscus and B. schoetensacki display the most positive scores, whereas Leptobos and 561 B. (Eobison) spp. show the most negative values. PC2 (14% variance) is mostly driven by mvDT and mvDW 562 (negative scores) and by mvPET, mvPEW, mvDET and mvDEW (positive scores); thus, the specimens with wide 563 and thick shaft have negative values, while the ones with a narrow diaphysis but robust epiphyses are 21 564 characterized by positive values. The VCS sample (positive, for PC1 and negative for PC2, in all the male 565 specimens except for IPS14917 ad IPS13928 which have positive PC2 values) is characterized by stout 566 metacarpals with a massive distal epiphysis and relatively large diaphysis. The specimens overlap with the B. 567 schoetensacki in both PCs. The two VCS female specimens are located at the edges of the B. schoetensacki 568 convex hull, close to the Durfort and Süssenborn female specimens and overlapping with the Eobison group. 569 The second PCA (Fig. 10b, Table S20), based on Scott and Barr’s (2014) method, shows no substantial 570 differences relatively to the first PCA. PC1 (58% variance) is affected principally by reLmax and reDW, and 571 separates long metacarpals with a relatively narrow diaphysis (positive scores) from shorter but stouter 572 metacarpals (negative scores). PC2 (14% variance) is mostly influenced by reLmax and the stoutness of the 573 shaft (reDW and reDT) regarding positive scores, while it is negatively affected by the dimensions of the 574 proximal epiphysis (rePET and rePEW), thus segregating long metacarpals with a wide diaphysis (positive 575 scores) from short metacarpals with a massive proximal epiphysis (negative scores). The VCS specimens display 576 negative scores for both PCs, except for the single female specimen (IPS107626, which displalys a positive for 577 PC1), and cluster with the B. schoetensacki and B. priscus scatter of points. The only exception is IPS13928, 578 which is characterized by an extremely stout distal epiphysis and hence overlaps with the lower range of B. 579 priscus from UK for PC2. The main convex hull of VCS is bordered below by the single metacarpal from 580 Süssenborn and above by the remains from Le Vallonnet and Mauer/Mosbach. The MANOVA performed 581 among the VCS sample, other species of Bison s.l., and Leptobos etruscus shows that the Iberian sample is 582 characterized by significant differences from all the others, with the exception of those attributed to B. 583 schoetensacki (Table S21). 584 585 4.2.10. Tibia 586 DescriptionA partial tibia (IPS107618) lacking the proximal epiphysis is the most complete specimen 587 from the VCS, whereas the rest of the sample is composed of distal fragments (Fig. 8; Table S22). The medial 588 malleolus is variably developed (from poorly marked in IPS92943 to strongly developed in IPS92942). It extends 589 distally below the trapezoidal process, separating the two articular grooves. The area immediately anterior to 590 the medial malleolus is characterized by a raised tubercle. The anterior malleolar facet is inclined anteriorly. 22 591 The posterior surface of the distal diaphysis is slightly convex. Between the medial malleolus and the 592 subtriangular process separating the two articular grooves, there is a shallow furrow that extends up to the 593 shaft ending distally to the midshaft (in IPS92942, this furrow is deeper than in the others and delimited by 594 two crests, the medial one being sharper). The lateral margins of the two malleolar facets are concave dorsally 595 and separated by a deep U-shaped notch. The lateral articular groove is antero-posteriorly longer but 596 mediolaterally narrower than the mesial one. The anterior margin of the lateral articular groove is pointed and 597 more anteriorly protruding than the mesial groove. Conversely, the posterior margins of the two grooves lie 598 almost at the same level. The postero-medial corner of the medial groove forms a right angle. In all the 599 specimens but IPS92943, on the posterior margin of the medial malleolus there is a marked a step. The lateral 600 malleolus is more developed than the medial and hosts the anterior and posterior malleolar facets. The 601 subcircular anterior facet is small and slightly concave, while the subtriangular posterior facet is larger and 602 more concave than the anterior one. 603 Remarks and comparisonsThe VCS tibiae attest to marked sexual dimorphism, with male specimens 604 (IPS107618, IPS92942, IPS114546) being larger and more robust than female ones (IPS92940, IPS92943). Gee 605 (1993) recognized several diagnostic criteria between the tibiae of Bos from Bison. However, we found that 606 the most reliable lies in the two facets for the malleolus (Brugal, 1985; Sala, 1986; Sher, 1997). In Bos, the 607 facets are often confluent, and the anterior one is quite smaller and flatter than the posterior one. The VCS 608 tibiae depart from this morphology and more closely resemble Bison, where the two facets are well separated 609 by a marked notch and the anterior facet is concave to some extent. The size of the distal epiphysis in the VCS 610 sample fits within the variation of Bison spp. (Table S23). 611 612 4.2.11. Astragalus 613 DescriptionSee Fig.7a–b and Table S24. The bone is large and stout, and the mediolateral diameter 614 is always larger than half of the maximum proximodistal length. The proximal lateral trochlea is wider and 615 higher than the medial one. The intertrochlear notch is narrow and smooth, and ends distally in a deep and 616 rough fossa. The distal trochleae are more similar to each other than the proximal ones, although the lateral 617 one is slightly wider. The distal intertrochlear notch is wider and shallower compared to the proximal one. 23 618 There is a small tubercle between the proximal and distal trochleae along the medial margin of the bone. Most 619 of the posterior surface of the talus is occupied by the wide and subrectangular calcaneal articular facet. This 620 facet is delimited dorsally by a transversal groove. The posterior margin of the calcaneal facet is separated 621 from the distal trochlea and the central tarsal facet by a deep and narrow L-shaped groove. The angle between 622 the two segments of this groove is slightly higher than 90°. The cubonavicular facet is located at the 623 distomedial corner of the calcaneal facet and is delimited laterally by a step, which is less marked in IPS92952 624 and IPS92951 than in the remaining specimens. 625 Remarks and comparisonsThe VCS astragali are homogenous in size and shape, with the only 626 exception of IPS92951, which is slightly smaller and slenderer than the others. The astragalus exhibits several 627 diagnostic features that distinguish Bison from Bos (Schertz, 1936a; Bibikova, 1958; Stampfli, 1963; Brugal, 628 1985; Sala, 1986; Gee, 1993; Sher, 1997). On the posterior side, the groove that separates the calcaneal facet 629 from the central tarsal facet is less accentuated in Bos than in Bison (Schertz, 1936a) and the L-shaped angle 630 described by this groove is almost right in Bos and obtuse in Bison (Brugal, 1985; Sala, 1986). These characters, 631 even if somewhat variable in both Bos and Bison, can help discriminating them based on large samples. All the 632 VCS specimens display a deep and obtuse groove, thus more closely resembling Bison. Furthermore, according 633 to Bibikova (1958), Bos would also differ from Bison in having a longer and thinner prolongation of the 634 calcaneal facet on the lateral side of the astragalus. However, this feature is highly variable in the VCS sample, 635 with some specimens (e.g., IP92953) displaying the supposedly Bos-like condition and others (e.g., IPS92949) 636 being instead Bison-like. Gee (1993) evidenced that in Bos the posterior lateral margin of the astraglus is 637 curved, whereas it is straight in Bison. We found that this character is generally reliable, as further shown by 638 the VCS astragalus sample, which shows very straight margins. 639 Leptobos and B. (Eobison) somewhat differ from Bison s.s. in the ratio between the width of the distal 640 trochlea and the total length of the astragalus (Fig. 9), with the former taxa having smaller and thinner bones. 641 However, the ranges given by the DEW/Lmax of all the species referred to Bison s.l. overlap widely, making 642 this feature not particularly diagnostic, especially for isolated remains. Within Bison s.s., B. priscus has larger 643 (both longer and wider) astragali compared to B. menneri and B. schoetensacki, but there are not many 644 evidence of differences in stoutness (Fig. 9d; Table S25). 24 645 646 4.2.12. Calcaneum 647 DescriptionSee Fig. 8c–d and Table S26. The calcaneum body displays an hourglass-shaped anterior 648 contour, with a constriction just above the substentaculum tali and a massive tuber calcanei (particularly 649 developed in IPS13936, IP92946, and IPS92948). The anterior and posterior edges of the tuber are slightly 650 convergent toward the proximal end which, in medial view, has subrounded margin. The angle between the 651 calcaneum body and the sustentaculum tali is almost right. The cubonavicular facet is elongate and shows a 652 sinuous concavity (oriented postero-medially in the proximal half and posteriorly in the distal). The anterior 653 and posterior margins of the calcaneum are convergent towards the tuber calcanei. The medial surface of the 654 anterior process bears a large tuberosity, edged posteriorly by a shallow depression. Posterior to the 655 sustentaculum tali, there is a very small crest (not visible, probably due by taphonomic damage, in IPS92945). 656 In distal view, the astragalus facet has a subsquare shape. This articulation expands also on the antero-medial 657 side of the sustentaculum tali and has a semicircular outline. This articulation forms a right angle with the 658 curved cubonavicular facet. Thisfacet is delimited on the medial margin of the postero-medial sustentaculum 659 tali by a subtriangular rough area. The facet for the malleolus is concave, in anterior view is proximodistally 660 elongated and has a subrectangular shape. The anterior process is pointed and projects distally. 661 Remarks and comparisonsThe VCS calcanei come from EVT7, apart from IPS13936, which comes 662 from CGR-D2. The EVT specimens are morphologically quite homogeneous, except minor differences in the 663 development of the tuber calcanei, probably related to sexual dimorphism or age. In contrast, IPS13936 is 664 much larger. Previous studies tried to find diagnostic features in the calcaneum, especially based on the 665 relative size of the central tarsal and astragalar articular facets (Bibikova, 1958; Stampfli, 1963; Sala, 1986). 666 However, the size of these facets is extremely variable even within a single bovine population (Sher, 1997) and 667 thus lacks any taxonomic value. The average dimensions of the VCS calcanei resemble those of Bison spp. 668 (Table S27). Like other postcranial elements, the calcaneum of Bison s.s. is generally larger than in the most 669 primitive species. However, it has to be noted that the calcaneum of female specimens of B. priscus and B. 670 schoetensacki is much smaller than that of the males, being similar to that of small-sized species of Leptobos 671 and B. (Eobison). 25 672 673 4.2.13. Other tarsal bones 674 See Fig. S3 and Table S15. As for the carpal bones, very few tarsal bones were recovered from the VCS 675 (one cuneiform and two cubonaviculars). The lack of comparative samples and the scarcity of the material do 676 not allow us to perform proper comparison with other bovid populations. 677 CuneiformThe only cuneiform (IPS92965) recovered from VCS comes from EVT10. It is complete, 678 although slightly damaged, and resembles in size and shape those of Bison and Bos (Sala, 1986: Figs 32, 33). 679 Central tarsal boneThe two specimens from EVT10 are quite similar to each other. IPS92957 is 680 slightly fragmented in some portions of the proximal and distal surfaces, whereas IPS92958 is complete. The 681 articular facets for the astragalus are ovoid, with the medial facet being slightly larger than the lateral. Overall, 682 their morphology is typical of Bison and Bos. The comparison with the material of B. schoetensacki from Isernia 683 shows that the VCS specimens are quite similar to those from the Italian site. 684 685 4.2.14. Metatarsals 686 DescriptionThe metatarsals are less abundant and morphologically more homogeneous than the 687 metacarpals in the VCS collection (Fig. 6d–f; Table 6). Half of the specimens are lacking the distal epiphysis and 688 most of them are damaged, but four specimens (IPS92934, IPS92931, IPS107634, and IPS92932) are very well 689 preserved. 690 The medial articular facet is posteriorly concave and shows an irregular pit joined to the posterior 691 margin. The lateral facet is higher and relatively flat. They are separated by a well-developed ridge. Three 692 specimens (IPS92932, IP92933, and IPS92934) show a small bulge in the middle of the medial margin, 693 posteriorly to the medial facet (particularly developed in IPS92932). The posterior articular facet is small, 694 antero-posteriorly short, strongly inclined toward lateral margin, and of variable maximum length. A secondary 695 facet for the articulation with the second metatarsal is present at the postero-medial corner of the proximal 696 epiphysis. There is a small crest along the postero-lateral corner of the articular surface. Two specimens 697 (IPS92932 and especially IPS92932) have a markedly slenderer diaphysis than the rest of the sample. A very 698 small and narrow proximal foramen is still visible only in IPS92931, IPS92936, IPS92937, IPS92938, and 26 699 IPS107634. The anterior side of the bone is characterized by the well-marked and deep vascular groove. In 700 their distal portion, the edges of the groove are shaped as sharp ridges. There is a secondary, shallower groove 701 located on the dorso-lateral aspect of the proximal portion of the shaft in most of the specimens (although it 702 is very faint in IPS92932 and IPS92932). An elliptical distal foramen is present in all the specimens. In the distal 703 epiphysis, the margins of the intertrochlear space are subparallel to distally convergent. As in the metacarpals, 704 the epitrochlear tubercles are well developed, so that the width at the level of the tubercles is about the same 705 as that measured between the outer trochlear margins. The only notable feature in the posterior side of the 706 proximal epiphysis is the pointed medial corner. Distally next to it, IPS92930 and IPS92932 show a small circular 707 facet, which is also present (although less marked) in IPS107634, corresponding to the posterior articular facet 708 for the second metatarsal. The large proximal foramen is located inside a deep subcircular depression distal 709 to the posterior margin of the proximal epiphysis. The proximal diaphysis is delimited by two crests, the medial 710 being usually stronger than the lateral, which do not extend beyond the proximal third of the shaft. The 711 posterior surface of the diaphysis is slightly concave to flat in its proximal half, where two very shallow 712 longitudinal grooves are present. These two grooves disappear toward the distal portion, which is completely 713 flat. The distal foramen is distinct and elliptical. As in the metacarpals, on the anterior side, proximally to each 714 trochlear crest, two depressions develop, with the outer one being larger and deeper than the inner. The distal 715 epiphysis is slightly curved posteriorly. The lateral and medial trochlear pits are deep and proximally 716 surrounded by radial rugosities. The two trochlear ridges are subparallel relative to the medial and lateral 717 margins of the distal end and converge anteriorly. 718 Remarks and comparisonsAs for the metacarpals, the metatarsals have often been considered 719 diagnostic for large bovids in the literature (Table S28). In the proximal articular surface, the angle between 720 the lateral and medial facets is lower in Bos (13–22°) than in Bison (22–40°; Schertz 1936a; Sala, 1986; Sher, 721 1997), even if the measurement can be somewhat biased depending on the methods (Sher, 1997). The VCS 722 sample most closely resembles Bison in this regard (20–30°). The anterior contact between the two proximal 723 facets is different in Bos and Bison (Brugal, 1985; Gee, 1993). In Bos there is no contact between these two 724 facets, often divided by a narrow channel. In Bison, the facets are confluent forming, in most of the cases, a 725 small but sharp ridge (Gee, 1993). In the VCS metatarsals, the two facets are in contact and a clear ridge is 27 726 present in all the specimens. Nevertheless, the major difference in metatarsal morphology between Bos and 727 Bison lies at the distal half of the bone. As for the metacarpals, the VCS metatarsals display the typical 728 morphology of Bison s.l., where the contact between the diaphysis and the distal epiphysis is medio-laterally 729 inflated (Sala, 1986; Sher, 1997)—so that medio-lateral width at this level is similar or even larger than than 730 that between the trochlear margins—whereas in Bos the medial and lateral margins of the distal end are 731 strongly divergent (Table S29). 732 Bison can readily be distinguished from Leptobos given the possession of larger and stouter 733 metatarsals, but given the considerable variation within Bison s.l. it is difficult to find reliable differences 734 among different species (Table S28). The “stoutness” diagram (Fig. 9b) shows that B. priscus and B. 735 schoetensacki have generally heavily-built metatarsals with wider ends compared to other samples, despite 736 considerable variation; in contrast, B. menneri has much slenderer metatarsals, while those of B. (Eobison) 737 spp. are both smaller and slenderer. Most of the VCS sample displays an intermediate size and slender built, 738 except for two specimens (IPS107634 and IPS92937 from EVT12) that are particularly large and massive, and 739 a quite slender specimen (IPS114553 from EVT7). The ANOVA shows that, although there are some differences 740 between syn-Jaramillo (EVT10 and EVT12) and post-Jaramillo (CGRD7 and EVT7) subsamples, these are not 741 significant for seven out of the eight analyzed variables (Table 4). The “stoutness” diagram and PCAs further 742 confirm that, despite the rather homogeneous morphology of the whole sample, a change from stouter to 743 slenderer metatarsals can be recognized from EVT10-EVT12 to CGRD7-EVT7. The MANOVA on five variables 744 confirms that the relative proportions of the VCS metatarsal sample are not characterized by significant 745 differences (Table 4). 746 As for the metacarpals, the PCAs enable the distinction of various groups of Bison s.l. despite 747 considerable overlap. In the first PCA (Fig. 10c, Table S30), based on seven shape variables, PC1 (40% 748 variance), positive scores are mainly influenced by msLmax and, to a lesser extent, msDET, whereas negative 749 scores are mostly driven by msDW, msPEW, and msDEW. In turn, PC2 (21% variance) is mainly determined 750 by msDEW and msDET (positive scores) and msDW, and to a lesser extent Lmax (negative scores). The VCS 751 sample is rather scattered in the diagram. The most massive specimens (IPS107634 and IPS92937), 752 characterized by a relatively stout appearance (short and wide bones), overlap with the bulk B. priscus form 28 753 UK and Taubach (negative PC1 scores). The VCS slender specimens (possibly females) overlap with the more 754 gracile B. (Eobison) group and B. menneri, which feature positive PC1 values due to their relatively long and 755 narrow diaphysis. 756 In the second PCA (Fig. 10d, Table S31), PC1 (39% variance) is almost totally influenced positively by 757 reMLmax, separating long metatarsals (positive scores) from short ones (negative scores). PC2 (25% variance), 758 on the contrary, separates metatarsals with more massive diaphysis (high positive values of reDT and reDW) 759 from those with a stouter distal epiphysis. The VCS metacarpals partially overlap with B. menneri from 760 Untermassfeld, B. (Eobison) cf. degiulii from the Mygdonia Basin, and B. priscus from UK, highlighting the high 761 variation of these bones in the VCS sample. The meager comparative data for B. schoetensacki do not allow us 762 to further assess the variation of this species. The MANOVA indicates that our sample differs significantly only 763 from the extremely slender samples from Venta Micena and Untermassfeld (Table. S32). This result obtained 764 for the metatarsals highlight less differences between taxa than the same analyses carried out on metacarpals 765 (see Section 4.2.9), pointing out that the bison hindlimbs are more variable. 766 767 4.2.15. Phalanges 768 Proximal phalangesThe proximal phalanges are overall well preserved and morphologically very 769 similar (Fig. S4; Table S33). The abaxial and interdigital margins are parallel so that the width of the phalanx is 770 similar throughout its length. The abaxial tuberosity is more developed than the interdigital one; in IPS14977 771 and IPS92925, the former constitutes a large bulging prominence. Between these tuberosities and the two 772 small sesamoid facets, there are two tubercles separated by a depression of variable depth. A depression, 773 developed mediolaterally, is present in the proximal part of the phalanx, between the posterior tuberosities 774 and the proximal end. In the proximal articulation, the interdigital glenoid cavity is higher and narrower than 775 the abaxial one. The abaxial sesamoid facet is larger and flatter than the interdigital one. 776 Intermediate phalanges See Fig. S4 and Table S33. The shaft of the phalanx is stout and massive. 777 The interdigital surface is concave, while the abaxial is convex. In the proximal articulation, the abaxial glenoid 778 cavity is slightly larger than the interdigital, and both are inclined dorsally and interdigitally. Below the articular 29 779 surface there is a small depression. The outline of the distal articular trochlea is triangular, with the abaxial 780 lobe larger than the interdigital; the groove that divides the two lobes is inclined dorsally and interdigitally. 781 Distal phalanges See Figure S4 and Table S33. In dorsal view, the abaxial margin is convex and the 782 interdigital is almost straight. Proximally, the two glenoid cavities are similar in size and inclined dorsally and 783 interdigitally, aligned obliquely relative to the sagittal plane. Posteriorly to them, a small subtriangular, 784 interdigitally-oriented sesamoid facet is visible. Below the sesamoid facet, there is a deep oblique groove, 785 delimited by a small crest. The abaxial surface is convex while the interdigital one is slightly concave. 786 Remarks and comparison (Phalanges)The phalanxes of Leptobos, Bison and Bos are quite similar and 787 no morphological differences were found. Discriminating between forelimb and hindlimb phalanxes, also 788 taking in account the sexual dimorphism in bovids, is not easy. Generally, the forelimb proximal phalanges are 789 shorter and more compact than the hindlimb ones (Revilliod et Dottrens, 1946), which are longer and relatively 790 slenderer (Sala, 1986); shorter and wider according to Sher (1997). The intermediate phalanxes are slightly 791 longer but considerably wider in the forelimbs than in the hindlimbs (Sala, 1986; Sher, 1997). According to Sala 792 (1986) the forelimb distal phalanxes are shorter and broader compared to the hindlimb ones, whereas Sher 793 (1997) states that, on the contrary, the forelimb distal phalanges are longer than the hindlimb onesThe fact 794 that in most cases the phalanges are found isolatedmake this discrimination difficult, which is further 795 aggravated by the fact that phalangeal measurements are seldom published and not all the authors agree in 796 how to distinguish the manual and pedal phalanges (e.g. Sala, 1986 and Sher, 1997). Regarding the sample 797 from VCS, the lack of comparative measurements, the scarcity of the material, and the lack of associated 798 specimens make it impossible to reliably discriminate between fore and hind phalanges or between male and 799 female specimens. 800 801 5. Results 802 5.1. The VCS sample: taxonomy and morphological variation 803 Based on the qualitative and quantitative comparisons reported above, it is possible to confidently 804 refer all described remains from the VCS to to the same species (Table 4). The analyses performed on the 805 metapodials suggest that two morphotypes are recognizable, characterized by slightly different proportions 30 806 probably due to ecophenotypic variation (see Section 6.1.). The study performed on the 220 remains allow us 807 to refer the large bovid from VCS to the genus Bison. Moreover, the large size and the stout morphology of 808 the limb bones support their assignment to subgenus Bison, which includes the largest species of the genus. 809 Some features, such as the relatively slender limbs, most closely resemble the early members of this subgenus, 810 namely B. menneri and B. schoetensacki, which are first recorded during the Epivillafranchian. The VCS bison 811 differs from the roughly coeval B. menneri in the markedly stouter limb bones, as well as the shorter and wider 812 metapodials (in particular, the metacarpals). The steppe bison B. priscus is characterized by even more robust 813 proportions (Fig. 9). The partial overlap in metapodial dimensions between the larger and stouter specimens 814 from the VCS and the smaller and slenderer specimens of B. priscus probably reflects sexual size dimorphism 815 in the two samples, with male individuals from VCS displaying a similar size to female individuals of B. priscus. 816 In sum, the morphological characters of the VCS bison allow the referral of this sample to Bison 817 schoetensacki, which was the most common bison species in Europe during the Middle Pleistocene, although 818 it originated during the latest Early Pleistocene (Flerov, 1975, 1979). The chronostratigraphic distribution of B. 819 schoetensacki might span from ca. 1.1 to 0.5 Ma (Grange et al. 2018). Bison schoetensacki has been regarded 820 as an early member of the lineage that ultimately led to the steppe bison and extant European wisent (Palacio 821 et al., 2017), but this needs to be clarified further. Although B. schoetensacki is quite common in the European 822 fossil record, only a few localities have yielded a large number of fossils, hindering a more detailed assessment 823 of its phylogenetic relationships. 824 The VCS sample is the largest collection of B. schoetensacki, together with that from Isernia, thus 825 substantially improving our knowledge on the anatomy of this species and allowing us to provide the 826 reconstruction depicted in Fig. 11. Moreover, almost all the fossils from VCS were recovered from layers dated 827 between 1.1 and 0.86 Ma, thus offering a broader chronological perspective on the morphology of this species, 828 which was likely influenced by major paleoenvironmental changes in the study area (see Section 6.1). 829 830 5.2. The Leptobos etruscus–Leptobos vallisarni lineage 831 Bison probably evolved in Asia from a derived species of Leptobos (Pilgrim, 1947; Tong et al., 2016). 832 Among the described species of the latter genus, L. etruscus from the Late Villafranchian of Europe is the 31 833 largest and one of the most derived, showing some cranial features that closely resemble those of Bison s.l. 834 (Masini, 1989; Bukhsianidze, 2005; Masini et al., 2013). The postcranial skeleton of L. etruscus is well known 835 thanks to the large collections from Senèze (France), Olivola and Upper Valdarno (Italy). Its metapodials are 836 particularly slender (length reaching 26.5 mm and epiphysis width not exceeding 70 mm; Table S16), being 837 similar in size to those of B. (Eobison) (Fig. 10). Other postcranial bones (e.g., radii, astragali) are overall smaller 838 and slenderer than those of Bison s.l. (Fig. 10). 839 Leptobos vallisarni, which is up to date the only Leptobos species reported both in Europe and Asia, is 840 characterized by even more derived features (Masini, 1989). This large-sized species was described by Merla 841 (1949) on the basis of a partial cranium from the Early Pleistocene of the Upper Valdarno, but two almost 842 complete skulls from the Gonghe Basin (central China) testify to the wide geographic distribution of this taxon 843 (Zheng et al., 1985). Unfortunately, the postcranial skeleton of L. vallisarni is poorly known. The very rich 844 sample from the late Villafranchian of Pietrafitta (central Italy), which is attributed to L. aff. vallisarni (Masini, 845 1989; Gentili and Masini, 2005) and also includes a large number of metapodials, displays a derived cranial 846 morphology that resembles that of Bison s.l.—thereby rendering the attribution to Leptobos unreliable. A few 847 metapodials from the Upper Valdarno, housed in the Natural History Museum of the University of Florence, 848 have been attributed to this species (Masini, 1989; Masini et al., 2013), but our analysis of the collection 849 suggests that they might had been mixed with material of other Leptobos species. For this reason, we refrain 850 from making further assumptions about the limb proportions of L. vallisarni until a complete revision of the 851 European Leptobos spp. postcranial collections is undertaken. 852 853 5.3. Early Bison species 854 The early occurrences of Bison s.l. correspond to B. (E.) palaeosinensis and B. (E.) sivalensis from Asia. 855 While many authors agree on an Asian origin of Bison s.l. (e.g., Flerov, 1972; Sala, 1986; Sher, 1997), there is 856 still no consensus on the chronology. The earliest known remains are referred to B. (E.) cf. sivalensis from the 857 Upper Siwaliks (northern Pakistan), dated to 3.3–2.6 Ma (Khan et al., 2010). These fossils mostly consist of 858 cranial material, and therefore we did not include them in our comparative analyses. However, the remains 859 attributed to B. (E.) sivalensis, apart from the lost holotype cranium, are too fragmentary to confirm that they 32 860 belong to a single species, particularly in the light of their unclear stratigraphic provenance (Kostopoulos et 861 al., 2018). 862 In turn, B. (E.) palaeosinensis is a small-sized and primitive species from the Early-Middle Villafranchian 863 of Asia, whose taxonomic status has been much debated (Teilhard de Chardin and Piveteau, 1930; Skinner and 864 Kaiser, 1947; Tong et al., 2016). The three incomplete crania and several postcranial bones of this species 865 come from different sites of Yushe and Nihowan Basins (China). Our analyses reveal that the metacarpals 866 where short and relatively slender, falling within the B. (Eobison) range (Fig. 9). One metacarpal from Nihowan 867 (NIH113; Masini, 1989) displays a slender morphology that fits with L. etruscus (Fig. 9), whereas the remaining 868 metacarpals are and being most similar to those of Bison sp. from Venta Micena and B. (Eobison) degiulii from 869 the Italian Peninsula (Fig. 9). The metatarsals, radii and astragali are small and slender, largely overlapping with 870 Leptobos but not with Bison s.s. 871 The earliest record of Bison s.l. from Europe corresponds to the small and primitive species B. (Eobison) 872 georgicus. The remains of this species come from the Late Villafranchian site of Dmanisi (Georgia; ca. 1.77 Ma) 873 and consist of a single neurocranium with horn cores and several postcranial bones (Bukhsianidze, 2005). Our 874 analyses shown that two metacarpals (GNM D2288, GNM D2812) display slender proportions similar to those 875 of L. etruscus, whereas another (GNM D3426) is clearly stouter and matches instead the variation of B. 876 (Eobison) spp. and also overlaps to some extent with B. schoetensacki (Fig. 10). The slenderer specimens are 877 about 8% longer than but more than 12% narrower distally than the stouter one. Given that, often, female 878 Bison metacarpals are shorter and narrower than those of males (Schertz, 1936a), the aforementioned 879 differences might simply be explained by sexual dimorphism. On the other hand, two complete radii from 880 Dmanisi (GNM D2962, GNM D2165) do not display the elongated and slender morphology of L. etruscus but 881 closely resemble that of Bison s.l. Taking into account the age of the site, Dmanisi might record the co- 882 occurrence of the last Leptobos and earliest of Bison s.l. with transitional characters, as already suggested by 883 Kostopoulos et al. (2018), but additional fossils would be required to adequately test such a possibility. 884 A large bovid from Venta Micena (southern Spain; ca. 1.6 Ma) was attributed to Bison sp. due to its 885 clearly “bisontine” size and proportions (Moyà Solà, 1987). The metapodials are shorter than those of L. 886 etruscus but slenderer than those of late forms of Bison s.s., resembling the samples of the B. (E.) degiulii and 33 887 B. (E.) palaeosinensis (Fig. 9). In metacarpal proportions (Fig. 9a), two specimens (VM-9033 and VM-925) at 888 the upper range of the Venta Micena sample overlap with those of L. etruscus from Olivola and Senèze. The 889 rest of the Spanish sample is otherwise quite homogeneous, albeit clearly showing differences attributable to 890 sexual dimorphism in the size and robusticity of the metacarpals. The metatarsals similarly plot with the 891 smaller and slenderer forms (i.e., Leptobos and Eobison; Fig. 9b). The PCAs, irrespective of the variables used, 892 indicate that the metapodials of the Venta Micena bovid are intermediate between Leptobos and Bison s.s., 893 partially overlapping with the former (Fig. 10). The three most complete radii from Venta Micena are quite 894 short and slender, being the smallest ones in our comparative sample (Fig. 9c). The identification of the bovid 895 postcranials from Venta Micena is further complicated by the attribution of cranial remains to both Bison sp. 896 and Hemibos aff. gracilis, of Asian origin (see Martínez-Navarro et al., 2011). A revision of the cranial and 897 postcranial bovid material from Venta Micena is pending, but our morphometric analyses indicate that the 898 metapodials fit well with the morphology of B. (Eobison) (Figs. 9–10). On geographic and chronological 899 grounds, the Venta Micena sample might be referable to B. (E.) degiulii, but more in-depth analyses would be 900 required to confirm such an attribution. 901 Current knowledge of B. (E.) degiulii is limited. The type material includes the partial cranium of an 902 elder individual (holotype) and six metapodials (five metacarpals and one metatarsal) from the latest 903 Villafranchian of Pirro Nord (southern Italy; ca. 1.6–1.4 Ma; Masini, 1989). Masini (1989) also attributed to 904 this species one metacarpal and one metatarsal from Capena (central Italy, Late Villafranchian) and three 905 metacarpals from Sainzelles (southern France, Late Villafranchian, Brugal, 1995). As noted above, the 906 metacarpals of B. (E.) degiulii are quite similar to those from Venta Micena, albeit they are slightly shorter 907 and stouter (Fig. 9). The bovid sample from the Mygdonia Basin (Greece), mainly including metapodial 908 remains and a few cranial elements from Kalamoto, Tsiotra Vryssi, Krimni, and Apollonia (dated to between 909 1.7 and 1.2 Ma), was attributed to B. (E.) cf. degiulii by Kostopoulos et al. (2018). The marked morphological 910 variation among metapodials was interpreted as the result of an increasing size and stoutness trend in 911 relation to progressive climate deterioration throughout the Late Villafranchian (Kostopoulos et al., 2018). 912 Several metacarpals from the younger locality of Apollonia (ca. 1.2–1.1 Ma) resemble male specimens of B. 913 schoetensacki in size and robucity, but most of the sample displays the typical proportions of B. (Eobison), 34 914 and only two (AUTH APL-677, AUTH KRM) are slightly slenderer and more Leptobos-like (Fig. 9). The PCAs 915 performed Maniakas and Kostopoulos (2017a) and Kostopoulos et al. (2018) showed considerable overlap 916 among Bison s.l. species, only distinguishing “slender” from “stout” forms. They revealed proportion 917 similarities between the Mygdonia bovid remains and those of B. menneri from Untermassfeld, as well as 918 between the latter and Bison sp. from Venta Micena. These similarities are also evident from our “stoutness” 919 and Log10 ratio diagrams (Fig. 9), which show that B. menneri and B. (Eobison) spp. mostly differ in size but 920 not in proportions. Kostopoulos et al. (2018) questioned the inclusion of B. degiulii into subgenus Eobison 921 based on some derived characters of the holotype and the cranium KLT-638 from Kalamoto. If the whole 922 sample from the Mygdonia basin belongs to a single species (Kostopoulos et al., 2018), the postcranial 923 remains further display a mosaic of derived and primitive features, because some metacarpals from 924 Apollonia (AUTH APL-745, AUTH APL-414, AUTH APL-578, AUTH APL-446, and AUTH APL-95) resemble in 925 robusticity the male specimens of B. schoetensacki from Mosbach, Durfort, Cromer Forest-bed, and VCS (Fig. 926 9). Nevertheless, our PCAs, ANOVAs and Log10 ratio diagrams confirm that, despite some overlap, the 927 Mygdonia sample is distinct from “priscoid” forms and show that, on average, the Greek metacarpals are 928 characterized by relatively shorter and slenderer proportions (Fig. 9e–g, Fig. 10a–b, Table S21). Such results 929 suggest suggeststhat the assignment to B. (E.) cf. degiulii by Kostopoulos et al. (2018) is well supported, at 930 least until a more detailed revision of this species is undertaken. 931 Two additional species of B. (Eobison) are poorly known. Bison (E.) tamanensis from the Taman 932 Peninsula and B. (Eobison) suchovi from central Ukraine, erected without a diagnosis by Verestchagin (1959) 933 and Alekseeva (1967), respectively. Both these late Early Pleistocene Eastern European species have a debated 934 taxonomic history and are described on quite scanty fossil material (see Kostopoulos et al., 2018 and 935 references therein). In the light of these issues the taxonomic status of this species cannot be properly 936 assessed and is not discussed in this paper. 937 Unlike the above, B. menneri is a well-known species from the German site of Untermassfeld (ca. 1.0 938 Ma; MIS31), being considered one of the earliest members of Bison s.s. (Sher, 1997). Bukhsianidze (2020) 939 recently referred this species to the subgenus Bison (Poephagus) based on purported closer cranial similarities 940 with extant yaks. The postcranial morphology of B. menneri has adequately been characterized (the holotype 35 941 itself, IQW 1982/17948, is a male metacarpal; Sher, 1997). The species is described as a long-legged bovid with 942 tall appearance, as well as a relatively small head and short horns (Sher, 1997; van Asperen and Kahlke, 2017; 943 Bukhsianidze, 2020). Among large bovids, B. menneri has the longest and most slender metacarpals, and 944 displays a mixture of Bison-like and, to a lesser extent, Bos-like features in the limb bones (Sher, 1997; 945 Bukhsianidze, 2020). The extremely elongated metapodials are similar in proportions to those of Leptobos and 946 B. (Eobison) (Fig. 9a–b), with considerable overlap with the latter in our PCAs (Fig. 10). Neverthelelss, Sher 947 (1997) ruled out the possibility that the Untermassfeld bovid could represent a boreal variant of a 948 Mediterranean/Asian Eobison species, due to its large size and metapodial built. The most complete skull 949 displays some primitive characters for the bison lineage, such as not very tubular orbits, elongated postcornual 950 portion of the cranium, and horn cores very backwardly orientated. According to Bukhsianidze (2020) these 951 elements are shared with the yak lineage (subgenus Poephagus). From the site of Untermassfeld, a juvenile 952 skull and a single metacarpal (IQW 1983/19 253 (Mei. 18 773)) are not referable to the aforementioned 953 species (Bukhsianidze, 2020); indeed, IQW 1983/19 253 (Mei. 18 773) resembles in stoutness the more 954 “priscoid” form B. schoetensacki (Fig. 9), although the scanty remains do not enable a specific attribution. 955 Bison menneri has also been reported from the North Sea seabed “Het Gat” site (Mol et al., 2003) and, 956 tentatively (B. cf. menneri), from layer TE9c of Sima del Elefante (Spain) (Huguet et al., 2017) and from 957 Cimichioi-III and Hadjimus (Moldova) (Croitor, 2016). 958 The three species from east Europe and Asia: Adjiderebos cantabilis, Protobison kushkunensis and 959 Probison dehmi (Dubrovo and Burchak-Abramovich, 1986; Burchak-Abramovich, Gadzhiev and Vekua, 1980; 960 Shani and Khan, 1968) are known for isolated cranial remains. Their affinities and relationships with Bison s.l. 961 group are still matter of debate, however their transitional morphology from Leptobos to Bison could shed 962 lights on the first forms of primitive bison. 963 964 5.4. Other samples of Bison schoetensacki 965 5.4.1. Le Vallonnet 966 The bison remains from Le Vallonnet (southeastern France, ca. 1.2–1.1 Ma), which is one of the few 967 European localities that record the earliest Epivillafranchian (de Lumley et al., 1988; Moullé, 1992; Moullé et 36 968 al., 2006; Michel et al., 2017), were attributed to B. schoetensacki by Moullé (1992). The metapodials from Le 969 Vallonnet are characterized by large size and quite slender proportions, resembling material of B. 970 schoetensacki from the type locality of Mauer, and partially overlapping with the specimens from Mosbach 971 and Durfort (Fig.9). Sexual dimorphism in particularly pronounced in the Le Vallonnet metacarpal sample, with 972 the putative female specimens (MPRM A8 B2 326 and MPRM B9 BJ7 346) being located on the left portion of 973 the stoutness diagram due to their slenderer morphology (Fig. 9a). Both the Le Vallonnet and the Mauer 974 samples partially overlap with that of B. menneri from Untermassfeld, which they resemble in the elongated 975 diaphysis and relatively narrow distal epiphysis, despite the larger diaphysis (Fig. 9a). Other postcranial bones 976 from Le Vallonnet, such as astragali and humeri, are within the size range of Bison s.s., even though the 977 astragali are slightly stouter than in other samples of B. schoetensacki (Fig. 9d). The two complete metatarsals 978 (MPRM B6278(G) and MPRM E7C186(D)) from Le Vallonnet most likely belong to large males, due to their 979 heavily-built morphology. Indeed, B6278(G) is one of the largest specimens in the entire sample of B. 980 schoetensacki and falls within the range of “priscoid” forms (Fig. 9b). Based on the Log10 ratio diagrams, PCAs, 981 and pairwise comparison, the sample of metacarpals from Le Vallonnet is similar to those of B. schoetensacki, 982 except for some diaphyseal measurements that appear relatively wider and thicker (Figs 9–10). Overall, the Le 983 Vallonnet sample fits well with the variation of B. schoetensacki, and in particular with the large morphotype 984 represented by the Mauer sample thereby confirming the presence of this species before the Jaramillo 985 subchron. Together with the specimens described here from roughly coeval VCS layers (CGRD2), they 986 represent the first occurrences of B. schoetensacki, conclusively indicating that Bison s.s. was recorded in 987 Europe since the Villafranchian-Epivillafranchian boundary. 988 989 5.4.2. Durfort and La Vassyére 990 The bison sample from Durfort (southwestern France; ca. 1.0–0.5 Ma) was referred to B. schoetensacki 991 on the basis of cranial (horn cores) and metapodial features (Brugal, 1995). The metapodials are 992 morphologically intermediate between those of B. menneri and B. priscus, and resemble the earlier fossils of 993 B. schoetensacki from Le Vallonnet at a slightly smaller size (particularly the metacarpals; Fig. 9). The 994 “stoutness”and Log10 diagrams and both PCAs (Figs 9–10) indicate that the Durfort sample fits with the 37 995 variation of B. schoetensacki, resembling the stout forms from Mosbach and the VCS—as the male specimens 996 (MNHN 010 D, MNHN 107 G, MNHN 105 G, MNHN 104 G) approach the “priscoid” scatter, and the putative 997 female specimens (MNHN 106 G, MNHN 108 G) overlap with B. (Eobison) (Figs 9a, 10a–b). The analysis of 998 metatarsals (Figs 9b, 10c–d) shows the same pattern, with the material from Durfort being intermediate 999 between the “priscoid” and slenderer forms. Our results confirm the attribution of the Durfort sample to B. 1000 schoetensacki, being characterized by slightly stouter metapodial proportions than the material from Le 1001 Vallonnet and Mauer, and most similar to the VCS bovid (Figs. 9–10). From the Early Pleistocene site of La 1002 Vassyére (southern France; ca. 0.6) several remains of a large bovid were recovered. The fossils were 1003 attributed to Bison cf. schoetensacki (Brugal and Fosse, 2003). The general morphology and proportions of the 1004 only complete metacarpal fit perfectly with the variation of B. schoetensacki (Fig. 9). 1005 1006 5.4.3. Mauer and Mosbach 1007 Bison schoetensacki was originally described by Freudenberg (1914) on the basis of remains from 1008 Mauer (Germany; Middle Pleistocene, ca. 0.4 Ma). The holotype is a partial cranial vault (which was lost during 1009 World War II; Sala, 1986), but Freudenberg (1914) also referred to the same species other cranial and few 1010 postcranial remains from Mauer and Cromer Forest-bed, and to B. cf. schoetensacki an almost complete skull 1011 from Mosbach. Multiple studies on this species have not entirely clarified its diagnostic features (e.g., 1012 Hilzheimer, 1918; Schertz, 1936a, b; Skinner and Kaisen, 1947; Flerov, 1969; Sala, 1986; Sher, 1997; Drees, 1013 2005). Sala (1986) described the cranial anatomy of B. schoetensacki based on the remains from Mauer and 1014 Isernia (Italy; Middle Pleistocene, ca. 0.55 Ma), but for the postcranium only summarized the most important 1015 differences between Bos and Bison. In contrast, Sher (1997) focused on metapodial proportions and rejected 1016 the common misconception that B. schoetensacki was a small-sized Bison (Sher, 1997), improving the original 1017 description given by Freudenberg (1914), in which only the relative slenderness of the limbs and the small 1018 horn cores were stressed. 1019 During the last century, the material from Mauer and Mosbach was scattered across more than five 1020 different institutions. We managed to measure some of the specimens, while for others we relied on the few 1021 published measurements (Freudenberg, 1914; Schertz, 1936a, b; Sher, 1997). However, it should be taken into 38 1022 account that the collections from Mauer/Mosbach might mix two different species, B. schoetensacki and B. 1023 priscus (Schertz, 1936a, b; Sher, 1997). Their stratigraphic context is not clear, especially for Mosbach in which 1024 two different levels with different ages and faunas are recognized (Breda and Marchetti, 2005), and both sites 1025 include layers dated to 0.6–0.5 Ma (Wagner et al., 2010; Kahlke et al., 2011), close to MIS11–9 (0.4–0.3 Ma), 1026 when the first occurrence of the large and stout B. priscus is recorded (Kahlke, 1999). According to Schertz 1027 (1936b) and Sher (1997), most of the material from Mosbach and some specimens from Mauer would indeed 1028 belong to B. priscus. This idea derives from the assumption that B. priscus has long metacarpals, whereas, in 1029 fact, the various well-known populations of B. priscus from Eurasia (except the gigantic ones from Taubach, 1030 Romain la Roche and, possibly, Tiraspol) are characterized by very short and extremely stout metacarpals (e.g., 1031 North Sea, Krasny Yar, Roter Berg, Chumysh, Kiputz IX, among others)—a misconception already remarked by 1032 Van der Made (2017). Furthermore, Sher (1997) did not consider the samples from Le Vallonnet and Durfort, 1033 which based on both morphology and chronology undoubtedly belong to B. schoetensacki even if they fall 1034 within the variation of most of the specimens that he attributed to B. priscus. The “stoutness” biplot, show 1035 that all the metacarpals from Mauer indeed share the same diagram area with other populations of B. 1036 schoetensacki, being, altogheter with Le Vallonnet sample, among the slenderest ones; whereas the 1037 metacarpals from Mosbach are characterized by a more pronounced stoutness (Fig. 9). Moreover, based on 1038 the specimens morphologically closer to B. schoetensacki, our Log10 diagrams indicate that the Mauer sample 1039 fits with the characters of B. schoetensacki and that the Mosbach specimens feature slightly stouter 1040 metacarpals than that from Mauer and other B. schoetensacki populations (Fig. 9). The more massive, B. 1041 priscus-like proportions of the Mosbach material further agrees with the derived and “priscoid” cranial 1042 features displayed by some cranial remains from this locality (Sala, 1986). In the two PCAs (Fig. 10), the 1043 Mauer/Mosbach metacarpals cluster in the area occupied by the species B. schoetensacki and B. priscus, with 1044 the Mauer specimens being recognizable by their slenderer proportions. Overall, our results confirm that the 1045 Bison sample from Mosbach display an increased robusticity, suggesting the presence of very large B. priscus- 1046 like morphologies in the German site. For this reason, we prefer to use open nomenclature when assigning 1047 Mosbach remains to B. cf. schoetensacki. 39 1048 In turn, the few analyzed metatarsals from Mauer and Mosbach are quite heterogeneous. In the 1049 “stoutness” diagrams (Fig. 9), all the Mosbach metatarsals are long and stout, resembling those of B. priscus. 1050 In turn, one of the two metatarsals from Mauer (DMSTD in Schertz, 1936b) fits with the variation of B. 1051 schoetensacki, whereas the other (MAU402) displays an extremely slender structure, similar to that of B. 1052 menneri. 1053 A complete revision of the cranial and postcranial material from Mauer and Mosbach would be 1054 required: (1) to clarify the diagnosis of B. schoetensacki (especially based on the sample from Mauer, which is 1055 the type locality) and (2) to confirm the possible first co-occurrence with larger “priscoid” forms (in the case 1056 of the larger specimens from Mosbach). An earlier co-occurrence of B. schoetensacki and B. priscus at Mosbach 1057 cannot be ruled out, as it is consistent with our results, which indicate relatively stout metacarpal proportions 1058 with a wide variation in the sample (Fig. 9). Moreover, it has to be considered that, even if most of the Mosbach 1059 material comes from the upper layers (Mosbach 2) roughly coeval to Mauer and Isernia, its lowermost 1060 sediments are dated to around 1.0 Ma (Koenigswald and Tobien, 1987). Unfortunately, the stratigraphic 1061 provenance of the remains was not recorded. Nonetheless, we disagree with Schertz (1936b) and Sher (1997) 1062 and state that the entire sample from Mauer and, possibly, some of the Mosbach specimens, are attributable 1063 to B. schoetensacki and not to B. priscus. 1064 1065 5.4.4. Süssenborn 1066 Flerov (1969) referred the large bovid remains from Süssenborn (Germany; Middle Pleistocene, ca. 1067 0.6 Ma) to two subspecies of B. schoetensacki (B. schoetensacki schoetensacki and B. schoetensacki 1068 lagenocornis) based on cranial characters without any reference to the metapodials. Bison schoetensacki 1069 lagenocornis was subsequently synonymized with the nominotypical one by Sala (1986), who found that the 1070 roughly coeval B. schoetensacki from Isernia displayed a range of cranial morphologies that encompassed the 1071 differences between the two purported subspecies, which were reinterpreted as resulting from sexual 1072 dimorphism. 1073 Sher (1997) concluded that the metacarpals from Süssenborn are quite heterogeneous and probably 1074 referable to more than a single species. Indeed, our results (Fig. 9) indicate that three out of five metacarpals 40 1075 fall in the range of B. schoetensacki from Mosbach, whereas one fits within B. menneri from Untermassfeld, 1076 and the remaining one more closely resembles Bison sp. from Venta Micena. However, Sher’s (1997) 1077 interpretation was based again on a misinterpretation of the size and proportions of B. priscus forelimbs. The 1078 “stoutness” diagram and both the PCAs (Fig. 9) shows that the three stoutest metacarpals from Süssenborn 1079 (IQW 1965/2330, IQW 1965/2325, IQW 1965/2319), referred to B. priscus by Sher (1997), fit well within the 1080 variation of male specimens of B. schoetensacki from Le Vallonnet, Durfort, and the VCS, while the smallest 1081 specimen from Süssenborn (IQW 1965/2331) is also very close to female specimens of B. schoetensacki from 1082 the same sites. The only exception is represented by the metacarpal IQW 1965/2333, here attributed to B. cf. 1083 menneri, which probably comes from earlier deposits than the rest of the assemblage due to the different type 1084 of preservation from most Süssenborn fossils and its extremely elongate and slender morphology. In turn, the 1085 metatarsals from Süssenborn fit quite well with the available sample of B. schoetensacki in the “stoutness”and 1086 the PCAs (Figs 9b, 10c–d). In summary, the bulk metapodials from Süssenborn can be confidently referred to 1087 a stout form of B. schoetensacki, very similar to those from the VCS and Durfort. 1088 1089 5.4.5. Isernia and Cesi 1090 The site of Isernia La Pineta (southern Italy; 0.58 Ma; Coltorti et al., 2005; Peretto et al., 2015) has 1091 yielded a rich sample of B. schoetensacki cranial and postcranial remains (Sala, 1986), although no complete 1092 long bones are preserved due to carcass exploitation by hominins (Sala, 1983). The distal epiphysis of some 1093 metapodials (e.g., IS.F.1979.t.3q(1) and IS.I.q.73.t.3(1)) has large mediolateral and anteroposterior diameters, 1094 resembling the proportions of heavily-built forms of B. schoetensacki. On the other hand, complete astragali 1095 and calcanei fit well with those of Bison s.s. and are overall larger than those of earlier populations of B. 1096 schoetensacki from Durfort and the VCS (Tables S25, S27). We therefore conclude that the rich sample from 1097 Isernia attests to the presence of a stout form of B. schoetensacki in Italy during the earliest Middle 1098 Pleistocene, exhibiting some postcranial features akin to the “priscoid” forms. Among the few remains of Bison 1099 from the early Middle Pleistocene site of Cesi (central Italy; Ficcarelli et al., 1997) the most informative is a 1100 complete metacarpal. The size and proportions fit with the ones observed in the B. schoetensacki sample in 41 1101 both the stoutness diagram and the PCA (Fig. 9), confirming that in the early Middle Pleistocene this species 1102 was already populating the Italian Peninsula. 1103 5.4.6. Cromer Forest-bed, Boxgrove, and Westbury 1104 The bovids from various sites of the Cromer Forest-bed Formation (eastern UK), Boxgrove and 1105 Westbury (western UK) collectively encompassing the whole early Middle Pleistocene (ca. 0.7–0.4 Ma; Breda 1106 et al., 2010), have been described by several authors (Freudenberg, 1914; Sala, 1986; Breda et al., 2010). Two 1107 horn cores (M/6559 and M/1426) from an unknown locality were attributed to B. schoetensacki (Freudenberg, 1108 1914; Flerov, 1969; Sala, 1986), and according to the latest review (Breda et al., 2010) multiple bovids would 1109 be recorded by teeth and a few postcranial remains in the Forest-bed Formation: B. cf. schoetensacki, B. 1110 priscus, and cf. Bos primigenius. 1111 The “stoutness” diagram (Fig. 9a) indicates that the metapodials from most of these sites (Boxgrove, 1112 Trimingham, Sidestrand, Ostend, and Palling) fit with the general proportions of stouter B. schoetensacki from 1113 Süssenborn, Durfort, Mosbach, and the VCS, whereas the three metacarpals from the yellow breccia of 1114 Westbury-sub-Mendip are stouter and more closely resemble B. priscus, as already suggested by Breda et al. 1115 (2010). 1116 The only available complete radius, which comes from Pakefield, fits with the variation of both B. 1117 schoetensacki and B. menneri, while the one from the younger site of Boxgrove falls far from these taxa and 1118 more closely resembles in length and stoutness those of B. priscus (Fig. 9a). The two radii from Westbury also 1119 overlap the borders of the “priscoid” variation (Fig. 9c). We therefore conclude that the material from the 1120 Cromer Forest-bed Formation most likely probably belongs to B. schoetensacki, whereas the assignment of 1121 the small sample from Boxgrove is more debatable, being attributed here to Bison sp. Finally, the Westbury 1122 sample is referred to B. cf. priscus due to the stoutness of both radii and metapodials. 1123 5.4.7. Châtillon-Saint-Jean and Siréjol cave 1124 Mourer- Chauvire (1972) described a large number of bovid remains from the late Middle Pleistocene 1125 site of Châtillon-Saint-Jean (France). The author referred to B. schoetensacki the smaller and slender 1126 postcranial specimens and to B. priscus the larger and stouter ones. Our morphometric analyses on the 1127 metacarpals question these attributions. On the whole, the Châtillon-Saint-Jean assemblage is characterized 42 1128 by the presence of a large bison featuring quite large metacarpals (Fig. 9) and long and very wide diaphysis 1129 (Fig. 10) fitting with the variation of B. priscus. The presence of two morphologically distinct groups is here 1130 interpreted as the result of intraspecific differences (i.e., sexual dimorphism) in the same sample of B. priscus 1131 (Fig. 9, Table S34), confirming the interpretation of Brugal (1985). The size and stoutness of the Châtillon-Saint- 1132 Jean bison metacarpals resemble those recorded for the very large-sized B. priscus from Romain-la-Roche, 1133 referred to B. priscus priscus together with the Taubach sample (Flerov, 1976; Vercoutère and Guérin, 2010), 1134 and for the bison from Tiraspol, referred to B. schoetensacki by Flerov (1972); and then to B. aff. priscus by 1135 Sher (1997). 1136 The sample from the Late Pleistocene cave of Siréjol (France), attributed to B. schotensacki (Guérin 1137 and Philippe, 1971), then reported to B. priscus with a nov. ssp. (Brugal, 1985, 1999) is characterized by 1138 relatively short and robust metacarpals. The “stoutness” and Log10 ratio diagrams (Fig. 9) as well as the PCA 1139 (Fig. 10) show that the distal and proximal epiphysis width is larger (relative to total length) than in B. 1140 schoetensacki, fitting with the variation of B. priscus. Nonetheless, the bison from Siréjol is distinguishable 1141 from latest Pleistocene B. priscus based on the significantly narrower metacarpal diaphysis compared with the 1142 distal and proximal epiphysis (especially the specimens n°100 000+658, n°100 000+664, n°100 000+665, n°100 1143 000+667, n°100 000+668; Figs. 9–10). This latter difference is particularly interesting in the light of recent 1144 molecular studies (Palacio et al., 2017; Grange et al., 2018), which recognize the Siréjol specimens as 1145 genetically distinct from B. priscus and close to the extant B. bonasus. Further discoveries from the same site 1146 and/or coeval sites are needed to investigate the intriguing presence of another bison species in the European 1147 Late Pleistocene, hitherto only suggested by molecular data. 1148 1149 6. Discussion 1150 6.1. Metapodials variation within Bison schoetensacki 1151 The metacarpal IPS14917 from CGRD2 (ca. 1.1 Ma) is quite long and slightly slenderer than the rest of 1152 the sample, which comes from younger layers (ca. 1.0–0.86 Ma). The “stoutness” diagram and Log10 ratio 1153 diagrams (Fig. 9a-e) show that IPS14917 fits with the size and proportions of B. schoetensacki from the type 1154 locality of the species, Mauer, and the pre-Jaramillo cave site of Le Vallonnet (ca. 1.2 Ma). The Mauer fauna is 43 1155 characterized by taxa typical of forest environments (Soergel, 1914; Breda and Marchetti, 2005; Kahlke et al., 1156 2013), the palynological study performed on the fossiliferous layers of Le Vallonnet suggest that the site might 1157 have been a tree refugia in a period of relatively dry and cold phase. The stout metapodials from EVT10-EVT12 1158 exhibit more “priscoid” proportions, similar to (but overall smaller than) the metapodials of the steppe- 1159 adapted Late Pleistocene B. priscus mediator. From EVT12, two slenderer metacarpals were also recovered, 1160 interpreted as the only female metacarpals of the collection. The EVT7-CGRD7 metacarpals are characterized 1161 by less stout limb proportions, similar to the remains from the sites of Durfort, Süssenborn, and Cromer forest- 1162 bed, all sites featuring a mixture of open and forested landscapes (Brugal, 1995; Stuart and Lister, 2001; van 1163 Asperen and Kahlke, 2011). 1164 The lowermost layers of VCS (CGRD2-CGRD4) span from 1.1 to 1.0 Ma (Madurell-Malapeira et al., 1165 2010). Paleobotanical data indicate that the VCS pre-Jaramillo period was characterized by a relatively warm- 1166 temperate and humid paleoenvironment with a mixture of habitats dominated by wooden landscapes, as 1167 suggested by the abundance of arboreal taxa (Mijarra et al., 2007). This fits with the aforementioned 1168 paleoenvironmental reconstructions of the nearly coeval and geographically close site of Le Vallonnet and of 1169 the younger site of Mauer. The dental wear pattern of the large herbivore remains from the VCS sin-Jaramillo 1170 layers (1.0–0.99 Ma; EVT10-EVT12) points to a generally abrasive diet, suggesting an environment with a 1171 significant predominance of open dry grasslands (Strani et al., 2019). This might explain the stouter and 1172 “priscoid” metapodials unearthed from these layers. The same study, performed on the post-Jaramillo layers 1173 of VCS dated to 0.86–0.78 Ma (EVT7-CGRD7), suggests an increase of average humidity and a relative 1174 expansion of forested habitats, in accordance with the study of the micromammal fauna (Lozano-Fernandez 1175 et al., 2015; Strani et al., 2019). This fits with the Durfort, Cromer Forest-bed, and Süssenborn inferred 1176 paleoenvironments mentioned above. In sum, the metapodial proportions of the VCS Bison sample, though 1177 characterized by an overall homogeneous morphology, shows a certain degree of variation when sub-samples 1178 from different stratigraphic contexts are compared. This variation shows a remarkable correlation with the 1179 dominant paleoenvironment conditions reconstructed along the composite section, with an increase of limbs 1180 stoutness corresponding to more arid conditions and slenderness to more humid ones, also tracing the trend 1181 observed in other B. schoetensacki samples. A similar interpretation has been even anticipated by Flerov 44 1182 (1979), who stated that the slenderer and taller forms of Bison are better adapted to closed and forested 1183 environments as opposite to the stouter ones, which are moresuited to open and arid conditions. This idea is 1184 furtherly confirmed by the extremely slender B. menneri from Untermassfeld and the massive but very tall B. 1185 priscus priscus from Taubach and Romain-la-Roche, all sites characterized by a mixture of close and open 1186 habitats (Sher, 1997; Argant, 2010; van Asperen and Kahlke, 2011). On the contrary, B. priscus mediator, 1187 featuring the typical “priscoid” proportions with short and large metapodials, was one of the most 1188 representative taxa of the so-called Mammoth Steppe fauna during the Late Pleistocene (Kahlke, 1999). 1189 1190 6.2. The metapodial morphology of Bison schoetensacki 1191 Our comparative analysis indicate that the metacarpals are the most useful postcranial skeletal 1192 elements for distinguishing the different species of Bison s.l. Most of the metacarpals attributable to B. 1193 schoetensacki are characterized by the following features (ranges based on all the samples attributed to B. 1194 schoetensacki in Table 16): total length between 225–270 mm, ratio of distal width to total length between 1195 27–34%, ratio of midshaft width to total length between 16–21%, and ratio of proximal width to total length 1196 between 28–36%. In early Middle Pleistocene sites with uncertain chronology and potentially 1197 redeposited/mixed material, such as Mauer, Mosbach, and Süssenborn, some metacarpals fall outside the 1198 aforementioned variation ranges and are not considered in this discussion. With the exception of these 1199 remains and the clearly different B. menneri from Untermassfeld, all the remaining bison samples from 1200 Epivillafranchian-early Middle Pleistocene European sites are distinguishable from earlier and later samples, 1201 and are here referred to B. schoetensacki. 1202 Our morphometric data show some morphometric variation both among samples from different localities, 1203 especially for the metacarpals (Fig. 9). This variation is probably due to ecophenotypic changes in limb 1204 robusticity and size through time and space in a single wide-ranging and long-lasting species (see Section 6.1.). 1205 However, excluding these small intraspecific differences, our results show that all the putative B. schoetensacki 1206 samples are showing no significant differences. After 0.5 Ma, with the first occurrence of the massive B. 1207 priscus, it is virtually impossible to discern female B. priscus from male B. schoetensacki individuals based on 1208 postcranial material, leading to possibly erroneous taxonomic attributions in sites where the two species 45 1209 purportedly co-occur, particularly in the Middle and Late Pleistocene (e.g., Chatillon-Saint-Jean and Sirejol 1210 cave; Mourer-Chauvire, 1972), but determined as male and female of B. priscus (Brugal, 1999; Grange et al., 1211 2018). This is also relevant in the light of molecular studies (Palacio et al., 2017; Grange et al., 2018; Vershinina 1212 et al., 2019) indicating the presence of various Late Pleistocene Bison lineages in Eurasia, which are not easily 1213 distinguishable on morphometric grounds. 1214 6.3. Sexual dimorphism 1215 Fifteen complete metacarpals from the sites of Mauer, Le Vallonnet, Durfort, EVT12 of VCS, 1216 Trimingham, and Süssenborn are slenderer and overall smaller than the rest of the sample attributed to B. 1217 schoetensacki with values of DEW/Lmax% less than 30 (Figs. 9–10). These metacarpals are, here, attributed to 1218 females, indicating a significant degree of sexual dimorphism, as it is frequently observed in large bovids like 1219 bison. To quantify the magnitude of these differences and to assess if it is similar to those recorded for extant 1220 species, we applied the equations by Schertz (1936a) (see Section 3) to all the available samples in which 1221 putative males and females were recognized (B. schoetensacki from VCS layer EVT12, Durfort, and Süssenborn; 1222 B. menneri from Untermassfeld; B. priscus from Châtillon-Saint-Jean, North Sea, and Krasny Yar; extant B. 1223 bonasus, and B. bison; Table S34). The results obtained for all analyzed EVT12 show that the sample has similar 1224 values to those of the extant and fossil Bison species (except for an unusual high value for DW). 1225 In most of the studied localities, including VCS, males represent the majority of the metacarpal record 1226 (e.g., 75% in VCS, 53% in Le Vallonnet, 67% in Durfort, 81% in Taubach, 56% in North Sea), apparently in 1227 contrast with what is observed in extant gregarious large mammals, in which female individuals are normally 1228 more abundant (Gower et al., 2019). This imbalance in the frequency of sex distribution has been reported for 1229 several fossil collections of large mammals, including carnivorans such as Ursus, and female-herd-based 1230 herbivores such as Mammuthus and Bison (Pečnerová et al., 2017; Gower et al., 2019). This might be explained 1231 by (1) bias in the fossil collection (e.g., focused on larger remains, especially in historical times), (2) taphonomic 1232 reasons (e.g., greater resistance of larger and more massive skeletal remains against destructive taphonomic 1233 agents), (3) different behaviors between the two sexes (i.e., segregation of males) and/or (4) seasonality factor 1234 for the bone accumulation. Gower et al. (2019) genetically sexed both cranial and postcranial bones belonging 46 1235 to 186 fossil individuals of Bison spp. from several Holarctic sites and found that the sample was composed by 1236 ca. 75% of males, a percentage very similar to that estimated for the VCS sample. 1237 1238 6.4. Final remarks on the postcranial differences among Leptobos and Bison 1239 Metatarsals are less diagnostic due to their greater variation, but their distal end is nevertheless useful 1240 to distinguish large and stout Bison s.s. from the slenderer Leptobos and B. (Eobison). Of the most common 1241 variables used in the literature, only a few are useful to discriminate among Bison species (Fig. 10), particularly 1242 metapodial total length, robustness of diaphysis and distal epiphysis width (Scott and Barr, 2014), despite 1243 some overlap. These differences in the proportions of metapodials, and the appendicular skeleton in general, 1244 are often linked to ecophenotypic variation in bovids, and must be therefore taken with caution when making 1245 taxonomic attributions or inferring phylogenetic relationships (Brugal, 1999; Scott and Barr, 2014). 1246 The astragalus is also often used as a diagnostic element for large bovids. The geometric morphometric 1247 analysis of astragali performed by Maniakas and Kostopoulos (2017b) on a large sample of Bison suggests that 1248 the shape of this bone is strongly (but not only) related to habitat preferences. Although there are some 1249 differences between the Bos and Leptobos/Bison, astragalus is not diagnostic enough for reliable taxonomic 1250 distinction between Leptobos and Bison. Size and proportions of the talus may help in distinguishing large and 1251 stout Bison s.s. from small and slender Leptobos and B. (Eobison), but the overlap between the ranges of 1252 variation is significant. In our opinion, the radius is one of the more useful long bones to distinguish Leptobos 1253 and Bison, and also different species of Bison s.l. In particular, B. menneri and B. (Eobison) have significantly 1254 slenderer radii than B. priscus, while B. schoetensacki is characterized by intermediate proportions (Fig. 9). 1255 Finally, phalanges, carpals, and tarsals (other than the astragalus) can only give some hints on the overall size 1256 and built, but are too homogeneous among large bovids to have any taxonomic value. Similarly, humeri and 1257 tibiae are taxonomically uninformative within the Leptobos/Bison group. 1258 Postcranial differences between Leptobos and Bison almost exclusively relate to dimensions and 1259 proportions. Some of these morphometric differences are also reflected in the identification of distinct groups 1260 within Bison s.l., which in some cases correspond to specific taxa. In particular, our results allow us to clearly 1261 distinguish among B. (Eobison) (Dmanisi, Venta Micena, Pirro Nord, Capena, Mygdonia Basin), B. menneri 47 1262 (Untermassfeld), B. schoetensacki, and early B. priscus. In turn, the last two species are very polymorphic, 1263 which results in the identification of local samples with peculiar postcranial proportions and/or dimensions, 1264 e.g., B. schoetensacki “slender” (CGRD2 of VCS, Mauer, Le Vallonnet) and “stout” (EVT7, EVT 10, EVT 12, and 1265 CGRD7 of VCS, Durfort, Süssenborn, Cromer Forest-bed, possibly Isernia and Mosbach) forms; B. p. priscus 1266 (Taubach, Romain-la-Roche, Tiraspol, Châtillon-Saint-Jean) and B. p. mediator (North Sea, Habarra, Kiputz IX 1267 and others) subspecies. At least for the case of B. schoetensacki (a deeper look into B. priscus is out of the 1268 scope of this manuscript) these differences might be related to ecophenotypic changes related to 1269 paleoenvironmental conditions (Section 6.1). At the state of the art, we are unable to establish, at least for B. 1270 schoetensacki, whether these differencies can be the result of ecophenotipycal variation within the same 1271 species, or whether they can correspond to different taxa or evolutive trend. From the late Middle Pleistocene 1272 onwards, the second hypothesis could find indirect support in the recognition of different haplotypes in the 1273 ancient DNA of European bison (Palacio et al., 2017), which to date do not correspond to species defined on 1274 a morphological basis. Regarding B. priscus, the two putative subspecies B. p. priscus and B. p. mediator which, 1275 until now, had been distinguished on the basis of cranial features and size, display substantial differences in 1276 proportions also in the metapodials (especially metacarpals; Grange et al., 2018; Fig. 9). 1277 1278 7. Conclusions 1279 The described bison sample from the VCS is attributed to B. schoetensacki. The records of this species in 1280 the lower layers of CGR and the roughly coeval site of Le Vallonnet in France represent the first occurrences 1281 of Bison s.s. in Europe, dating to the beginning of the Epivillafranchian (ca. 1.2–1.1 Ma). Bison schoetensacki 1282 can therefore be considered a biochronological marker for the beginning of this biochron (Kahlke et al., 1283 2007; Bellucci et al., 2015), in association with the first occurrence of Megaloceros savini (see Madurell- 1284 Malapeira et al., 2019) and the reappearance of Sus strozzii in Europe after the latest Villafranchian “suid 1285 gap” (Cherin et al., 2020). According to our results, B. schoetensacki was widely distributed across Eastern 1286 and Central Europe (from UK to Italy, and from Iberia to Germany) between 1.2 and 0.6–0.5 Ma, 1287 representing the most common large bovid during the Epivillafranchian and early Middle Pleistocene. In the 1288 absence of sufficiently complete cranial remains, the morphometric analysis on postcranial bones (especially 48 1289 metapodials) represents the most powerful tool for diagnosing Pleistocene bovine taxa. In particular, 1290 metapodial proportions enable a reliable distinction among the genera Leptobos, Bison, and Bos, as well as 1291 among different Bison species. For the two most polymorphic and geographically widespread species, 1292 namely B. schoetensacki and B. priscus, it is even possible to distinguish several morphotypes, whose 1293 biological meaning (local ecophenotypic variation vs taxonomic differences) should be subject to further 1294 research. The relatively wide range of morphometric variation exhibited by the VCS sample of B. 1295 schoetensacki may be related to ecophenotypic changes and/or local adaptations in response to 1296 environmental changes that affected the VCS area during the Epivillafranchian (Strani et al., 2019). In 1297 particular, these changes may be related to the important ecological transitions that occurred during the 1298 EMPT. Moreover, the increased size and robusticity observed in the long bones from the VCS intermediate 1299 layers (EVT10 and EVT12) represent the earliest record of “priscoid” characters in Bison s.s. at ca. 1.0 Ma. 1300 1301 Acknowledgments 1302 This work isfunded by the Agencia Estatal de Investigación–European Regional Development Fund of the 1303 European Union (CGL2016-76431-P and CGL2017-82654-P, AEI/FEDER-UE) and the Generalitat de Catalunya 1304 (CERCA Program). LS is supported by the FI AGAUR fellowship (ref. 2020 FI_B1 00131) funded by the Secretaria 1305 d’Universitats i Recerca de la Generalitat de Catalunya and the European Social Fund. D.M.A. and J.M.M. are 1306 members of consolidated research group 2017 SGR 116 (AGAUR, Generalitat de Catalunya). We would like to 1307 thank Marzia Breda for sending us measurements of the UK bovids; Marisa Blume and Oliver Sandrock for 1308 measurements of the bovids from Mosbach and Mauer; Lionel Cavin and Corinne Charvet for sending us 1309 measures of the bovids from Romain-la-Roche; Paola Romi (Soprintendenza Archeologia Belle Arti e Paesaggio 1310 dell’Umbria) for the possibility to study the Pietrafitta bovid collection at the MPLB; Benedetto Sala and Claudio 1311 Berto for the possibility to study the Isernia bovid collection at the UNIFE and MPPPL; Alessandro Blasetti for 1312 the possibility to study the Cesi bovid collection at the MuPA; Elisabetta Cioppi and Luca Bellucci for the 1313 possibility to study the Valdarno bovid collection at the IGF. Comments by Marzia Breda and an anonymous 1314 reviewer helped to significantly improve the manuscript. 49 1315 References 1316 Agadzhanyan, A.K., Vislobokova, I.A., Shunkov, M.V. and Ulyanov, V.A., 2017. Pleistocene mammal fauna of 1317 the Trlica locality, Montenegro. Fossil Imprint 73, 93–114. 1318 Alekseeva, L.I., 1967. To the history of subfamily Bovinae during the Pleistocene in the European USSR. In: 1319 Paleontologiya, geologiya i poleznye iskopaemye Moldavii Vyp. 2. Shtiintsa, Kishinev, pp. 123–127 (in 1320 Russian). 1321 Alekseeva, L.I., 1977. Theriofauna of the Early Anthropogene of Eastern Europe. Trudy Geol. Inst. Akad. 300, 1322 1–214 (in Russian). 1323 Ambert P., Brugal J.P., Houles N., 1996. Le maar du Riège (Hérault, France): géologie, paléontologie, 1324 perspectives de recherches. C. R. Acad. Sc. Paris. 322, 125–132. 1325 Argant, J., 2010. Palynologie des coprolithes d’hyène de Romain-la-Roche (Doubs, France): apport 1326 paléoenvironnemental. Rev. Paléobiol. 29, 473–476. 1327 Asperen van, E.N., Kahlke, R.D., 2017. Dietary traits of the late early Pleistocene Bison menneri (Bovidae, 1328 Mammalia) from its type site Untermassfeld (Central Germany) and the problem of Pleistocene 1329 ‘wood bison’. Quat. Sci. Rev. 177, 299–313. 1330 Ayrolles P., 1973. Essai de distinction des genres Bos et Bison d’apèrs deux métacarpiens. Etudes 1331 préhistoriques 7, 13–15. 1332 Bellucci, L., Sardella, R. and Rook, L., 2015. Large mammal biochronology framework in Europe at Jaramillo: 1333 the Epivillafranchian as a formal biochron. Quat. Int. 389, 84–89. 1334 Bibikova, V.I., 1958. Some distinguishing features in the bones of the genera Bison and Bos. Bull. Mosk. 1335 Obschtschestwa Isp. Privoda NS Otdel Biol. 63, 23–35. 1336 Breda, M., Collinge, S.E., Parfitt, S.A., Lister, A.M., 2010. Metric analysis of ungulate mammals in the early 1337 Middle Pleistocene of Britain, in relation to taxonomy and biostratigraphy: I: Rhinocerotidae and 1338 Bovidae. Quat. Int. 228, 136–156. 50 1339 Brugal J.P., -1983. Application des analyses multidimensionnelles à l'étude systématique du squelette des 1340 membres des Grands Bovidés Pléistocènes (Grottes de Lunel-Viel, Hérault); Perspectives évolutives. 1341 Ph.D. Dissertation, Université Aix-Marseille II. 1342 Brugal, J.-P., 1985. Le Bos primigenius Boj.,1827 du Pléistocène moyen des grottes de Lunel-Viel (Hérault). 1343 Bull. Mus. Anthr. Préh. Monaco 28, 7–62. 1344 Brugal, J.-P., 1995. Le bison (Bovidae, Artiodactyla) du Pléistocène moyen ancien de Durfort (Gard, France). 1345 Bull. Mus. Natl. Hist. Nat. 16C 2–4, 349–381. 1346 Brugal, J.P., 1999. Etude des populations de grands Bovidés européens: intérêt pour la connaissance des 1347 comportements humains au Paléolithique. In: Brugal, J.P., David, F., Enloe, J.G., Jaubert, J. (Eds.), Le 1348 Bison: gibier et moyen de subsistance des hommes du Paléolithique aux Paléoindiens des Grandes 1349 Plaines, Ed. APDCA, Antibes (1999), pp. 85–104. 1350 Brugal J.P., Fosse P. 2005. Les grands bovidés (Bison cf. schoetensacki) du site pléistocène moyen de La 1351 Vayssière (Aveyron, France). In: Crégut-Bonnoure, E. (Ed.), Les Ongulés Holarctiques du Pliocène et 1352 du Pléistocène Quaternaire H.S. 2, 75–80 1353 Bukhsianidze, M., 2005. The fossil Bovidae of Dmanisi. Ph.D. Dissertation, Università degli Studi di Ferrara. 1354 Bukhsianidze, M., 2020. New Results on Bovids from the Early Pleistocene site of Untermassfeld. In: Kahlke, 1355 R.D. (Ed.), Das Pleistozänvon Untermaßfeld bei Meiningen (Thüringen). Teil 4. Habelt-Verlag, Bonn, 1356 pp. 1169–1195. 1357 Burchak-Abramovich, N.I., Gadzhiev, D.V., Vekua, A.K., 1994. On a new Pleistocene Bovine from Eastern 1358 Georgia. Teriologii Paleoteriologija 253-261. (in Russian). 1359 Caloi, L., Palombo, R., 1980. La fauna quaternaria di Venosa, i Bovidi. Boll. Geol. It. 100, 101–140. 1360 Cherin, M., D’Allestro, V., Masini, F., 2019. New bovid remains from the Early Pleistocene of Umbria (Italy) 1361 and a reappraisal of Leptobos merlai. J. Mamm. Evol. 26, 201–224. 51 1362 Cherin, M., Alba, D.M., Crotti, M., Menconero, S., Moullé, P.É., Sorbelli, L., Madurell-Malapeira, J., 2020. The 1363 post-Jaramillo persistence of Sus strozzii (Suidae, Mammalia) in Europe: New evidence from the 1364 Vallparadís Section (NE Iberian Peninsula) and other coeval sites. Quat. Sci. Rev. 233, 106–234. 1365 Clark, P.U., Archer, D., Pollard, D., Blum, J.D., Rial, J.A., Brovkin, V., Mix, A.C., Pisias, N.G., Roy, M., 2006. The 1366 middle Pleistocene transition: characteristics, mechanisms, and implications for long-term changes 1367 in atmospheric pCO2. Quat. Sci. Rev. 25, 3150–3184. 1368 Coltorti, M., Feraud, G., Marzoli, A., Peretto, C., Ton-That, T., Voinchet, P., Bahain, J.J., Minelli, A., 1369 Hohenstein, U.T., 2005. New 40Ar/39Ar, stratigraphic and palaeoclimatic data on the Isernia La 1370 Pineta Lower Palaeolithic site, Molise, Italy. Quat. Int. 131, 11–22. 1371 Croitor, R., 2010. Critical remarks on genus Bison (Bovidae, Mammalia) from Pleistocene of Moldova. Revista 1372 Archeologica 5, 172–188. (in Russian) 1373 Croitor R., 2016. Genus Bison (Bovidae, Mammalia) in Early Pleistocene of Moldova. In: Coropceanu E. (Ed.), 1374 Materialele Conferinţei știinţifi ce naţionale cu participare internaţională “Mediul și dezvoltare 1375 durabilă”, Ediţia a III-a, Chișinău, pp. 14–20. 1376 David, A.I., Svistun, V.I., 1981. Bison remains from Upper Pliocene and Lower Pleistocene deposits of 1377 Moldova and South Ukraine.In: Biostratigraphy of Anthropogene and Neogene of South-West of 1378 USSR. Shtiintsa; Kishinev pp. 3–15. (in Russian) 1379 Delpech F., 1972. Fouilles de sauvetage dans le gisement magdalénien de Fongaban. Commune de Saint- 1380 Emilion (Gironde), 3éme partie: la Faune. L’Anthropologie 76, 615–629. 1381 Lumley, H. de, Kahlke, H.D., Moigne, A.M., Moullé, P.É., 1988. Les faunes de grands mammifères de la grotte 1382 du Vallonnet. L’Anthropologie 92, 465–495. 1383 Demirel, F.A., Mayda, S., 2014. A new early Pleistocene mammalian fauna from Burdur Basin, SW Turkey. 1384 Russ. J. Theriol. 13, 55–63. 52 1385 Drees, M., 2005. Sexual dimorphism in Pleistocene Bison priscus (Mammalia, Bovidae) with a discussion on 1386 the position of Bison schoetensacki. Senck. Leth. 85, 153–157. 1387 Duvernois, M. P., 1990. - Les Leptobos (Mammalia, Artiodactyla) du Villafranchien d'Europe occidentale. Doc. 1388 Lab. Géol., Lyon, 113, 1–213 pp. 1389 Ficcarelli, G., Abbazzi, L., Albianelli, A., Bertini, A., Coltorti, M., Magnatti, M., Masini, F., Mazza, P., 1390 Mezzabotta, C., Napoleone, G., Rook, L., 1997. Cesi, an early Middle Pleistocene site in the Colfiorito 1391 Basin (Umbro‐Marchean Apennine), central Italy. J. Quat. Sci. 12, 507–518. 1392 Fischer, K.H., 1965. Bisonreste (Bison schoetensacki voigtstedtensis ssp. n.) aus den altpleistozänen Tonen 1393 von Voigtstedt in Thüringen. Paläontol. Abh. A Paläozool. 2, 364–377. 1394 Flerov, K.K., 1969. Die Bison-Reste aus den Kiesen von Süßenborn bei Weimar. Paläontol. Abh. 3, 489–520. 1395 Flerov, C.C., 1972. The Most Ancient Bisons and the History of Genus Bison. In: Teriologiya, vol 1. Nauka, 1396 Novosibirsk, pp. 81–86 (in Russian). 1397 Flerow, K.K., 1975. Die Bison-reste aus den travertinen von Weimar-ehringsdorf. Abh. Zentr. Geol. Inst. 23, 1398 171–199. 1399 Flerov, K.K., 1979. Morphology, systematic, evolution, ecology. In: Sokolov, E.V. (Ed.), European Bison. 1400 Nauka, Moscow, pp. 9–127 (in Russian). 1401 Freudenberg, W., 1914. Die Säugetiere des älteren Quartärs von Mitteleuropa. Geol. Paleont. Abh. NF 12, 1402 533–549. 1403 Froese, D., Stiller, M., Heintzman, P.D., Reyes, A.V., Zazula, G.D., Soares, A.E., Meyer, M., Hall, E., Jensen, B.J., 1404 Arnold, L.J., MacPhee, R.D., 2017. Fossil and genomic evidence constrains the timing of bison arrival 1405 in North America. Proc. Natl. Acad. Sci. USA 114, 3457–3462. 1406 Gee, H., 1993. The distinction between postcranial bones of Bos primigenius Bojanus, 1827 and Bison priscus 1407 Bojanus, 1827 from the British Pleistocene and the taxonomic status of Bos and Bison. J. Quat. Sci. 8, 1408 79–92. 53 1409 Gentili, S., Masini F., 2005. An outline of Italian Leptobos and a first sight on Leptobos aff. vallisarni from 1410 Pietrafitta (early pleistocene, Perugia) Quaternarie Hors Series, 2, 81–89 1411 Geraads, D., 1992: Phylogenic analysis of the tribe Bovini (Mammalia: Artiodactyla). Zoological Journal of the 1412 Linnean Society 104, 193–207. 1413 Gower, G., Fenderson, L.E., Salis, A.T., Helgen, K.M., van Loenen, A.L., Heiniger, H., Hofman-Kamińska, E., 1414 Kowalczyk, R., Mitchell, K.J., Llamas, B., Cooper, A., 2019. Widespread male sex bias in mammal fossil 1415 and museum collections. Proc. Natl. Acad. Sci. USA 116, 19019–19024. 1416 Grange, T., Brugal, J.P., Flori, L., Gautier, M., Uzunidis, A., Geigl, E.M., 2018. The evolution and population 1417 diversity of bison in Pleistocene and Holocene Eurasia: Sex matters. Divers. 10, 65. 1418 Gromov, I.M., Baranova, G.I. (Eds.), 1981. Catalogue of Mammals of the USSR (Pliocene-Recent). Nauka, 1419 Leningrad, pp. 1–456 (in Russian). 1420 Hammer, Ø., Harper, D.A., Ryan, P.D., 2001. PAST: Paleontological statistics software package for education 1421 and data analysis. Paleontol. Electron. 4, 1–9. 1422 Head, M.J., Gibbard, P.L., 2005. Early-Middle Pleistocene transitions: an overview and recommendation for 1423 the defining boundary. Geol. Soc. London Special Publications 247, 1–18. 1424 Hilzheimer, M., 1918. Dritter Beitrag zur Kenntnis der Bisonten. Arch. Naturg. 84, 41–87. 1425 Huguet, R., Vallverdú, J., Rodríguez-Álvarez, X.P., Terradillos-Bernal, M., Bargalló, A., Lombera-Hermida, A.D., 1426 Menéndez, L., Modesto-Mata, M., Van der Made, J., Soto, M., Blain, H.A., 2017. Level TE9c of Sima 1427 del Elefante (Sierra de Atapuerca, Spain): A comprehensive approach. Quat. Int. 433, 278–295. 1428 Jungers, W.L., Falsetti, A.B., Wall, C.E., 1995. Shape, relative size, and size-adjustments in morphometrics. 1429 Yrbk. Phys. Anthropol. 38, 137–161. 1430 Kahlke, R.D., 1999. The History of the Origin, Evolution and Dispersal of the Late Pleistocene Mammuthus- 1431 Coelodonta Faunal Complex in Eurasia (large mammals). Fenske Companies, Rapid City. 54 1432 Kahlke, R.-D., 2007. Late Early Pleistocene European large mammals and the concept of an Epivillafranchian 1433 biochron. In: Kahlke, R.-D., Maul, L.C., Mazza, P. (Eds.), Late Neogene and Quaternary Biodiversity 1434 and Evolution: Regional Developments and Interregional Correlations. Volume II. Proceedings of the 1435 18th International Senckenberg Conference (VI International Palaeontological Colloquium in 1436 Weimar). Cour. Forsch.-inst. Senck. 259, 265–278. 1437 Kahlke, R.D., García, N., Kostopoulos, D.S., Lacombat, F., Lister, A.M., Mazza, P.P., Spassov, N. and Titov, V.V., 1438 2011. Western Palaearctic palaeoenvironmental conditions during the Early and early Middle 1439 Pleistocene inferred from large mammal communities, and implications for hominin dispersal in 1440 Europe. Quat. Sci. Rev. 30, 1368–1395. 1441 Khan, M.A., Kostopoulos, D.S., Akhtar, D.S., Nazir, M., 2010. Bison remains from the upper Siwaliks of 1442 Pakistan. N. Jahrb. Geol. Palaontol. Abh. 258, 121–128. 1443 Kostopoulos, D.S., Maniakas, I., Tsoukala, E., 2018. Early bison remains from Mygdonia Basin (Northern 1444 Greece). Geodiversitas 40, 283–319. 1445 Von Koenigswald, W., Tobien, H., 1987. Bemerkungen zur Altersstellung der pleistozänen Mosbach-Sande 1446 bei Wiesbaden. Geologisches Jahruch Hessen 115, 227–237. 1447 Lehman U., 1949. Der Ur in diluvium Deutschlands und seine verbreitung. Neues Jb. Miner. Abh. 90, 163– 1448 266. 1449 Lopatin, A.V., Vislobokova, I.A., Lavrov, A.V., Startsev, D.B., Gimranov, D.O., Zelenkov, N.V., Maschenko, E.N., 1450 Sotnikova, M.V., Tarasenko, K.K. and Titov, V.V., 2019. The Taurida Cave, a new locality of Early 1451 Pleistocene vertebrates in Crimea. Dokl. Biol. Sci. 485, 40–43. 1452 Madurell-Malapeira, J., Minwer-Barakat, R., Alba, D.M., Garcés, M., Gómez, M., Aurell-Garrido, J., Ros- 1453 Montoya, S., Moyà-Solà, S., Berástegui, X., 2010. The Vallparadís section (Terrassa, Iberian Peninsula) 1454 and the latest Villafranchian faunas of Europe. Quat. Sci. Rev. 29, 3972–3982. 55 1455 Madurell-Malapeira, J., Alba, D. M., Minwer-Barakat, R., Aurell-Garrido, J., & Moyà-Solà, S., 2012. Early 1456 human dispersals into the Iberian Peninsula: A comment on Martínez et al. (2010) and Garcia et al. 1457 (2011). J. of Hum. Evol., 62, 169–173. 1458 Madurell-Malapeira, J., Ros-Montoya, S., Espigares, M.P., Alba, D.M., Aurell-Garrido, J., 2014. Villafranchian 1459 large mammals from the Iberian Peninsula: paleobiogeography, paleoecology and dispersal events. J. 1460 Iber. Geol. 40, 167–178. 1461 Madurell-Malapeira, J., Alba, D.M., Espigares, M.P., Vinuesa, V., Palmqvist, P., Martínez-Navarro, B., Moyà- 1462 Solà, S., 2017. Were large carnivorans and great climatic shifts limiting factors for hominin 1463 dispersals? Evidence of the activity of Pachycrocuta brevirostris during the Mid-Pleistocene 1464 Revolution in the Vallparadís Section (Vallès-Penedès Basin, Iberian Peninsula). Quat. Int. 431, 42–52. 1465 Madurell-Malapeira, J., Sorbelli, L., Bartolini Lucenti, S., Rufí, I., Prat-Vericat, M., Ros-Montoya, S., Espigares, 1466 M.P., Martínez-Navarro, B., 2019. The Iberian latest Early Pleistocene: glacial pulses, large 1467 carnivorans and hominins. In: Martínez- Navarro, B., Palmqvist, P., Espigares, M.P., Ros-Montoya, S. 1468 (Eds.), Libro de Resúmenes XXXV Jornadas de la Sociedad Española de Paleontología, pp. 155–159. 1469 Maniakas, I., Kostopoulos, D.S., 2017a. Morphometric-palaeoecological discrimination between Bison 1470 populations of the western Palaearctic. Geobios 50, 155–171. 1471 Maniakas, I., Kostopoulos, D.S., 2017b. Assessing astragalar morphology and biomechanics in western 1472 Palaearctic Bison populations with geometric morphometrics. C. R. Palevol 16, 783-794. 1473 Martin, T., 1987. Artunterschiede an den Langknochen großer Artiodactyla des Jungpleistozäns 1474 Mitteleuropas. Cour. Forsch-inst. Senck. 96, 1–121. 1475 Martínez-Navarro, B., Pérez-Claros, J.A., Palombo, M.R., Rook, L., Palmqvist, P., 2007. The Olduvai buffalo 1476 Pelorovis and the origin of Bos. Quat. Res. 68, 220–226. 56 1477 Martínez-Navarro, B., Ros-Montoya, S., Espigares, M.P., Palmqvist, P., 2011. Presence of the Asian origin 1478 Bovini, Hemibos sp. aff. Hemibos gracilis and Bison sp., at the early Pleistocene site of Venta Micena 1479 (Orce, Spain). Quat. Int. 243, 54–60. 1480 Masini, F., 1989. I bovini villafranchiani dell’Italia. Ph.D. Dissertation, Università di Modena-Bologna-Firenze- 1481 Roma. 1482 Masini, F., Palombo, M.R., Rozzi, R., 2013. A reappraisal of the Early to Middle Pleistocene Italian Bovidae. 1483 Quat. Int. 288, 45–62. 1484 Merla, G., 1949. I Leptobos Rütim. italiani. Palaeontogr. Ital. 46, 41–155. 1485 Michel, V., Shen, C.-C., Woodhead, J., Hu, H.-M., Wu, C.-C. Moullé, P.-É., Khatib, S., Cauche, D., Moncel, M.- 1486 H., Valensi, P., Chou, Y.-M., Gallet, S., Echassoux, A., Orange, F., de Lumley, H., 2017. New dating 1487 evidence of the early presence of hominins in Southern Europe. Sci. Rep. 7, 10074. 1488 Mijarra, J.M.P., Burjachs, F., Manzaneque, F.G., Morla, C., 2007. A palaeoecological interpretation of the 1489 lower–middle Pleistocene Cal Guardiola site (Terrassa, Barcelona, NE Spain) from the comparative 1490 study of wood and pollen samples. Rev. Palaeobot. Palynol. 146, 247–264. 1491 Minwer-Barakat, R., Madurell-Malapeira, J., Alba, D.M., Aurell-Garrido, J., De Esteban-Trivigno, S., Moyà-Solà, 1492 S., 2011. Pleistocene rodents from the Torrent de Vallparadís section (Terrassa, northeastern Spain) 1493 and biochronological implications. J. Vertebr. Paleontol. 31, 849–865. 1494 Mol, D., Post, K., Reumer, J.W.F., de Vos, J., Laban, C., 2003. Het Gat: preliminary note on a Bavelian fauna 1495 from the North Sea with possibly two mammoth species. Deinsea 9, 253–266. 1496 Mourer-Chauvire, C., 1972. Etude de nouveaux restes de vertébrés provenant de la carrière Fournier à 1497 Châtillon-Saint-Jean. III. Artiodactyles, chevaux, oiseaux. Quaternaire 9, 271–305. 1498 Moullé, P.E., 1992. Les grands mammifères du Pléistocène inférieur de la grotte du Vallonnet (Roquebrune- 1499 Cap-Martin, Alpes-Maritimes). Étude paléontologique des Carnivores, Équidés, Suidés et Bovidés. 1500 Ph.D. Dissertation, Muséum national d'Histoire naturelle, Paris. 57 1501 Moullé, P.É., Lacombat, Echassoux, A., 2006. Apport des grands mammifères de la grotte du Vallonnet 1502 (Roquebrune-Cap-Martin, Alpes-Maritimes, France) à la connaissance du cadre biochronologique de 1503 la seconde moitié du Pléistocène inférieur d’Europe. L’Anthropologie 110, 837–849. 1504 Moyà-Solà, S., 1987. Los bóvidos (Artiodactyla, Mammalia) del yacimiento del Pleistoceno inferior de Venta 1505 Micena (Orce, Granada, Espana). Paleontol. Evol. Mem. Esp. 1, 181–236. 1506 Palacio, P., Berthonaud, V., Guérin, C., Lambourdière, J., Maksud, F., Philippe, M., Plaire, D., Stafford, T., 1507 Marsolier-Kergoat, M.C., Elalouf, J.M., 2017. Genome data on the extinct Bison schoetensacki 1508 establish it as a sister species of the extant European bison (Bison bonasus). BMC Evol. Biol. 17, 48. 1509 Pečnerová, P., Díez-del-Molino, D., Dussex, N., Feuerborn, T., von Seth, J., van der Plicht, J., Nikolskiy, P., 1510 Tikhonov, A., Vartanyan, S., Dalén, L., 2017. Genome-based sexing provides clues about behavior and 1511 social structure in the woolly mammoth. Curr. Biol. 27, 3505–3510. 1512 Peretto, C., Arnaud, J., Moggi-Cecchi, J., Manzi, G., Nomade, S., Pereira, A, Falguères, C., Bahain, J.-J., 1513 Grimaud-Hervé, D., Berto, C., Sala, B., Lembo, G., Muttillo, B., Gallotti, R., Thun Hohenstein, U., 1514 Vaccaro, C., Coltorti, M., Arzarello, M., 2015. A human deciduous tooth and new 40Ar/39Ar dating 1515 results from the Middle Pleistocene archaeological site of Isernia La Pineta, Southern Italy. PLoS ONE 1516 10, e0140091. 1517 Pilgrim, G.E., 1947. The evolution of the buffaloes, oxen, sheep and goats. Zool. J. Linn. Soc. Lond. 41, 272– 1518 286. 1519 Prat F., 1968. Observations sur quelques ossements découverts dans la Basse Terrasse de l’Oise à Moru 1520 (Rhuis, Oise). In La Préhistoire, Problèmes et Tendances. CNRSParis, pp. 337–348 1521 Revilliod P., Dottrens E., 1946, La faune néolithique de la couche profonde de St-Aubin. Etude préliminaire 1522 des phalanges osseuses de Bos taurus domesticus, Revue Suisse de Zoologie, Genève, 53, 739–774. 1523 Rook, L., Martínez-Navarro, B., 2010. Villafranchian: the long story of a Plio-Pleistocene European large 1524 mammal biochronologic unit. Quat. Int. 219, 134–144. 58 1525 Sala, B., 1986. Bison schoetensacki Freud. from Isernia La Pineta (early Mid-Pleistocene – Italy) and revision 1526 of the European species of bison. Palaeont. It. 74, 113–170. 1527 Schertz, E., 1936a. Zur Unterscheidung von Bison priscus Boj. und Bos primigenius Boj. an Metapodien und 1528 Astragalus, nebst Bemerkungen über einige diluviale Fundstellen. Seckenbergiana 18, 37–71. 1529 Schertz, E., 1936b Der Geschlechts-Unterschied an Metapodien von Bison. Senkenbergiana 18, 357–381. 1530 Scott, R.S., Barr, W.A., 2014. Ecomorphology and phylogenetic risk: implications for habitat reconstruction 1531 using fossil bovids. J. Hum. Evol. 73, 47–57. 1532 Shapiro, B., Drummond, A.J., Rambaut, A., Wilson, M.C., Matheus, P.E., Sher, A.V., Pybus, O.G., Gilbert, 1533 M.T.P., Barnes, I., Binladen, J., Willerslev, E., 2004. Rise and fall of the Beringian steppe bison. 1534 Science 306, 1561–1565. 1535 Sher, A.V., 1997. An Early Quaternary Bison population from Untermaßfeld: Bison menneri sp. nov. In: 1536 Kahlke, R.D. (Ed.), Das Pleistozänvon Untermaßfeld bei Meiningen (Thüringen). Teil 2. Habelt-Verlag, 1537 Bonn, pp. 101–180. 1538 Simpson, G.G., 1941. Large Pleistocene felines of North America. Am. Mus. Novit. 1136, 1–27. 1539 Skinner, M.R., Kaisen, O.C., 1946. The fossil Bison of Alaska and preliminary revision of the genus. Bull. Am. 1540 Mus. Nat. Hist. 89, 123–256. 1541 Soubrier, J., Gower, G., Chen, K., Richards, S.M., Llamas, B., Mitchell, K.J., Ho, S.Y., Kosintsev, P., Lee, M.S., 1542 Baryshnikov, G., Bollongino, R., 2016. Early cave art and ancient DNA record the origin of European 1543 bison. Nat. Commun. 7, 13158. 1544 Stampfli, H.R., 1963. Wisent, Bison bonasus (Linné) 1758, Ur, Bos primigenius Bojanus, 1827, und Hausrind 1545 Bos taurus (Linné) 1758. In: J. Boessneck, J.-P. Jéquier, H.R. Stampfli (Eds.), Burgäschisee-Süd, Teil 3: 1546 Die Tierreste, 2. Acta Bernensia, Beiträge zur prähistorischen, klassischen und jüngeren Archäologie 1547 II. Verlag Stämpfli & Cie, Bern, pp. 117–196. 59 1548 Strani, F., DeMiguel, D., Alba, D.M., Moyà-Solà, S., Bellucci, L., Sardella, R., Madurell-Malapeira, J., 2019. The 1549 effects of the “0.9 Ma event” on the Mediterranean ecosystems during the Early-Middle Pleistocene 1550 transition as revealed by dental wear patterns of fossil ungulates. Quat. Sci. Rev. 210, 80–89. 1551 Stuart, A.J. and Lister, A.M., 2001. The mammalian faunas of Pakefield/Kessingland and Corton, Suffolk, UK: 1552 evidence for a new temperate episode in the British early Middle Pleistocene. Quat. Sci. Rev. 20, 1553 1677–1692. 1554 Teilhard de Chardin, P., Piveteau, J., 1930. Les mammifères fossiles de Nihowan (Chine). Ann. Paleontol. 19, 1555 1–134. 1556 Tong, H.W., Chen, X., Zhang, B., 2016. New fossils of Bison palaeosinensis (Artiodactyla, Mammalia) from the 1557 steppe mammoth site of early Pleistocene in Nihewan Basin, China. Quat. Int. 445, 450–468. 1558 Van der Made, J., Rosell, J., Blasco, R., 2017. Faunas from Atapuerca at the Early–Middle Pleistocene limit: 1559 The ungulates from level TD8 in the context of climatic change. Quat. Int. 433, 296–346. 1560 Vershinina, A.O., Kapp, J.D., Soares, A.E.R., Heintzman, P.D., Lowson, C., Cassatt-Johnstone, M., Shidlovskiy, 1561 F.K., Kirillova, I.V., Shapiro, B., 2019. Ancient DNA analysis of a Holocene bison from the Rauchua 1562 River, Northwestern Chukotka, and the existence of a deeply divergent mitochondrial clade. Zool. J. 1563 98, 1091–1099. 1564 Verestchagin, N.K., 1959. The Mammals of the Caucasus. A History of the Fauna. Academia Nauk, Leningrad 1565 (translated from Russian; Israel Program for Scientific Translations, Jerusalem 1967). 1566 Wagner, G.A., Krbetschek, M., Degering, D., Bahain, J.J,, Shao, Q., Falguères, C., Voinchet, P., Dolo, J.M., 1567 Garcia, T., Rightmire, G.P., 2010. Radiometric dating of the type-site for Homo heidelbergensis at 1568 Mauer, Germany. Proc. Natl. Acad. Sci. USA 107, 19726–19730. 1569 Zheng, S.H., Wu, W.Y., Li, Y., Wang, G.D., 1985. Late Cenozoic mammalian faunas of Guide and Gonghe 1570 Basins, Qinghai Province. Vert. PalAs. 23, 89–134 (in Chinese). 1571 60 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 61 1588 Figure captions 1589 Fig. 1. a, Quaternary time scale showing the chronological/biochronological distribution of the studied taxa in 1590 Europe and the most important Early-Middle Pleistocene sites mentioned in the text. b, Map of Europe with 1591 the location of some of the sites mentioned in the text. c, Focus on the Black Sea area. d, Geographic location 1592 of the studied sites of Cal Guardiola and Vallparadís Estació in the town of Terrassa (Iberian Peninsula). 1593 Abbreviations: Ma, million years ago; GC, geochronology; LMA, Land Mammal Age; C, glacial/interglacial 1594 cyclicity; MS, magnetostratigraphy; CF, Cava Filo (Italy); CBF, Cromer Forest-Bed sites (United Kingdom); Ce, 1595 Cesi (Italy); CSL, Châtillon-Saint-Jean (France); Dma, Dmanisi (Georgia); Dur, Durfort (France); Isen, Isernia La 1596 Pineta (Italy); KIX, Kiputz IX (Spain); Mau, Mauer (Germany); MB, Mygdonia Basin sites (Greece); Mos, Mosbach 1597 (Germany); Olv, Olivola (Italy); Pf, Pietrafitta (Italy); Pir, Pirro (Italy); RlR, Romain-la-Roche (France); Se, Senèze 1598 (France); Si, Siréjol cave (France); Sus, Süssenborn (Germany); Tau, Taubach (Germany); Tir, Tiraspol (Ukraine); 1599 Unt, Untermassfeld (Germany); Val, Le Vallonnet (France); VCS, Vallparadís Composite Section (Spain); VM, 1600 Venta Micena (Spain); We, Westbury (United Kingdom). 1601 1602 Fig. 2. Some of the measurements taken on postcranial bones in this work: a, radioulna in anterior (1), lateral 1603 (2) proximal (3) and distal (4) views; b, ulna in lateral (1) and anterior (2) views; c, tibia in distal view; d, 1604 calcaneum in anterior (1), medial (2) and posterior (3) views; e, cubonavicular in distal (1) and anterior (2) 1605 views; f, semilunar in proximal (1) and anterior (2) views; g, astragalus in anterior (1), lateral (2) and medial (3) 1606 views; h, cuneiform in distal (1) and medial (2) views; i, pyramidal in lateral (1) and proximal (2) views; j, 1607 unciform in proximal (1) and lateral (2) views; k, scapula in lateral (1) and distal (2) views; l, phalanx in proximal 1608 (1), distal (2) and anterior (3) views; l, distal phalanx in anterior view; m, metacarpal in anterior (1), posterior 1609 (2), proximal (3) and distal (4) views; n, metatarsal in proximal view; o, humerus in lateral (1), posterior (2), 1610 anterior (3) and distal (4) views; p, thoracic vertebra in anterior (1) and lateral (2) views. Abbreviations as in 1611 Table 1. 1612 1613 62 1614 Fig. 3. Dentognathic remains of Bison schoetensacki from the Vallparadís Composite Section: a, left dp4 1615 IPS93008 in occlusal (1) and lingual (2) views; b, left p4 IPS93015 in occlusal (1) and lingual (2) views; c, right 1616 m1 IPS93027 in occlusal view; d, left m2 IPS93044 in occlusal view; e, left m3 IPS92989 in occlusal (1) and 1617 buccal (2) views; f, left P2 IPS92986 in occlusal (1) and buccal (2) views; g, left P3 IPS92998 in occlusal (1) and 1618 buccal (2) views; h, right P4 IPS13557 in occlusal (1) and buccal (2) views; i, left M1 IPS93029 in occlusal (1) 1619 and buccal (2) views; j, left M2 IPS93005 in occlusal (1) and lingual (2) views; k, right M2 IPS93032 in occlusal 1620 (1) and lingual (2) views; l, right M3 IPS20182 in occlusal (1) and lingual (2) views; m, left maxillary fragment 1621 with M2–M3 IPS92993 in occlusal view. Scale bar: 30 mm. 1622 1623 Fig. 4. Craniodental and postcranial remains of Bison schoetensacki from the Vallparadís Composite Section: 1624 a, left hemimandible with m1–m3 IPS92973 in occlusal (1) and lingual (2) views; b, juvenile mandible with dp2– 1625 dp4 IPS92968 in left lateral (1) and occlusal (2) views; c, horn core IPS92970 in dorsal (1), ventral (2), and 1626 anterior/posterior (undetermined laterality) (3) views; d, right scapula IPS107637 in distal (1) and lateral (2) 1627 views. Scale bars: 50 mm. 1628 1629 Fig. 5. Forelimb remains of Bison schoetensacki from the Vallparadís Composite Section: a, distal fragment of 1630 left humerus IPS107620 in anterior (1) and posterior (2) views; b, distal fragment of right humerus IPS50672 1631 in anterior (1) and posterior (2) views; c, left radioulna IPS107617 in anterior (1), medial (2), posterior (3), distal 1632 (4), and proximal (5) views; d, right radius IPS39893 in anterior (1), lateral (2), posterior (3), and distal (4) views 1633 and associated right ulna in anterior (5) and lateral (6) views. Scale bar: 100 mm. 1634 1635 Fig. 6. Metapodials of Bison schoetensacki from the Vallparadís Composite Section: a, left metacarpal IPS14917 1636 (male) in anterior (1), proximal (2), distal (3) and posterior (4) views; b, left metacarpal IPS107635 (male) in 1637 anterior (1), proximal (2), distal (3) and posterior (4) views; c, left metacarpal IPS92910 (female) in anterior (1), 1638 proximal (2), distal (3) and posterior (4) views; d right metatarsal IPS92934 (male) in anterior (1), proximal (2), 1639 distal (3) and posterior (4) views; e, right metatarsal IPS107634 (male) in anterior (1), proximal (2), distal (3) 63 1640 and posterior (4) views; f, right metatarsal IPS92932 (female) in anterior (1), proximal (2), distal (3) and 1641 posterior (4) views. Scale bar: 100 mm. 1642 1643 Fig. 7. Hindlimb remains of Bison schoetensacki from the Vallparadís Composite Section: a, right tibia 1644 IPS107618 in posterior (1), medial (2), lateral (3), distal (4) and anterior (5) views; b, left tibia IPS92942 in 1645 posterior (1), medial (2), lateral (3), distal (4) and anterior (5) views. Scale bar: 100 mm. 1646 1647 Fig. 8. Tarsals of Bison schoetensacki from the Vallparadís Composite Section. a, left astragalus IPS92953 (male) 1648 in anterior (1), posterior (2), and lateral (3) views; b, right astragalus IPS92952 (female) in anterior (1), posterior 1649 (2), and lateral (3) views; c, right calcaneum IPS92945 (female) in medial (1), lateral (2), posterior (3), and 1650 anterior (4) views; d, right calcaneum IPS13936 (male) in medial (1), lateral (2), posterior (3), and anterior (4) 1651 views. Scale bar: 50 mm. 1652 1653 Fig. 9. a–d, Bivariate plots of maximum length (Lmax) vs distal end width (DEW) / maximum length (Lmax) % 1654 to evaluate stoutness of metacarpals (a), metatarsals (b), radii (c), and astragali (d) in several Leptobos and 1655 Bison s.l. species. e, Log10 ratio diagrams of seven selected metacarpal variables, in which the VCS bison is 1656 compared with Leptobos etruscus, B. (Eobison) spp., and B. menneri (1), B. schoetensacki from several sites (2), 1657 and B. priscus from several sites (3). 1658 1659 Fig. 10. Bivariate plots of the firts two principal component (PC) scores resulting from principal components 1660 analyses of metacarpal (a) and metatarsal (b) variables, based on the seven variables used in this study (1), as 1661 well as in those used by Scott and Barr (2004) (2). Significant PCs with variance and coefficient of variables are 1662 in Tables S19, S20, S30, S31. Abbreviations as in Table 2. 1663 1664 Fig. 11. Reconstruction of the skeleton (a) and external appearance (b) of an adult male of Bison schoetensacki 1665 based on the remains recovered from VCS and other European sites. Scale bar: 300 mm. 1666 64 1667 Fig. 12. Comparison of metacarpals among Leptobos and Bison s.l. species. a, Leptobos etruscus male IGF 2452 1668 (1) and female IGF 2471 (2) from Olivola (Italy); b, Bison (Eobison) palaeosinensis IVPP V 22655 from 1669 Shanshenmiaozui (China; taken from Tong et al., 2016); c, Bison (Eobison) sp., male IPHES VM 83.C-3.G-85 (1) 1670 and female IPHES VM 83.C-3.G-7.11 (2) from Venta Micena (Spain; taken from Moyà-Solà, 1987); d, Bison 1671 (Eobison) georgicus males GNM D2812 (1) and GNM D2288 (2) from Dmanisi (Georgia; taken from 1672 Bukhsianidze, 2005); e, Bison (Eobison) cf. degiulii male AUTH KLT-646 (1) and female AUTH APL-373 (2) from 1673 Mygdonia Basin (Greece; taken from Kostopoulos et al., 2018); f, Bison schoetensacki males MNHN DUR-105 1674 (1), MNHN DUR-107 (2) and female MNHN DUR-106 (3) from Durfort (France); g, Bison schoetensacki males 1675 IPS14917 (1), IPS13928 (2), IPS14815 (3), IPS13547 (4), IPS92907 (5), IPS92912 (6), IPS92911 (7), IPS107635 1676 (8), IPS107636 (9) and females IPS107626 (10), IPS107626 (11) from VCS (Spain); h, Bison schoetensacki males 1677 MPRM A8 B1 3748 (1), MPRM A7 AH14 8424 (2), MPRM E7 C 242 (3), MPRM C6 359 (4), MPRM C4 CE13 209 1678 (5), MPRM C6 200 (6) and females MPRM A8 B2 326 (7): MPRM B9 BJ7 346 (8) from Le Vallonnet (France); i, 1679 Bison schoetensacki HLMD-Mau-401 from Mauer (Germany; kindly provided by Marisa Blume and Oliver 1680 Sandrock); j, Bison menneri male IQW 1980/15235 (1) and female IQW 1980/16658 (2) from Untermassfeld 1681 (Germany; taken from Sher, 1997); k, Bison schoetensacki HLMD-Mb-433 (1), HLMD-Mb-435 (2) HLMD-Mb- 1682 785 (3) from Mosbach (Germany; kindly provided by Marisa Blume and Oliver Sandrock); l, Bison priscus males 1683 NAS F-3072 (1), NAS F-1155 (2), NAS F-1275 (3) and females NAS F-1353 (4), NAS F-1339 (5), NAS F-1348 (6) 1684 from Siberia (Russia; taken from Kirilova et al., 2015); m, Bison priscus: MCM BOV 1520 M 13-34 Ro 87 from 1685 Romain-la-Roche (France; kindly provided by Lionel Cavin and Corinne Charvet). Scale bar: 100 mm. 1686 65 Table 1. Abbreviations of the measurements taken and shown in Fig. 2. Abbreviation Measurement taken ABETl Abaxial emicondyle thickness (lateral emicondyle) ABETm Abaxial emicondyle thickness (medial emicondyle) AETm Axial emicondyle thickness (medial emicondyle) AL Articular surface length AT Articular surface thickness BLm Calcaneum minimum length (medial view) BLp Calcaneum body length (posterior view) CBHmin Calcaneum body minimum height CBWmin Calcaneum body minimum width CH Centrum height CL Centrum length CW Centrum width DDT Distal diaphysis thickness DDW Distal diaphysis width DEAAW Distal epiphysis astragalus articulation width DEAT Distal end articular thickness DEATmin Distal end articular thickness (without the distal end of ulna) DEAW Distal end articular width DET Distal end thickness DETl Distal end thickness (lateral trochlear crest) DETm Distal end thickness (medial trochlear crest) DEW Distal end width DT Diaphysis thickness (midshaft) DTmin Diaphysis minimum thickness DW Diaphysis width (midshaft) DWmin Diaphysis minimum width GL Glenoid cavity length GPL Glenoid process length GW Glenoid cavity width Hmax Maximum height HST Height at sustentaculum tali level HTC Height of tuber calcanei L Length Ll Lateral length Lmax Maximum length Lmin Minimum length MFL Malleolar facet length MFW Malleolar facet width NCFL Cubonavicular articular facet length NSL Neural spine length NSW Neural spine width OL Olecranon length OTmin Olecranon minimum thickness OW Olecranon fossa width PEAT Proximal end articular thickness PEATl Lateral facet articular thickness (proximal end) PEAW Proximal end articular width PET Proximal end thickness PEW Proximal end width PFOl Metatarsal proximal articular facet oblique diameter (lateral facet) Con formato: Español (España) PFOm Metatarsal proximal articular facet oblique diameter (medial facet) Con formato: Español (España) PFWl Metacarpal proximal articular facet width (lateral facet) PFWm Metacarpal proximal articular facet width (medial facet) SCW Spinal canal width SOL Oblique length of the sole STT Sustentaculum tali thickness T Thickness TCH Trochlea crest height THl Lateral trochlear height THm Medial trochlear height Tl Lateral thickness TTl Trochlea thickness (lateral epicondyle) TTm Trochlea thickness (medial epicondyle) TTW Intertrochlear width TWl Trochlear lateral articulation width TWm Trochlear medial articulation width Wmax Maximum width Wmin Minimum width WTC Width of tuber calcanei Table 2. Shape variables used for the principal component analysis. Abbreviations: DETl, distal epiphysis thickness (lateral trochlear crest); DETm, distal epiphysis thickness (medial trochlear crest); DEW, distal epiphysis width; DT, diaphysis thickness (midshaft); DW, diaphysis width (midshaft); Lmax, maximum length; MGSV, metapodial global size variable (for the equation see Section 3); ms, Mosimann shape variable; PET, proximal epiphysis thickness; PEW, proximal epiphysis width; re, relative. Selected Mosimann shape Relative dimension variables, Fig. 10a Fig. 10b variables after Scott and Barr (2014) msLmax = Log(Lmax/GM) 1a reLmax = Log(Lmax/MGSV) 1b msPET = Log(PET/GM) 2a reDW= Log(DW/MGSV) 2b msPEW = Log(PEW/GM) 3a reDT = Log(DT/MGSV) 3b msDT = Log(DT/GM) 4a reDETl = Log(DETl/MGSV) 4b msDW = Log(DW/GM) 5a reDETm = Log(DETm/MGSV) 5b msDETm = Log(DETm/GM) 6a reDEW = Log(DEW/MGSV) 6b msDEW = Log(DEW/GM) 7a rePEW = Log(PEW/MGSV) 7b rePET = Log(PET/MGSV) 8b Table 3. Measurements (mm) of the teeth of Bison schoetensacki from Vallaradís Estació (EVT) and Cal Guardiola (CGR) layers. Abbreviations: L, length; W, width. ID Specimen Layer Tooth L W IPS92968 EVT12 dp2 11.4 7.1 IPS92968 EVT12 dp3 19.6 10.3 IPS14965 CGRD7 dp4 29.4 16.0 IPS92968 EVT12 dp4 32.5 14.1 IPS92997 EVT7 dp4 25.0 12.8 IPS93008 EVT7 dp4 31.0 14.3 IPS92977 EVT7 dp4 20.3 13.2 IPS93015 EVT7 p4 20.5 12.4 IPS93045 EVT7 p4 17.4 11.7 IPS93056 EVT7 p4 19.4 11.3 IPS93027 EVT7 m1 24.8 15.7 IPS93042 EVT7 m1 24.6 17.0 IPS92973 EVT7 m1 25.0 16.5 IPS92977 EVT7 m1 23.9 15.1 IPS93009 EVT7 m2 25.3 16.1 IPS93014 EVT7 m2 24.0 17.2 IPS93019 EVT7 m2 26.2 18.4 IPS93030 EVT7 m2 24.8 17.1 IPS93044 EVT7 m2 27.7 18.7 IPS92973 EVT7 m2 24.7 18.6 IPS92977 EVT7 m2 28.5 17.6 IPS92990 EVT7 m2 26.8 24.5 IPS93043 EVT7 m3 41.9 17.6 IPS92977 EVT7 m3 38.2 16.3 IPS92973 EVT7 m3 41.7 18.4 IPS92987 EVT7 m3 46.5 18.4 IPS92989 EVT7 m3 40.3 16.7 IPS92986 EVT7 P2 16.3 13.0 IPS92996 EVT10 P3 19.4 16.2 IPS92998 EVT7 P3 13.8 19.9 IPS13557 CGRD7 P4 14.7 19.2 IPS92995 EVT10 P4 15.6 21.8 IPS93013 EVT3 P4 19.5 21.4 IPS93020 EVT7 P4 15.8 21.7 IPS92992 EVT10 M1 25.2 25.6 IPS93016 EVT6 M1 27.2 24.2 IPS93017 EVT6 M1 24.7 25.9 IPS93029 EVT7 M1 21.9 26.8 IPS93036 EVT7 M1 21.0 23.8 IPS93047 EVT7 M1 20.6 23.4 IPS92988 EVT7 M2 25.8 25.3 IPS92991 EVT10 M2 28.1 28.8 IPS92994 EVT10 M2 26.9 27.6 IPS93005 EVT7 M2 27.6 28.2 IPS93003 EVT7 M2 24.9 25.1 IPS93022 EVT7 M2 29.3 21.2 IPS93032 EVT7 M2 28.3 27.3 IPS93034 EVT7 M2 23.3 23.7 IPS93038 EVT7 M2 25.7 26.1 IPS20182 CGRD7 M3 30.6 26.0 IPS92993 EVT10 M3 30.4 27.6 IPS92994 EVT10 M3 31.0 27.5 IPS93007 EVT7 M3 27.3 24.7 IPS93011 EVT7 M3 28.0 24.3 IPS93018 EVT7 M3 27.8 26.0 IPS93021 EVT7 M3 29.1 25.7 IPS93023 EVT7 M3 23.5 25.0 IPS93028 EVT7 M3 32.3 24.5 IPS93033 EVT7 M3 27.0 26.5 Table 4. Univariate and multivariate analyses of variance of M2 and M3, metacarpals, and metatarsals from the two main VCS chronologies: 1.07–0.99 Ma (EVT10-12) and 0.86–0.78 (CGRD7-EVT7) (only complete metapodials included). Values with p<0.05 are in bold. Abbreviations: DET, distal epiphysis thickness (maximum); DEW, distal epiphysis width; DT, diaphysis thickness (midshaft); DW, diaphysis width (midshaft); F, F statistic; L, length; Lmax, maximum length; p, significance; PET, proximal epiphysis thickness; PEW, proximal epiphysis width; W, width. Molars ANOVA Variable F p M2: W/L% 0.5891 0.4719 M3: W/L% 0.05129 0.8273 Metacarpals ANOVA Variable F p Lmax 18.55 0.00259 PEW 0.03753 0.849 PET 0.1259 0.728 DW 2.737 0.1263 DT 1.765 0.2109 DEW 2.107 0.19 DET 1.017 0.3467 DEW/Lmax% 0.04082 0.8456 Metacarpals MANOVA F p 4.009 0.1412 Metatarsals ANOVA Variable F p Lmax 0.06263 0.8107 PEW 1.466 0.2513 PET 3.524 0.08726 DW 0.004166 0.9497 DT 0.898 0.3637 DEW 5.131 0.0641 DET 3.995 0.1021 DEW/Lmax% 7.579 0.03316 Metatarsals MANOVA F p 2.049 0.3597 Table 5. Measurements (mm) and descriptive statistics of the metacarpals of Bison schoetensacki from Vallparadís Estació (EVT) and Cal Guardiola (CGR) layers. Abbreviations: ABETl, abaxial emicondyle thickness (lateral emicondyle); ABETm, abaxial emicondyle thickness (medial emicondyle); AETl, axial emicondyle thickness (lateral emicondyle); AETm, axial emicondyle thickness (medial emicondyle); DETl, distal epiphysis thickness (lateral trochlear crest); DETm, distal epiphysis thickness (medial trochlear crest); DEW, distal epiphysis width; DDW, distal diaphysis width (above the epiphysis); DT, diaphysis thickness (midshaft); DW, diaphysis width (midshaft); F, female; L, left; Lf, functional length; Lmax, maximum length; M, male; PET, proximal epiphysis thickness; PEW, proximal epiphysis width; PFWl, proximal articular facet width (lateral facet); PFWm, proximal articular facet width (medial facet); R, right. ID Specimen Layer Side Sex Lmax Lf PEW PET DW DT DDW DEW DETm DETl PFWl PFWm ABETm ABETl AETm AETl IPS796 CGRD7 R M 75.0 50.4 34.6 43.5 29.7 IPS13547 CGRD7 R M 254.7 243.9 81.3 47.6 52.9 33.6 78.6 82.4 43.7 43.8 43.3 35.9 32.8 30.1 39.8 38.5 IPS13911 CGRD2 L M 53.9 36.6 IPS13928 CGRD7 R M 250.2 239.4 79.3 47.0 49.55 30.0 72.6 79.3 45.9 43.6 45.9 32.6 36.7 33.8 39.9 39.4 IPS14102 CGRD7 R F 70.6 42.1 39.0 27.3 IPS14702 CGRD7 L F? 73.5 42.2 40.4 29.5 IPS14815 CGRD7 L M 242.5 77.0 46.3 51.8 34.6 37.5 44.4 28.0 IPS14917 CGRD2 L M 262.7 251 83.9 51.5 52 33.4 78.5 83.4 45.4 41.4 49.7 31.2 33.1 29.9 40.7 40.3 IPS14985 CGRD7 R M 83.0 50.2 46.1 31.4 IPS15002 CGRD7 R M 79.0 46.9 46.5 46.6 44.2 30.8 IPS92905 CGRD7 R M 46.2 IPS92907 EVT10 L M 230.3 220.9 83.2 47.0 50.2 33.4 75.2 77.1 41.7 41.7 46.2 28.7 31.6 30.5 37.5 37.0 IPS92909 EVT12 R M 85.1 49.0 48.7 33.4 74.0 45.2 31.2 IPS92910 EVT12 L F 230.5 223.6 67.7 41.3 38.42 29.9 62.8 66.7 36.5 36.5 37 24.1 25.7 IPS92911 EVT7 R M 246 232.4 78.2 45.3 50.3 33 74.4 79.1 44.3 34.1 29.6 28.4 IPS92912 EVT7 R M 239.5 230.4 77.3 45.5 49.3 32.5 74.6 77.0 41.2 41.6 43.1 32.1 32.9 30.7 39.6 38.0 IPS92913 EVT12 L M 75.8 45.6 42.9 31.5 IPS107626 EVT12 L F 228.9 69.5 44.5 36.3 27.9 66.8 35.3 35.0 29.2 26.3 32.3 32.8 IPS107635 EVT12 L M 238.0 224.01 78.3 45.8 50.4 32.9 75.8 80.1 42.5 41.9 42.1 34.2 32.5 29.9 40.8 39.9 IPS107636 EVT12 R M 235.7 224.7 79.0 45.0 50.7 33.5 75.6 80.2 42.4 42.6 42.2 31.9 32.2 30.6 40..5 39.0 Mean 241.7 232.3 77.6 46.1 48.8 33.7 74.2 77.2 41.2 40.9 38.6 39.2 31.6 28.8 37.9 38.4 Minimum 228.9 220.9 67.7 41.3 36.3 27.9 62.8 66.7 35.3 35 32.7 29 25 20.8 29.5 32.8 Maximum 262.7 251 85.1 51.5 53.9 46.6 78.6 83.4 45.9 43.8 42.6 43.1 36.7 33.8 40.8 40.3 SD 10.8 10.4 5.0 2.7 5.0 4.14 4.4 5.9 3.7 3.1 4.2 5.8 3.1 3.5 4.1 2.5 N 11 9 18 17 15 15 10 10 10 9 7 5 10 11 9 8 Table 6. Measurements and descriptive statistics of the metatarsals of Bison schoetensacki from Vallparadís Estació (EVT) and Cal Guardiola (CGR) layers. Abbreviations: ABETl, abaxial emicondyle thickness (lateral emicondyle); ABETm, abaxial emicondyle thickness (medial emicondyle); AETl, axial emicondyle thickness (lateral emicondyle); AETm, axial emicondyle thickness (medial emicondyle); DETl, distal epiphysis thickness (lateral trochlear crest); DETm, distal epiphysis thickness (medial trochlear crest); DEW, distal epiphysis width; DDW, distal diaphysis width (above the epiphysis); DT, diaphysis thickness (midshaft); DW, diaphysis width (midshaft); F, female; L, left; Lf, functional length; Lmax, maximum length; M, male; PET, proximal epiphysis thickness; PEW, proximal epiphysis width; PFOl, proximal articular facet oblique diameter (medial facet); PFOm, proximal articular facet oblique diameter (lateral facet); R, right. ID Specimen Layer Side Sex Lmax Lf PEW PET DW DT DDW DEW DETm DETl PFOm PFOl ABETm ABETl AETm AETl IPS92906 EVT7 L F 54.5 56.5 37.5 42.0 46.5 IPS92930 CGRD7 R F 280.5 265.6 58.1 53.2 36.6 34.8 59.0 65.0 37.4 37.6 IPS92931 EVT7 L F 269.7 253.0 57.0 56.8 39.4 40.3 62.5 63.8 34.5 35.1 39.3 41.1 25.5 IPS92932 EVT12 R F 275.2 258.6 57.3 57.5 34.0 37.9 65.3 65.0 41.4 38.7 40.0 36.9 26.5 25.8 32.2 33.7 IPS92933 EVT7 L M? 62.6 61.6 42.8 44.3 33.5 52.5 30.7 28.5 35.8 37.3 IPS92934 EVT7 R M? 296.2 275.3 63.8 59.2 42.2 43.0 72.1 69.9 41.9 41 45.5 43.0 IPS92935 EVT7 R M? 63.7 62.0 42.0 40.5 44.5 42.8 31.5 28.9 35.0 40.2 IPS92936 EVT7 L F 40.3 39.5 IPS92937 EVT12 R M 300.2 287.4 68.3 66.0 45.5 45 76.5 78 43.6 41.1 46.0 53.3 IPS92938 EVT10 L F 61.0 60.0 41.3 42.0 34.4 32.1 39.1 40.4 IPS92939 EVT7 R F 59.0 55.5 39.9 40.5 37.5 40.0 IPS107634 EVT12 R M 279.0 258.3 60.4 60.0 39.5 42.4 70.7 73.6 42.5 41.9 44.0 40.2 IPS114552 EVT7 L F 272.3 268.3 53.0 52.7 36.9 34.2 60.9 61.3 39.5 37.0 37.0 39.0 31.9 30.2 38.7 39.6 IPS114553 EVT7 R F 293.8 287.4 57.0 55.9 37.7 37.7 63.3 61.8 38.9 44 36.4 30.2 27.7 33.0 32.0 Mean 283.4 269.2 59.7 58.2 39.6 40.2 65.8 67.3 40.3 39.1 41.3 42.1 30.1 28.8 35.6 37.2 Minimum 269.7 253.0 53.0 52.7 34.0 34.2 55.0 61.3 34.5 35.1 33.5 36.4 25.5 25.8 32.2 32.0 Maximum 300.2 287.4 68.3 66.0 45.5 45.0 76.5 78.0 43.6 41.9 46.5 53.3 34.4 32.1 39.1 40.4 SD 11.7 13.1 4.2 3.7 3.1 3.3 6.9 6 3.1 2.7 4.2 5.5 3.1 2.2 2.9 3.6 N 8 8 13 13 13 13 8 8 7 6 13 12 7 6 6 6