This is the accepted version of the article: Mecozzi, Beniamino. «The tale of a short-tailed cat : new outstanding Late Pleistocene fossils of Lynx pardinus from southern Italy». Quaternary Science Reviews, Available online 15 March (2021), art. 106840. This version is available at https://ddd.uab.cat/record/237630 under the terms of the license 1 The tale of a short-tailed cat: new outstanding Late Pleistocene 2 fossils of Lynx pardinus from southern Italy 3 4 Beniamino Mecozzi1, Raffaele Sardella1 *, Alberto Boscaini2, Marco Cherin3, Loïc 5 Costeur4, Joan Madurell-Malapeira5,6, Marco Pavia7, Antonio Profico8, Dawid A. 6 Iurino1,3 * 7 PaleoFactory, Dipartimento di Scienze della Terra, Sapienza Università di Roma, Piazzale A. Moro 5, 1 8 00185 Roma, Italy. 9 Instituto de Ecología, Genética y Evolución de Buenos Aires (IEGEBA – CONICET), DEGE, Facultad de 2 10 Ciencias Exactas y Naturales, Universidad de Buenos Aires. Int. Guiraldes 2160, Buenos Aires, Argentina. 11 Dipartimento di Fisica e Geologia, Università degli Studi di Perugia, Via A. Pascoli, 06123 Perugia, Italy. 3 12 Naturhistorisches Museum Basel, Augustinergasse 2, 4001 Basel, Switzerland. 4 13 Institut Català de Paleontologia Miguel Crusafont, Universitat Autònoma de Barcelona, Edifici ICTA-ICP, c/ 5 14 Columnes s/n, Campus de la UAB, 08193 Cerdanyola del Vallès, Barcelona, Spain. 15 6 Department of Geology, Universitat Autònoma de Barcelona, Facultat de Ciències, Edifici C, Campus de la 16 UAB, 08193 Cerdanyola del Vallès, Barcelona, Spain 17 Museo di Geologia e Paleontologia, Dipartimento di Scienze della Terra, Università degli Studi di Torino, 7 18 Via Valperga Caluso 35, 10125 Torino, Italy. 19 PalaeoHub, University of York, Wentworth Way Heslington, YO10 5NG York, United Kingdom. 8 20 *Corresponding authors: dawid.iurino@uniroma1.it and raffaele.sardella@uniroma1.it 21 22 Abstract 23 The pardel lynx Lynx pardinus is today restricted to small populations living in southern 24 Iberian Peninsula. However, this endangered species was widely spread throughout Iberia 1 25 until historical times and is currently the subject of intense conservation programs. 26 Paleontological data suggest that its past geographical range was much wider, including 27 also southern France and northern Italy. Here, we report exceptionally preserved fossil 28 remains of L. pardinus from the Late Pleistocene (about 40’000 years) of Ingarano (Italy), 29 which represent the largest sample of fossil lynx currently known in Europe. This new 30 evidence allows (1) to revise the taxonomy of European fossil lynxes, (2) to extend far 31 southeast the paleobiogeographical distribution of L. pardinus, and (3) to offer new 32 insights on the evolutionary history (e.g., relationships with other extinct and extant lynx 33 species) and paleobiology (e.g., intraspecific variation, body mass) of this iconic European 34 felid. 35 36 Keywords: Europe, Pleistocene, Carnivora, Felidae, Lynx, Evolution, Taxonomy, 37 Paleobiogeography, Paleoecology 38 39 1. Introduction 40 With a geographical distribution limited to a few restricted areas of southwestern 41 Iberian Peninsula (Macdonald et al., 2010), the pardel (or Iberian) lynx Lynx pardinus is 42 considered one of the most threatened living felids by the International Union for 43 Conservation of Nature (IUCN; Simón, 2012; Simón et al., 2012). Over the last decades 44 many efforts have been made to better characterize the ecological and biological profile of 45 this rare felid, mainly with the aim of developing conservation projects to increase 46 population abundance and to expand its distribution. According to IUCN data, not more 47 than 160 mature individuals exist today in southern Iberia (Rodríguez and Calzada, 2 48 2015). Currently, the pardel lynx is one of the best-known living felids, but despite this, its 49 evolutionary history is still strongly debated among the specialists. 50 Since the mid-1970s, different scenarios on the evolution of genus Lynx have been 51 proposed. Ficcarelli and Torre (1977) considered the Plio-Pleistocene Lynx issiodorensis 52 as the ancestor of L. pardinus. This hypothesis was shared by Kurtén (1978) and partially 53 revised by Werdelin (1981). The earliest records of L. issiodorensis date back to the 54 Ruscinian European Land Mammal Age (ELMA), roughly corresponding to the early 55 Pliocene (ca. 4 Ma). These records include scanty remains from Cuevas del Almanzora 56 (Spain; Montoya et al., 2001a), Layna (Spain; “Caracal depereti” in Morales et al., 2003), 57 Serrat d’en Vacquer (France; “Caracal brevirostris” in Depéret, 1890), and Çalta (Turkey; 58 Ginsburg, 1998). Lynx issiodorensis then became a common element in terrestrial 59 mammal assemblages during the Villafranchian ELMA, encompassing the late Pliocene 60 and most of the Early Pleistocene (ca. 3.3-1.2 Ma). Werdelin (1981) divided L. 61 issiodorensis into two chronologically distinct subspecies, namely L. issiodorensis 62 issiodorensis (early-middle Villafranchian; ca. 3.3–2.1 Ma) forms and L. issiodorensis 63 valdarnensis (late Villafranchian; ca. 2.1–1.2 Ma). The first includes, among others, the 64 nearly complete skeleton from Les Etouaires (France; Kurtén, 1978) and Saint Vallier 65 (France; Viret, 1954; Argant, 2004); the second includes several well-preserved fossils 66 from Italian sites such as Olivola, Upper Valdarno, and Pantalla (Fabrini, 1896; Del 67 Campana, 1925; Cherin et al., 2013). The last occurrence of L. issiodorensis is from the 68 Epivillafranchian site of Untermassfeld (Germany; ca. 1.0 Ma; Hemmer, 2001). According 69 to Werdelin (1981), L. issiodorensis has a pivotal role in lynx evolution, as it probably gave 70 rise to both the lineages of L. pardinus in Europe and Lynx lynx in Asia in the Early 71 Pleistocene. This is corroborated by molecular evidence, according to which the 72 divergence between L. pardinus and L. lynx occurred between ca. 2.2 Ma (Bininda- 73 Emonds et al., 1999) and 1.6 Ma (Johnson et al., 2004). The evolution of L. lynx from L. 3 74 issiodorensis in Asia probably involved L. issiodorensis shansius as an intermediate form 75 (Werdelin, 1981; Kurtén and Werdelin, 1984). The Eurasian lynx then reached Europe at 76 the beginning of the Late Pleistocene (Eemian, MIS 5e; Werdelin, 1981) and became a 77 common element of the carnivore guild during the whole Last Glacial Period, especially in 78 central and northern Europe (Kahlke, 1999; Sommer and Benecke, 2006). On the other 79 side, the long and gradual evolution from L. issiodorensis to L. pardinus involved the 80 aforementioned intermediate subspecies L. i. issiodorensis and L. i. valdarnensis, followed 81 by the Middle Pleistocene ‘cave lynx’ L. pardinus spelaeus (Werdelin, 1981). This 82 anagenetic evolutionary trend characterized by a progressive reduction in body size and 83 relative lengthening of the M1 is accepted by several authors (Kurtén and Granqvist, 84 1987; García and Arsuaga, 1998; Cherin et al., 2013), although the taxonomic status of 85 the ‘cave lynx’ is still debated. This taxon was established on fossil remains from Grotta 86 del Principe (Ventimiglia, northwestern Italy) by Boule (1919) as “Felis (Lynx) pardinus 87 race spelaea” mainly based on the strong craniodental similarity with the extant L. 88 pardinus. However, the fossil form showed larger body size than the modern one. 89 The ‘cave lynx’ is reported from several Middle-Late Pleistocene sites of 90 southwestern Europe, especially in southern France and northern Italy. In agreement with 91 Werdelin’s (1981) interpretation, many authors considered this taxon as a subspecies and 92 probable ancestor of the modern L. pardinus, i.e., L. pardinus spelaeus (Kurtén and 93 Granqvist, 1987; García-Perea, 1997; Palombo et al., 2008; Garrido and Arribas, 2008; 94 Rodríguez-Hidalgo et al., 2020; Fosse et al., 2020). Conversely, other authors preferred 95 to classify the ‘cave lynx’ as a distinct species, i.e., Lynx spelaeus (Bonifay, 1971; 96 Ficcarelli and Torre, 1977; Arribas, 1994; Rustioni et al., 1995; Capasso Barbato et 97 al., 1998; Testu, 2006; Ghezzo et al., 2015). 4 98 The debate on the taxonomy and distribution of Quaternary lynxes in southwestern 99 Europe has been enriched by recently described specimens. Boscaini et al. (2015) 100 reported on a fragmentary cranium from the site of Avenc Marcel (Spain; 1.7–1.6 Ma) as 101 the earliest fossil evidence of L. pardinus and they suggested an Early Pleistocene origin 102 of this species from L. issiodorensis. Boscaini et al. (2016) analyzed the extensive 103 collection of fossil lynx from the latest Early Pleistocene of Vallparadís Estació and Cueva 104 Victoria (Iberian Peninsula) and the authors proposed to consider L. spelaeus as a junior 105 subjective synonym of L. pardinus, since the two taxa do not show significant skeletal 106 morphological differences other than the body size. The poor reliability of taxonomic 107 identifications based on dimensional criteria, especially for Late Pleistocene material, was 108 confirmed by the recent molecular study by Rodríguez-Varela et al. (2015). These 109 authors analyzed the mtDNA of fossils from Arene Candide (Savona, northwestern Italy), 110 which were previously attributed to L. lynx or L. cf. spelaeus based on their overall large 111 dimensions (Cassoli and Tagliacozzo, 1994a). Molecular results allowed to refer some of 112 the specimens to L. lynx and some to L. pardinus. Interestingly, some of the remains 113 genetically attributed to the pardel lynx show dimensions relatively larger than the average 114 values recorded for the ‘cave lynx’ (Rodríguez-Varela et al., 2015). The identification of L. 115 pardinus in the latest Pleistocene (ca. 25–17 ka) deposits of Arene Candide represents the 116 first direct evidence of the occurrence of this species in Italy, suggesting that its ancient 117 distribution should exceed the Iberian Peninsula. However, taking into account the 118 absence of diagnostic morphological differences between the pardel lynx and ‘cave lynx’, it 119 has been proposed that Middle-Late Pleistocene samples from Mediterranean Europe 120 originally referred to Lynx sp., L. spelaeus, or L. pardinus spelaeus, can be instead 121 attributed to L. pardinus (Boscaini et al., 2016). For these reasons, unless otherwise 122 stated, here we prefer using the terminology L. pardinus sensu lato (s.l.) to indicate extant 123 and fossil pardel lynx, that is, including ‘cave lynx’. 5 124 In this paper we report on the lynx remains from the Late Pleistocene site of Ingarano 125 (Foggia, southeastern Italy), which represent the largest sample of fossil lynx in Europe, 126 consisting of 415 remains including two almost complete crania. These remains were 127 previously attributed to L. lynx mainly due to their large dimensions (Capasso Barbato et 128 al.,1992; Petronio et al., 1996; Petronio and Sardella, 1998). More recently, Iurino et al. 129 (2015) carried out a comparative study on a natural brain endocast from Ingarano, which 130 was referred to Lynx sp. since the endocranial features did not show reliable interspecific 131 differences. Here, we analyze the morphological and biometric features of the lynx 132 craniodental material from Ingarano, we discuss its taxonomic attribution, as well as its 133 paleoecological and paleobiogeographical implications. 134 135 2. Geological and paleontological framework 136 The Late Pleistocene fossiliferous site of Ingarano is located in the northwestern area 137 of the Gargano promontory (270 m s.l.m.) close to the Gargano railway, nearby Apricena 138 town (Foggia, southeastern Italy; Figure 1). The Ingarano deposit is a karstic filling 139 succession within the Jurassic-Cretaceous limestone of the “Calcare di Sannicandro” 140 Formation. The stratigraphic succession was exposed by quarrying activities and is 141 composed by five different layers, A–E from bottom to top (Bedetti and Pavia, 2007). 142 Geochemical analysis of the phosphatic material from layer B with the 329Th/234U method 143 provided dating of 40 ± 2 ka in agreement with the putative age of some Mousterian 144 artifacts found between layers D-E (Petronio et al., 1996). Paleoecological and 145 taphonomic studies confirmed that the Ingarano succession, at least from layers B to E 146 (e.g., those bearing vertebrate remains), was deposited in a short time-span during MIS 3 147 (60–28 ka; Bedetti and Pavia, 2007). 6 148 In the last decades, several authors studied the abundant vertebrate fauna from 149 Ingarano, which includes 41 taxa: Amphibians - Rana sp.; Reptiles - Lacerta sp.; Birds - 150 Circus aeroginosus, Buteo rufinus, Aquila chrysaëtos, Falco cherrug, Falco columbarius, 151 Alectoris graeca, Perdix perdix, Columba livia, Otus scops, Nyctea scandiaca, Athene 152 noctua, Pyrrhocorax graculus, Pyrrhocorax sp., Corvus corone, C. corone vel C. 153 frugilegus, Corvus corax; Mammals - Erinaceus europaeus, Myotis blythi, Oryctolagus 154 cuniculus, Lepus europaeus, Arvicolidae indet., Microtus sp., Microtus ex gr. 155 arvalis/agrestis, Terricola savii, Apodemus sylvaticus, Eliomys quercinus, Canis lupus, 156 Vulpes vulpes, Ursus arctos, Mustela nivalis, Martes sp., Meles meles, Gulo gulo, Crocuta 157 crocuta, Felis sp., Lynx lynx (revised in this paper), Panthera pardus, Equus hydruntinus, 158 Cervus elaphus, Dama dama, Capreolus capreolus, Rupicapra sp., Bos primigenius 159 (Capasso Barbato et al., 1992; Petronio et al., 1996; Petronio and Sardella, 1998; 160 Curcio et al., 2005; Bedetti and Pavia, 2007; Iurino, 2014; Iurino et al., 2015). An 161 additional layer was identified by Petronio et al. (1996), including fossil remains attributed 162 to Panthera spelaea, Hippopotamus amphibius, Coelodonta antiquitatis, Stephanorhinus 163 hemitoechus, and Palaeololoxodon antiquus. This layer was composed by sandy clay in 164 doubtful stratigraphic correlation with the infilling succession (Petronio and Sardella, 165 1998). 166 167 Figure 1 168 169 3. Materials and methods 170 The fossil material studied in this work were discovered in the 1990s by the research 171 team of the Departments of Earth Sciences of Sapienza University of Rome and University 172 of Turin during several geo-paleontological surveys at Ingarano. The collection counts 415 7 173 lynx remains represented by 346 postcranial and 68 craniodental remains housed in the 174 PaleoFactory Laboratory of the Department of Earth Sciences of Sapienza University of 175 Rome, except the skull MGPT-PU 135415, which is stored in the Museo di Geologia e 176 Paleontologia of the University of Turin. In this paper, we focused on craniodental remains 177 listed in Supplementary note 1. All the specimens from Ingarano are marked with 178 different alphanumeric codes: FG indicates the remains coming from the first geo- 179 paleontological surveys (1989-90); specimens without field labels/numbers are indicated 180 with ING-, INGND-, and M- and come from successive geo-paleontological surveys. 181 The crania from Ingarano (MGPT-PU 135415, ING75), the cranium of L. issiodorensis 182 from Pantalla (SABAP_UMB 337653) and a skull of extant L. lynx (MC 85) from the Museo 183 Civico di Zoologia of Rome were acquired using a Philips Brilliance CT 64-channel 184 scanner at M.G. Vannini Hospital (Rome). The slice thickness is 0.55 mm and interslice 185 space is 0.27 mm. A cranium of extant L. pardinus (MNB 8865) and that of L. issiodorensis 186 from Olivola (MNB Ol 1) were scanned at the Biomaterials Science Center of the 187 University of Basel using the advanced microCT-system nanotom® m (phoenix x-ray, 188 GE Sensing & Inspection Technologies GmbH, Wunstorf, Germany); slice thickness and 189 interslice space are both 66 µm. CT image processing was performed using Mimics 20.0, 190 while the digital restoration process of the missing portions of the skulls was made with 191 ZBrush 4R6. All the missing parts of the specimens were cloned and mirrored from those 192 preserved, and their alignment and positioning were performed following the sagittal plane 193 of the skull (Cherin et al., 2018; Iurino et al., 2020). Finally, the reconstructed portions 194 were highlighted using different colors (Figures 3 and 10). 195 Measurements were taken with a digital caliper to the nearest 0.1 mm, following Von 196 den Driesch (1976) and Testu (2006). To build a comparative dataset, we used the 197 available literature on late Pliocene-Pleistocene lynxes from several European sites 198 (Supplementary table 9). We also collected morphological and biometric data on extant 8 199 specimens of L. pardinus and L. lynx, as well as on fossil remains of L. pardinus from Cau 200 d’en Borràs, Cau del Duc, Cova Toll, Cova Toixoneres, Cova de l’Arbreda, El Escarche, El 201 Muscle, Turó del Moro (Spain), Monte Tignoso (Italy), and L’Escale (France); L. 202 issiodorensis from Les Etouaires, Serrat d’en Vaquer, Saint Vallier (France),Olivola, 203 Pantalla, Figline, Matassino, Poggio Rosso, and other Upper Valdarno localities (Italy), La 204 Puebla de Valverde and Almanzora (Spain); Lynx sp. from Monte Argentario and Pirro 205 Nord (Italy). 206 Biometric comparisons and statistical analysis were performed considering the length 207 (L) and width (W) of the upper (P4) and lower (M1) carnassial. First, we used the Shapiro- 208 Wilk test to check if the data were normally distributed. Given the negative results (P4L, p- 209 value < 0.01; M1L, p-value < 0.01), non-parametric tests were preferred to evaluate 210 differences in size between the Ingarano remains and the comparative sample. We carried 211 out a Kruskall-Wallis test on each subsample for each variable, testing the null hypothesis 212 that the samples originate from the same distribution. The alternative hypothesis is that 213 they differ in at least one distribution. All statistical analyses were performed using the 214 software R (R Core Team, 2019). 215 Body mass was estimated using the method by Van Valkenburgh (1990), who 216 provided a series of predictive equations for several groups of carnivorans, based on 217 different linear measurements. Among these, we tested the two equations for which we 218 had more data available, namely the one based on the condylobasal length (CBL) and the 219 one based on the length of the lower carnassial (M1L), in their “Felidae only” version. By 220 applying the two formulas to a control sample for which we had both linear measurements 221 and body mass available (i.e., extant L. pardinus from Doñana, Spain; Beltrán and 222 Delibes, 1993), we verified that the CBL-based equation provided estimates very close to 223 the actual average body mass (estimated body mass - females: 10.4 kg, males: 12.3 kg; 224 actual body mass - females: 9.3 kg, males: 12.8 kg). The same equation was also tested 9 225 on a sample of extant L. lynx from Croatia (Gomerčić et al., 2010) and we obtained a 226 similar correspondence between estimated and actual values (estimated body mass - 227 females: 17.0 kg, males: 21.0 kg; actual body mass - females: 18.4 kg, males: 21.9 kg). 228 This confirmed Van Valkenburgh’s (1990) claim that CBL is the most reliable proxy for 229 estimating carnivoran body mass. Conversely, estimates based on M1L resulted in body 230 mass values 43 to 70% higher than actual average body mass for males and females, 231 respectively. 232 The extraction of ancient DNA from Ingarano lynx specimens was attempted at the 233 Archaeological Research Laboratory, University of Stockholm (Sweden), but it gave no 234 results (Supplementary note 4). 235 236 4. Results 237 4.1. Description of the cranium and upper teeth 238 Ten cranial remains were recovered from Ingarano (Supplementary note 1), among 239 which the specimens MGPT-PU 135415 and ING75 are the most complete (Figures 2-3). 240 Measurements are reported in Table 1 (see also Supplementary note 2-3, 241 Supplementary figure 1 and Supplementary tables 1-3). Complete fusion of cranial 242 sutures (where not obliterated by diagenesis) and eruption of permament dentition indicate 243 that all remains belong to adult individuals. 244 245 Figure 2 246 247 In dorsal view, the studied specimens display laterally expanded frontals with well- 248 developed triangular-shaped zygomatic processes, triangular nasals, and wide and 10 249 semicircular zygomatic arches. The marked temporal ridges are lyre-shaped and converge 250 posteriorly in a short sagittal crest, which in turn merges to a well-developed nuchal crest 251 (Figures 2-3). In lateral view, the cranium is anteroposteriorly elongated with a rounded 252 and domed dorsal profile. The orbits are large and rounded. The incisive line of the 253 premaxillae is projected anteriorly, displaying a well-developed notch at the level of the I3- 254 C1 diastema. In ventral view, the outline of the tympanic bullae appears elliptical and 255 anteroposteriorly elongated. At the posteromedial corner of the tympanic bullae, the 256 jugular and hypoglossal foramina are located into the same depression. In posterior view, 257 the oval-shaped foramen magnum is slightly compressed dorsoventrally. 258 259 Figure 3 260 261 The upper teeth from Ingarano show a rather low overall morphological variation. The 262 upper incisors are preserved in both the complete crania MGPT-PU 135415 and ING75; 263 among them, the I3 is the largest in size and it shows a marked cingulum. The C1 is 264 buccolingually compressed and smooth on the lingual side, whereas two longitudinal 265 grooves are present on the buccal surface (Figure 2). The P3 is buccolingually 266 compressed, with a high and distally projected paracone, a small distal accessory cusp, 267 and a marked distal cingulum. Four specimens (FG130, ING75, ING82, and ING86) 268 possess an incipient mesiolingual accessory cusp, which is absent in the other remains. 269 The P4 shows a small protocone located mesiolingually. The parastyle is buccolingually 270 compressed and mesiodistally shorter than the paracone and metastyle. The latter are well 271 developed and separated by a deep grove, especially in the buccal side. Four remains, 272 (FG101, FG130, ING82, and ING8) show a marked groove connecting the paracone and 273 protocone. Only two remains (FG103 and ING76) exhibit a weak ectoparastyle. The three 11 274 preserved M1 (FG101, FG102, and ING75) are buccolingually expanded with two very 275 small lingual and buccal cusps. 276 277 Table 1 278 279 4.2. Description of the mandible and lower teeth 280 Several complete or nearly complete hemimandibles as well as mandibular 281 fragments, were discovered at Ingarano (Figure 4, Supplementary note 1). 282 Measurements are reported in Table 1 (see also Supplementary table 4). All remains 283 belong to adult individuals. 284 285 Figure 4 286 287 The hemimandibles and lower teeth show a more marked variation in both size and 288 shape than that observed for cranium and upper dentition remains. Indeed, a remarkable 289 variation is observed for what concerns mandibular dimensions, number of the accessory 290 cuspids, development of the cingula in the P3 and P4, and presence/absence of the 291 metaconid in the M1. Also, the mandibular mental foramina are variable in number and 292 position. In particular, the majority of the specimens shows two mental foramina: one 293 located posteriorly to the distal border of the C1 and one ventral to the mid length of the P3; 294 several specimens possess a third foramen often placed at the level of the distal border of 295 the P3 (FG93, ING68, ING69, ING70, INGND318) or close to the distal border of the P4 296 (INGND318, INGND868, INGND869). The anterior portion of the masseteric fossa 297 generally reaches the level of the distal portion of the M1 in lateral view, except for ING68, 298 ING70, ING71, and ING72, in which it lies more posteriorly. The ventral border of the 299 mandibular corpus is almost straight but anteriorly, in correspondence with the alveolus of 12 300 the lower canine, it sharply bends dorsally. In medial view, the mandibular foramen opens 301 near to the ventral border of the corpus and about halfway between the distal margin of 302 the M1 and the posterior end of the dentary. The ramus is high and inclined 303 posterodorsally with an angle of ca. 45° with respect to the corpus long axis. The coronoid 304 process is relatively long at its base and the angular process is also elongated and 305 projects posteroventrally. In dorsal view, the mandibular condyle has a straight posterior 306 border and is inclined in posterolateral direction. 307 No lower incisors are preserved. The C1 is buccolingually compressed, smooth on 308 the lingual side, and with a weak longitudinal groove on the buccal one. It is separated 309 from the P3 by a long diastema. The P3 is buccolingually compressed and has a high and 310 straight protoconid, a weak crest connecting the mesial cingulum to the protoconid, a weak 311 distal accessory cuspid, and a marked distal cingulum; in addition, ten specimens 312 (IN15NS, ING68, ING71, ING72, INGND144, INGND145, INGND312, INGND318, 313 INGND373, and INGND870) show a small paraconid. The P4 is also buccolingually 314 narrow, with a high protoconid, which is however slightly projected distally. A paraconid, a 315 distal accessory cuspid, and a distal cingulum are also recognizable in all specimens; in 316 five specimens (ING63, ING68, ING69, INGND318, and INGND373) the distal cingulum is 317 well developed and similar in size to the distal accessory cuspid, whereas in the other 318 specimens (n=22) it is less prominent. The M1 (Figure 5) has a protoconid that is similar in 319 length, but higher than the paraconid. Generally, these cuspids are separated by a weakly 320 marked incision, with the exception of ING61, ING63, ING68, ING72, INGND144, 321 INGND319, INGND372, and INGND373, in which the incision is more pronounced. A 322 distinct metaconid separated from the protoconid blade (Boscaini et al., 2016: Fig. 5A), is 323 never observed in the Ingarano sample (n=27); a weak metaconid adjacent to the 324 protoconid blade (Boscaini et al., 2016: Fig. 5B) is observed in FG107, FG108, ING70, 325 ING73, ING79, INGND373, INGND868, and INGND869, whereas a small cuspulid-like 13 326 enamel inflection (Boscaini et al., 2016: Fig. 5C) is observed in IN15NS, ING67, ING71, 327 ING72, INGND145, and INGND319. However, the majority of the specimens (ING61, 328 ING62, ING63, ING68, ING69, ING74, INGND144, INGND312, INGND369, INGND372, 329 INGND867, INGND870, and MGPUTV135415) shows a completely smooth distal margin 330 of the protoconid blade (Boscaini et al., 2016: Fig. 5D). 331 332 Figure 5 333 334 4.3. Morphological comparisons 335 The crania from Ingarano display: (i) a short sagittal crest, (ii) long, well-separated, 336 and lyre-shaped temporal ridges, and (iii) the confluence in the same cavity of the jugular 337 and hypoglossal foramina, situated just behind the posteromedial border of the tympanic 338 bulla. All these characters fit the morphology of extant and fossil L. pardinus (Boule, 1919; 339 Boule and Villeneuve, 1927; García-Perea et al., 1985; García-Perea, 1996; Boscaini 340 et al., 2015; Figure 6). Conversely, this morphology contrasts with Early Pleistocene L. 341 issiodorensis (Viret, 1954; Kurtén, 1978; Cipullo, 2010; Cherin et al., 2013) (no Pliocene 342 neurocrania of this species are available) and with extant and fossil L. lynx (García-Perea 343 et al., 1985; García-Perea, 1996; Cassoli and Tagliacozzo, 1994b; Boscaini et al., 344 2015), which are characterized by a longer and more marked sagittal crest and shorter 345 temporal ridges. The jugular and hypoglossal foramina are separated by a septum in L. 346 lynx (Cassoli and Tagliacozzo, 1994b; Larivière and Walton, 1997; Boscaini et al., 347 2015), whereas the condition observed in L. issiodorensis from Olivola and Pantalla 348 (Figure 6) is similar to that of L. pardinus (i.e., confluent foramina). Unfortunately, it is not 349 possible to evaluate this character in the other analyzed specimens of L. issiodorensis 350 (Les Etouaires and Saint Vallier) due to preservation reasons (Supplementary figure 2). 14 351 As for the upper teeth, the P4 of the Ingarano sample never shows a distinct 352 ectoparastyle. According to Kurtén (1978) the ectoparastyle is generally present in L. 353 issiodorensis and L. lynx. 354 355 Figure 6 356 357 Concerning lower dentition, a detached distal cingulum is visible in all P3 (n=23) from 358 Ingarano. Conversely, only 5 out of 27 specimens show a P4 distal cingulum. A prominent 359 P4 distal cingulum was considered as characteristic for ‘L. spelaeus’ from Middle 360 Pleistocene French sites such as L’Escale, Lunel Viel, and Aragó Cave (Bonifay, 1971; 361 Testu, 2006). However, the same authors recognized a certain variation in this character’s 362 development, which should be considered of little taxonomic value (Boscaini et al., 2016). 363 In the Ingarano sample, a distinct M1 metaconid is absent in 19 out of 27 specimens 364 (70.4%), whereas a weak metaconid adjacent to the protoconid blade is visible in the 365 remaining M1 (29.6%). A detached M1 metaconid is observable in extant and fossil L. lynx 366 (Kurtén, 1963; Werdelin, 1987), whereas it is totally absent or barely developed in extant 367 L. pardinus (respectively, 83% and 14% of the 54 pardel lynx skulls analyzed by García- 368 Perea et al., 1985). In fossil remains of L. pardinus from southwestern Europe, the M1 369 metaconid shows a progressive trend of reduction from the late Early Pleistocene up to the 370 extremely low percentages of presence observed in extant populations (Bonifay, 1971; 371 García-Perea et al., 1985; Testu, 2006; Boscaini et al., 2016). In L. issiodorensis, the 372 development of a well-defined M1 metaconid shows an almost inverse chronological trend. 373 This cuspid is absent in Ruscinian and Early Villafranchian forms, starts to occur in Middle 374 Villafranchian forms, and is common in Late Villafranchian and Epivillafranchian forms 375 (Kurtén, 1963; Boscaini et al., 2016; Supplementary table 5). 15 376 In sum, the morphology of craniodental remains from Ingarano more closely 377 resembles L. pardinus s.l. than L. lynx and L. issiodorensis. In particular, the short sagittal 378 crest, the lyre-shaped temporal ridges, the absence of P4 ectoparastyle, and the absence 379 of a distinct M1 metaconid are characters clearly related to the pardel lynx. At the same 380 time, particularly noteworthy is the fact that the Ingarano crania share with both L. pardinus 381 and L. issiodorensis the anatomy of the jugular and hypoglossal foramina, which open in 382 the same depression. This character is here described in L. issiodorensis for the first time. 383 384 4.4. Biometric comparisons 385 Comparative cranial measurements of European extant and fossil lynxes are 386 reported in Supplementary table 6. Extant L. lynx and L. pardinus are well distinguishable 387 according to cranium size, with the first showing a total length normally exceeding 140 mm 388 and the second rarely reaching this value in the available comparative samples. Late 389 Pleistocene specimens of L. lynx are on average similar in size to L. issiodorensis, that is 390 to say very large overall. However, when considering fossil remains of L. pardinus from the 391 Pleistocene of southwestern Europe, the scenario becomes more complicated due to 392 general overlapping in cranium size (Supplementary table 6). Indeed, most of the 393 available fossils of L. pardinus show cranial dimensions closer to those of the Eurasian 394 lynx. In particular, the two crania from Ingarano, MGPT-PU 135415 and especially ING75, 395 are amongst the largest individuals in our comparative sample, with total lengths 396 exceeding the average values of each sub-samples. On the contrary, the crania from 397 Grotta del Principe and Grotte de l’Observatoire occupy an intermediate position, falling 398 between the ranges of extant L. lynx and L. pardinus (Supplementary table 6). The same 399 reasoning applies to the zygomatic width, although for this measurement the degree of 400 dimensional overlapping between extant populations is much higher, thus preventing 401 reliable taxonomic distinctions. Conversely, the length of the sagittal crest, which in turn 16 402 reflects the development of temporal ridges, is of high taxonomic importance being 403 significantly higher in L. lynx and L. issiodorensis, than in L. pardinus. The length of the 404 sagittal crest of the specimens from Ingarano are close to the average length of the extant 405 L. pardinus and to the few available lengths of fossil L. pardinus (Supplementary table 6). 406 Due to the relative scarcity of lynx well-preserved cranial findings in Europe, many of 407 the comparative analyses conducted so far are based on measurements of teeth (e.g., 408 Bonifay, 1971; Ficcarelli and Torre, 1977; Werdelin, 1981; Testu, 2006; Cipullo, 2010; 409 Ghezzo et al., 2015), among which the P4 and M1 seem to be the most informative 410 (Supplementary figure 3). The length of the P4 from Ingarano is undistinguishable from 411 that of fossil L. pardinus (p.value > 0.01), smaller than those of L. issioderensis and L. lynx 412 (p.value < 0.01), and larger than that of extant L. pardinus (p.value < 0.01). The range of 413 the M1 length of the Ingarano sample is very close to those of fossil L. pardinus and L. 414 issiodorensis (p.value > 0.01), falling between the ranges of extant L. pardinus and L. lynx 415 (Supplementary table 7). 416 To sum up, based on morphometric comparisons, the sample from Ingarano shows 417 remarkably large cranial dimensions. On the contrary, dental measurements are more in 418 line with the pardel lynx, and lower than those of L. issiodorensis and L. lynx. 419 420 5. Discussion 421 5.1. Taxonomy 422 The large lynx sample from Ingarano shows a distinctive craniodental morphology 423 including: short sagittal crest, long and lyre-shaped temporal ridges, confluence of the 424 hypoglossal and jugular foramina into the same cavity, absence of ectoparastyle in the P4, 425 and absence of a distinct metaconid in the M1. These features are typical of L. pardinus s.l. 426 (Boule, 1919; Boule and Villeneuve, 1927; Van den Brink, 1971; García-Perea et al., 427 1985; García-Perea, 1996; Larivière and Walton, 1997; Boscaini et al., 2015, 2016), 17 428 and allow an unambiguous attribution of the studied craniodental material to this species. 429 Opposite craniodental characters (i.e., long sagittal crest, separated hypoglossal and 430 jugular foramina, presence of P4 ectoparastyle, and presence of M1 metaconid) are 431 characteristics of L. lynx (Kurtén, 1963; García-Perea et al., 1985; Tumlison, 1987; 432 Boscaini et al., 2015, 2016). The more reliable criteria to separate L. pardinus from L. 433 issiodorensis are currently represented by the P4 length and the development of the 434 sagittal crest and the consequent pattern of the temporal ridges (although the distance 435 between the temporal ridges is subject to inter-individual variation in the extant pardel lynx 436 being usually wider in juveniles and females, the general lyre-shaped pattern is very 437 consistent in the species; García-Perea et al., 1985). Moreover, L. pardinus and L. 438 issiodorensis share the confluence of the hypoglossal and jugular foramina (Figure 7). 439 Considering L. issiodorensis as the putative ancestor of L. pardinus (Werdelin, 1981), this 440 can therefore be interpreted as a plesiomorphic character. This hypothesis is also 441 supported by the fact that confluent foramina are also found in the bobcat Lynx rufus 442 (Larivière and Walton, 1997), which is commonly considered the sister taxon to all the 443 other extant lynxes (Werdelin, 1981; Bininda-Emonds et al., 1999; Johnson et al., 444 2004, 2006). On the contrary, the Canadian lynx Lynx canadensis shares with L. lynx the 445 presence of a clear septum separating the two foramina (Lavoie et al., 2019). However, 446 the phylogenetic relationships of L. canadensis are not clear, as it is considered as the 447 sister taxon of L. lynx by some authors (Werdelin, 1981; Bininda-Emonds et al., 1999), 448 or as the sister taxon of the clade formed by L. lynx and L. pardinus by others (Johnson et 449 al., 2004, 2006). 450 451 Figure 7 452 18 453 The occurrence of L. pardinus in the Late Pleistocene (MIS 3) of Ingarano offers 454 some key cues for elucidating the natural history of this species in Europe. The taxonomic 455 status of the pardel lynx has long been a subject of debate, until molecular studies 456 (Beltrán et al., 1996) confirmed L. pardinus as a valid species. On the other hand, the so- 457 called ‘cave lynx’ L. spelaeus (or L. p. spelaeus) from the Middle-Late Pleistocene of 458 southwestern Europe, has been recently considered a junior synonym of L. pardinus 459 (Boscaini et al., 2016). The latter authors illustrated how the morphological characters 460 generally used to separate the two species are inconsistent. This is also supported by the 461 results of molecular analyses (Rodríguez-Varela et al. 2015), according to which Late 462 Pleistocene remains of “L. cf. spelaeus” from northwestern Italy were re-assigned to L. 463 pardinus based on haplotype similarity. Conversely, many authors (Werdelin, 1981; 464 Kurtén and Granqvist, 1987; García-Perea, 1997; Palombo et al., 2008; Garrido and 465 Arribas, 2008; Rodríguez-Hidalgo et al., 2020) in the last decades preferred to maintain 466 a subspecific distinction (i.e., L. p. spelaeus) for Middle-Late Pleistocene pardel lynxes 467 from southwestern Europe, following in some way the pioneering initial denomination (i.e., 468 “Felis (Lynx) pardinus race spelaea”) by Boule (1919). This taxonomic view is largely 469 based on the allegedly larger dimensions of the fossil forms compared to the extant ones. 470 In particular, according to the comprehensive study by Werdelin (1981), the sequence of 471 chronosubspecies formed by L. i. issiodorensis (early-middle Villafranchian) - L. i. 472 valdarnensis (late Villafranchian) - L. p. spelaeus (Middle-Late Pleistocene) - L. p. pardinus 473 (recent), would represent a phylogenetic lineage, characterized by a progressive reduction 474 in body size and an increase of relative length of M1. However, the fossil record of 475 European lynxes documented between ca. 1.6 and 0.6 Ma, which should include the L. 476 issiodorensis - L. pardinus split, is mostly represented by isolated teeth and mandibles of 477 unclear diagnostic value. Two remarkable exceptions at the limits of this range, that is, the 478 neurocranium from the Avenc Marcel (1.7–1.6 Ma; Boscaini et al., 2015) and the two 19 479 crania from L’Escale (ca. 0.6 Ma; Bonifay, 1971), are reliably referable to L. pardinus. The 480 lack of cranial remains for more than 1 Ma has forced research over the past fifty years to 481 focus on dental morphology, but despite many efforts, only some tooth measurements and 482 the presence/absence of the P4 ectoparastyle and the M1 metaconid have been suggested 483 as possible diagnostic traits (e.g., Bonifay, 1971; Ficcarelli and Torre, 1977; Werdelin, 484 1981; Testu, 2006; Cipullo, 2010; Ghezzo et al., 2015). However, the latter features 485 show a marked interindividual variation when observable in large samples (Boscaini et 486 al., 2016). The fossils from Ingarano represents an excellent example, given that, based 487 only on the frequency of the aforementioned dental characters, it could equally be 488 attributed to L. pardinus or L. issiodorensis (Figure 7; Supplementary table 5). This 489 confirms the inconsistency of tooth morphology, especially if considered alone and on 490 small fossil samples, for taxonomic attributions of European Plio-Pleistocene lynx remains. 491 Therefore, unless samples are large enough to evaluate the frequency of the 492 aforementioned dental characters (e.g., Epivillafranchian L. pardinus from Vallparadís 493 Estació and Cueva Victoria; Boscaini et al., 2016), we suggest the use of the open 494 nomenclature Lynx sp. for all European remains approximately comprised in the interval 495 from 1.6 to 0.6 Ma, previously referred to L. issiodorensis or to ‘cave lynx’. These include 496 the records from Pirro Nord (Italy; Petrucci et al., 2013), Monte Argentario (Italy; Cherin 497 et al., 2018), Valdemino (Italy; Ghezzo et al., 2015), La Sartanette (France; Palombo and 498 Valli, 2004), Sainzelles (France; Palombo and Valli, 2004), Le Vallonnet (France; Moullé 499 et al., 2006), Solehilac (France; Kurtén, 1957), Sierra de Quibas (Spain; Montoya et al., 500 1999, 2001b), Venta Micena (Spain; Moyà-Solà et al., 1981; Boscaini et al., 2016), 501 Untermassfeld (Germany; Hemmer, 2001), Mauer (Germany; Voelcker, 1930), Mosbach 502 (Germany; Voelcker, 1930), and Apollonia 1 (Greece; Koufos, 1992; Koufos and 503 Kostopoulos, 1997). Details are reported in Supplementary table 5. 20 504 Additional information is provided by biometric analyses of upper and lower 505 carnassials (Figure 7; Supplementary tables 3, 4, 6). The lengths of the P4 and M1 show 506 similar decreasing trends from Villafranchian L. issiodorensis (before 1.7 Ma) to later 507 forms, that is, Lynx sp. (1.6–0.6 Ma) and L. pardinus s.l. (including the Ingarano sample; 508 Supplementary figure 3). This would support the hypothesis that dentognathic remains of 509 Lynx sp. in the 1.6–0.6 Ma interval (see above) may be related to L. pardinus, but the 510 absence of sufficiently complete crania encourages a cautious approach. The carnassials 511 of the extant pardel lynx are usually smaller, whereas those of the Eurasian lynx are 512 considerably larger, reaching and sometimes exceeding the size of some L. issiodorensis 513 specimens (Supplementary figure 3). This trend indicates that the M1 has undergone a 514 slight decrease in length in the L. issiodorensis - L. pardinus lineage, rather than an 515 increase as proposed by Werdelin (1981). Moreover, the larger size of the P4 observed in 516 L. issiodorensis does not seem to correspond to a relatively elongated cranium in this 517 species, which is similar in length to specimens MGPT-PU 135415 and ING75 from 518 Ingarano and to that of extant L. lynx (Supplementary tables 1, 3, 6, 7). This is also 519 supported by our results on body mass estimates (see below), which evidence a 520 substantial uniformity among European Pleistocene lynxes (Supplementary table 8). 521 In sum, in light of our results, we suggest a more parsimonious and simplified 522 taxonomy for European fossil lynxes (Supplementary table 9). Lynx issiodorensis occurs 523 in the interval between the Ruscinian and early late Villafranchian. The validity of the two 524 subspecies L. i. issiodorensis and L. i. valdarnensis, which are only differentiated based on 525 tooth dimensions (Werdelin, 1981), is questionable based on our data. In the early late 526 Villafranchian (ca. 1.7–1.6 Ma), L. pardinus appears in western Mediterranean Europe 527 (Avenc Mancel). The possible co-existence between L. issiodorensis and L. pardinus 528 cannot be verified on the basis of available data. During the following period (i.e., latest 529 Villafranchian to earliest Middle Pleistocene), with few exceptions (i.e., L. pardinus from 21 530 Vallparadís Estació and Cueva Victoria), an undeterminable form (Lynx sp.) probably 531 related to L. pardinus, is found in Europe. The occurrence of L. pardinus s.l., at least in 532 southwestern Europe, is attested from the early Middle Pleistocene onwards. 533 534 5.2. Size variation in European fossil lynxes 535 The Ingarano sample, with its well-preserved cranial remains, allows to better explore 536 the evolution of body size in European fossil lynxes, using condylobasal lengths as a proxy 537 to estimate body mass following Van Valkenburgh (1990) (see Materials and methods). 538 Of course, it is worth remembering that since the estimates are based on individual fossil 539 specimens from different geographical and chronological contexts, our results do not 540 presume to provide precise data, but only to reconstruct possible overall trends in the 541 evolution of body size (and to compare these trends with what is suggested in the 542 literature for fossil lynxes; see below). This is in agreement with the purpose with which the 543 considered prediction equations were originally proposed (Van Valkenburgh, 1990). 544 Having ascertained that the Ingarano individuals are all adults as shown by the fully 545 erupted permanent dentition and fusion of the cranial sutures (see Description), an aspect 546 worth mentioning before discussing the data on body mass is sexual dimorphism. It has 547 been proposed that sexual dimorphism may explain intraspecific differences in body size 548 in some European Plio-Pleistocene felids (e.g., Acinonyx pardinensis, Panthera 549 gombaszoegensis; Petrucci et al., 2013; Cherin et al., 2018). However, as in extant taxa 550 (Macdonald et al., 2010), the effect of sex on size is more pronounced in large felid 551 species. As for extant L. pardinus, differences in cranial measurements between adult 552 males and females are relatively low (e.g., mean condylobasal length is about 5% larger in 553 males) and are reflected in a weight difference of about 27% between sexes (Beltran and 554 Delibes, 1993). Similar if not smaller differences are found in extant L. lynx (e.g., mean 555 condylobasal length about 6% larger in males, weight difference of about 16% between 22 556 sexes; Wiig and Andersen, 1986; Červený and Koubek, 2000; Gomerčič et al., 2010). 557 In addition to these low average differences, the majority of cranial measurements show 558 largely overlapping ranges of variation between males and females in both extant species, 559 so much so that, at least for L. pardinus, distinguishing the sex of an individual on 560 craniometric grounds is not reliable (Beltran and Delibes, 1993). Finally, Van 561 Valkenburgh's (1990) prediction equation used in this paper, although originally obtained 562 by regression starting from a mixed sample of males and females, does not allow for sex- 563 differentiated estimates. For all these reasons, sexual dimorphism cannot be taken into 564 consideration either for a priori sex separation of our samples on a biometric basis, or to 565 speculate on the possible causes of differences in estimated body masses. 566 Figure 8 shows the estimates obtained from the Ingarano sample and other well- 567 preserved crania available in the European Plio-Pleistocene fossil record. As for L. 568 issiodorensis, what first stands out is the remarkable body mass (almost 35 kg) estimated 569 for individual MNB Prr 200 from Les Etouaires (France; latest Pliocene, ca. 2.8 Ma; dating 570 from Nomade et al., 2014). Although the cranium is slightly crushed dorsoventrally 571 (Supplementary figure 2), length measurements should be considered as reliable 572 (Kurtén, 1978). This weight value is at least 10 kg higher than the maximum reached by 573 males of the extant Eurasian lynx (Heptner and Sludskii, 1992; Gomerčić et al., 2010), 574 and is more similar to those recorded for larger felids (e.g., maximum body mass of female 575 puma; Macdonald et al., 2010). However, rare reports of extremely large individuals of L. 576 lynx are present in the literature (32 kg in Ognev, 1935; 38 kg in Kazlauskas and 577 Matuzevicius, 1970). On the other hand, the second individual from Les Etouaires (MNB 578 Prr 411) for which it was possible to estimate the body mass (21.8 kg), is significantly 579 smaller and very similar to MNB Ol 1 from Olivola (Early Pleistocene, ca. 2.1 Ma; dating 580 from Napoleone et al., 2003). It is worth mentioning that Werdelin (1981) referred some 581 specimens from Olivola to L. i. valdarnensis, considered to be smaller than L. i. 23 582 issiodorensis (which includes the Les Etouaires material, among others) on the basis of 583 dental measurements. The cranium SABAP_UMB 337653 from Pantalla (2.1–1.9 Ma; 584 dating from Cherin et al., 2019) referred to L. i. valdarnensis based on morphological and 585 biometric similarity with the samples from Olivola and Upper Valdarno (Cherin et al., 586 2013), retrieved a slightly lower body mass (19.7 kg). However, the individual DE 04 Liv II 587 from Pirro Nord (1.6–1.3 Ma; Arzarello et al., 2007; Lopez-García et al., 2015), 588 tentatively referred to L. issiodorensis (Petrucci et al., 2013) but here re-assigned to Lynx 589 sp. due to the lack of clear craniodental characters, resulted as significantly larger (25.4 590 kg). In sum, with the exception of the extremely large-sized MNB Prr 200 from Les 591 Etouaires, all other specimens of L. issiodorensis show estimated body masses falling in 592 the upper part of the range of L. lynx (Heptner and Sludskii, 1992; Gomerčić et al., 593 2010). Similar values (19.3–22.9 kg) were also calculated for some Late Pleistocene 594 individuals of the latter species (Supplementary table 8). This is in agreement with the 595 reconstruction of the life appearance of L. issiodorensis by Kurtén (1978), according to 596 which this extinct felid would have been quite similar to the extant L. lynx, but with a 597 relatively larger skull, slightly longer body (ca. 4%), and shorter limbs. 598 599 Figure 8 600 601 Body mass data on L. pardinus s.l. are noteworthy for the purposes of this study. The 602 estimates for the two better preserved crania from the early Middle Pleistocene of L’Escale 603 (France; Bonifay, 1971) are higher than those from the late Middle(?)-Late Pleistocene of 604 Grotte de l’Observatoire (Monaco; Boule and Villeneuve, 1927) and Late Pleistocene of 605 Grotta del Principe (Italy; Boule, 1919). These results alone, although based on a limited 606 number of crania, seem to disprove the statement that the Middle-Late Pleistocene ‘cave 607 lynx’ underwent a further progressive reduction in size along the evolutionary lineage that 24 608 led to the pardel lynx (Werdelin, 1981). In addition to this, the two complete crania from 609 Ingarano (MGPT-PU 135415 and ING75) resulted in remarkably high body mass 610 estimates (23.7 and 25 kg, respectively). The relatively recent (ca. 40 ka) Ingarano 611 individuals are therefore the largest ever reported for L. pardinus s.l. 612 Kurtén and Granqvist (1987: 42) defined the Middle-Late Pleistocene L. pardinus 613 as “plastic in size”. These geographical and chronological body mass variations indicate 614 that this species has not actually experienced a progressive reduction in size, thus 615 supporting the inconsistency of the fossil (chrono)subspecies L. p. spelaeus, as proposed 616 by Boscaini et al. (2016) on morphological grounds. 617 Once recognized the body mass variation in the fossil record of L. pardinus, it 618 remains to be clarified how this relates with the populations living today in southwestern 619 Spain. The average body mass of the extant pardel lynx ranges between ca. 9 kg 620 (females) and 13 kg (males; Beltrán and Delibes, 1993). These values are much closer to 621 those of Nearctic species, that is, L. rufus (ca. 7/10 kg in females/males; Banfield, 1987) 622 and L. canadensis (ca. 9/11 kg in females/males; Sunquist and Sunquist, 2002), than to 623 those of L. lynx (ca. 17/20 kg in females/males; Heptner and Sludskii, 1992). However, 624 according to studies on conservation (Simón, 2012, and references therein) and molecular 625 biology (Johnson et al., 2004), L. pardinus suffered a recent strong demographic 626 contraction (which however would not appear to be associated with a substantial reduction 627 in genetic variation, at least in the last 50 ka; Rodríguez et al. [2011]). The dramatic 628 historical decline of pardel lynx populations prevents us to consider the few survivors as a 629 reliable representation of the species’ dimensional variation at the geological time scale. At 630 the state of the art, we can only point out that the body mass estimates obtained for some 631 fossil individuals, namely those from Grotte de l’Observatoire and Grotta del Principe 632 (Figure 8), are close to the maximum values recorded for extant males from southern 633 Spain (ca. 14.5 kg; Beltrán and Delibes, 1993). 25 634 635 5.3. Paleoecology 636 Whether the body size variations observed in European fossil lynxes are related to 637 paleoclimatic factors, such as glacial/interglacial dynamics and/or latitude-driven factors, is 638 still unclear. Actually, the majority of the lynx remains from Ingarano (including the 639 aforementioned crania) come from the lower part of the stratigraphic succession (Petronio 640 and Sardella, 1998), which is correlated to a cold phase during MIS 3 (Bedetti and Pavia, 641 2007). However, recent works conducted on other large mammal taxa from the same 642 region, suggest a cautious approach in recognizing correlations between Middle-Late 643 Pleistocene climatic fluctuations and body size in such a peripheral Mediterranean area 644 (cf. Sardella et al. [2014] for Canis lupus and Iannucci et al. [2020] for Sus scrofa). 645 The extant pardel lynx also shares with North American lynx species some ecological 646 adaptations, among which the feeding preference upon lagomorphs. The European rabbit 647 Oryctolagus cuniculus and the snowshoe hare Lepus americanus account respectively for 648 80–99% (Ferreras et al., 2010) and 60–97% (Sunquist and Sunquist, 2002) of the diet 649 of L. pardinus and L. canadensis, and also L. rufus preys primarily on lagomorphs 650 (Sunquist and Sunquist, 2002). These diet preferences strongly influence the pardel 651 lynx’s habitat selection, as the species is only found today in Mediterranean shrublands 652 with scrub-pasture ecotones and dense rabbit populations (Palomares et al., 2000; 653 Palomares, 2001). Unfortunately, the available data do not allow obtaining clear 654 information on the paleoecology of Pleistocene L. pardinus. Some anatomical features 655 such as the very short sagittal crest and slender dentition, were interpreted as indicators of 656 a diet based on small prey, throughout the history of L. pardinus s.l. (Werdelin, 1981; 657 García-Perea, 1996; Boscaini et al., 2015). Some data even seem to suggest a close 658 correlation between the evolution of the pardel lynx and that of the European rabbit 659 (Kurtén, 1968), whose geographical distributions followed the same contraction dynamics 26 660 according to Pleistocene glacial-interglacial oscillations (Lopez-Martinez, 2008; Ferreras 661 et al., 2010). Moreover, recent taphonomic analysis carried out on Late Pleistocene (MIS 662 3) fossils from Cova del Gegant (Spain) pointed out a biogenic origin of the paleontological 663 accumulation of layer IIIa, characterized by the very abundant occurrence of lynx and 664 rabbit remains and interpreted as a L. pardinus den (Rodríguez-Hidalgo et al., 2020). 665 However, we cannot currently confirm if the pardel lynx has always been a specialized 666 rabbit hunter, or if it also relied on larger prey in the past, as the large body mass that we 667 estimated for some fossil individuals would seem to suggest. 668 Similarly, we do not have a clear picture of the habitat preferences of extinct 669 populations of L. pardinus, due to the lack of paleoecological information on the sites from 670 which it has been reported. The few available data on Ingarano comes from the analysis of 671 the fossil bird assemblage, which suggests a paleoenvironment dominated by open areas, 672 with secondary presence of wetlands and woods (Bedetti and Pavia, 2007). This 673 reconstruction only partially fits the aforementioned restricted Mediterranean habitat of the 674 extant pardel lynx. Based solely on the geological contexts in which fossil remains of L. 675 pardinus have been found, largely represented by caves and other karst cavities, we can 676 speculate that the species has always maintained the need to occupy areas with the 677 presence of natural cavities, used as natal dens (Fernández et al., 2002, 2006; 678 Rodríguez-Hidalgo et al., 2020). In general, also taking into account the much wider 679 geographical distribution, it is reasonable to think that, at least outside the Iberian 680 Peninsula, the species had more diversified food and habitat preferences in the 681 Pleistocene than today. 682 683 5.4. Paleobiogeography 684 According to the available data, the first appearance of L. pardinus in Europe is 685 rooted in the Early Pleistocene. The record from Avenc Mancel (1.7–1.6 Ma; Boscaini et 27 686 al., 2015) suggests that the origin of this species may be part of the important faunal 687 turnover occurred at the Gelasian-Calabrian boundary (Sardella et al., 2018). The glacial 688 pulse recorded around 1.8 Ma (MIS 64-62) could have played a key role in the separation 689 of populations of L. issiodorensis, triggering the speciation of L. pardinus (Boscaini et al., 690 2015). Unfortunately, as stated above, the characters useful for a reliable attribution of 691 fossils to L. issiodorensis or L. pardinus are only found in well-preserved cranial remains, 692 whereas dental morphology can be considered only if analyzed in large samples. For 693 these reasons, most European records from the latest Villafranchian-Epivillafranchian 694 should be prudentially referred to Lynx sp. (Figure 9; Supplementary tables 5, 9). Due to 695 this uncertainty, it is currently hard to evaluate the impact on lynx evolution of another 696 crucial turnover, that is, the Early-Middle Pleistocene Transition (EMPT), also known as 697 the “Mid-Pleistocene Revolution” (ca. 1.2–0.4 Ma; Head and Gibbard, 2005). During this 698 interval, global-scale climatic changes strongly affected the faunal and floral composition 699 of European ecosystems also in Mediterranean areas (Kahlke et al., 2011; Magri and 700 Palombo, 2013; Strani et al., 2019; Cherin et al., 2020), although in a more mitigated 701 way if compared to higher latitudes (Head and Gibbard, 2015). However, starting from ca. 702 0.6 Ma, i.e, from the L'Escale record (Bonifay, 1971), L. pardinus s.l. is undoubtedly 703 recognized in southwestern Europe (Iberian Peninsula, southern France, and Italy; Figure 704 9). If new discoveries from Untermassfeld (Germany, ca. 1.0 Ma) confirm the presence of 705 L. issiodorensis at the site (Hemmer, 2001), an interesting latitudinal separation between 706 the last representatives of this species in central Europe and L. pardinus in Mediterranean 707 areas during the EMPT would be expected. 708 To date, the outstanding sample from Ingarano represents the southernmost record 709 of the species in Italy and the easternmost in Europe (Figure 9). The lynx fossil record 710 along the eastern coast of the Adriatic Sea is still fragmentary. Middle-Late Pleistocene 711 lynx remains were reported from several caves in Slovenia and Croatia (Miracle, 1991, 28 712 1995), but nevertheless these historical reports of L. pardinus (e.g., Veternica cave) were 713 recently denied (Miracle et al., 2010). Although these coastal areas presumably showed 714 paleoecological conditions favorable to the presence of the pardel lynx, new discoveries 715 and a reappraisal of preexisting collections are needed to confirm the possible occurrence 716 of the species. 717 718 Figure 9 719 720 One of the main issues affecting the reconstruction of the paleobiogeographical 721 history of L. pardinus in Europe is the co-occurrence with L. lynx. This species spreads in 722 Europe at the beginning of the Late Pleistocene (MIS 5e) and it is recorded together with 723 L. pardinus in some Mediterranean localities, e.g., Grotte de l’Observatoire layers A-F 724 (Boule and Villeneuve, 1927), Grotte de Cotencher (Dubois and Stehlin, 1933), and 725 Arene Candide cave (Rodríguez-Varela et al., 2015). However, the size-based 726 classifications performed in the past have probably contributed to overestimate the 727 presence of L. lynx in the Late Pleistocene of southern Europe, as it is the case for 728 Ingarano. Molecular data indicate that the Eurasian lynx was spread in the Mediterranean 729 area during the Last Glacial Period (MIS 2) (e.g., Grotta della Madonna, Arene Candide, 730 and Rascaño; Rodríguez-Varela et al., 2015, 2016) (Supplementary table 8). Moreover, 731 L. lynx is mainly documented from isolated postcranial remains until the Last Glacial stage, 732 and only a few exceptionally preserved crania were found (e.g., the layers A-F of Grotte de 733 l’Observatoire, Boule and Villenueve, 1927; Col d’Aran, de Beaufort, 1965; Rèseau de 734 Cèbèri, Clot & Besson, 1974; Arene Candide, Cassoli and Tagliacozzo, 1994a). 735 Similarly to what recognized from 1.7 Ma to 0.6 Ma, the early Late Pleistocene (MIS 5 to 736 MIS 3) is characterized by the lack of cranial remains clearly ascribable to L. lynx, 737 therefore, the earliest occurrence of this species during the MIS 5e cannot be confirmed. 29 738 The high ecological plasticity of the Eurasian lynx probably favored the replacement 739 of L. pardinus in most the continent, with the exception of the Iberian Peninsula 740 (Rodríguez-Varela et al., 2015), which probably acted as a refugium for the latter species. 741 As a matter of fact, L. lynx is reported from the latest Pleistocene until historical times in 742 northern Spain (Ferreras et al., 2010; Clavero and Delibes, 2013; Rodríguez-Varela et 743 al., 2016). Nevertheless, several authors reported the presence of L. pardinus in French 744 Bronze Age deposits (Gagniere, 1926; Vigne, 1996; Vigne and Pascal, 2003) thus 745 suggesting that the contraction of its geographical range has occurred in recent times, and 746 was probably further driven by human pressure. A similar decline has been documented 747 for the Eurasian lynx in the Italian Peninsula, where human impact led to the complete 748 eradication of Apennine populations in 1850 and Alpine ones in 1920 (Bologna and 749 Mingozzi, 2003). 750 The unreliability of size-based criteria for the taxonomic recognition of European 751 fossil lynxes confirmed in this work opens many questions on the paleobiogeography of L. 752 pardinus and L. lynx, especially in the Late Pleistocene. Some answers may come in the 753 future from genetic analyses, which have proved useful in differentiating the two species 754 (Rodríguez-Varela et al., 2015, 2016), as well as from the discovery of sufficiently 755 complete cranial material, such as those from Ingarano (Figure 10). 756 757 Figure 10 758 759 Acknowledgments 760 The authors are grateful to M. Danti (G. Vannini Hospital, Rome) for the access, 761 assistance, and facilities for CT scans; to R. Carlini (Museo Civico di Zoologia, Rome) for 762 the access to the comparative specimens; to R. Rodríguez-Varela for conducting 763 paleoDNA analyses of the Ingarano lynx sample; to L. Werdelin for providing raw 30 764 measurements of extant lynxes; to A. Cipullo and P. D’Aguanno for their help. RS wishes 765 to thank V. Simeonovski for the useful suggestions. MC wishes to thank M. Antón and the 766 whole crew (Sara, Miguel, Puri, Jan, Rafa, Diego, Edurne, Marisa, Joserra, and Manu) of 767 the December 2018 trip to the Parque Natural de Sierra de Andújar, which allowed him to 768 admire closely the wonderful pardel lynx. LC thanks G. Schulz and B. Müller for granting 769 access to the CT-scan facility at the Biomaterials Science Center of the University of 770 Basel. 771 772 Founding 773 “Grandi Scavi” Sapienza University of Rome 2018-2019 (Raffaele Sardella); “Fondo 774 Ricerca di Base 2019” Department of Physics and Geology, University of Perugia (Marco 775 Cherin and Dawid A. Iurino via Roberto Rettori); Spanish Agencia Estatal de 776 Investigación/European Regional Development Fund of the European Union (CGL2017- 777 82654-P, AEI/FEDER EU) (Joan Madurell-Malapeira); ex 60% Grant 2019-2020 778 Department of Earth Sciences, University of Turin (Marco Pavia). 779 780 Additional files 781 Supplementary information 782 783 References 784 Andersen, T., Wiig, Ø. (1984). Growth of the skull of Norwegian lynx. Acta Theriol. 28, 89- 785 100. 786 Anelli, F. (1959). Prime ricerche paleontologiche della grotta Masseria del monte 787 Conversano Murge in Bari. Le Grotte d’Italia, 3, 1–34. 788 Argant, A., (2004). Les Carnivores du gisement Pliocène final de Saint-Vallier (Drôme, 789 France). Geobios, 37, S133–S182. 31 790 Arribas, A. (1994). Los macromamíferos del yacimiento mesopleistoceno de Villacastín 791 (Segovia, España). Boletín Geológico y Minero, 105, 344–361. 792 Arzarello, M., Marcolini, F., Pavia, G., Pavia, M., Petronio, C., Petrucci, M., Rook, L., 793 Sardella, R. (2007). Evidence of earliest human occurrence in Europe: the site of 794 Pirro Nord (Southern Italy). Sci. Nat., 94, 107–112. 795 Banfield, A. W. F. (1987). The mammals of Canada. University of Toronto Press, Ontario, 796 pp. 438. 797 Bedetti, C., Pavia, M. (2007). Reinterpretation of the late Pleistocene Ingarano cave 798 deposit based on the fossil bird associations (Apulia, South-Eastern Italy). Riv. It. Pal. 799 Strat., 113, 487–507. 800 Beltrán, J.F., Delibes, M. (1993). Physical characteristics of Iberian lynxes (Lynx pardinus) 801 from Doñana, southwestern Spain. J. Mammal., 74, 852–862. 802 Beltrán, J. F., Rice, J. E., Honeycutt, R. L. (1996). Taxonomy of the Iberian lynx. Nature, 803 379, 407–408. 804 Bininda Emonds, O.R.P., Gittleman, J.L., Purvis, A. (1999). Building large trees by 805 combining phylogenetic information: a complete phylogeny of the extant Carnivora 806 (Mammalia). Biol. Rev., 74, 143–175. 807 Bologna, M. A., Mingozzi, T. (2003). Distribuzione geografica. In: Boitani, L., Lovari, S. & 808 Vigna Taglianti, A. (Eds.), Fauna d’Italia. Mammalia III. Carnivora-Artiodactyla. 809 Edizioni Calderini, Bologna, pp. 226–234. 810 Bonifay, M. F. (1971). Carnivores quaternaires du Sud-Est de la France. Mémoires du 811 Muséum National d’Histoire Naturelle, Paris, série C, t. XXI, pp. 377. 812 Boscaini, A., Madurell-Malapeira, J., Llenas, M., Martínez-Navarro, B. (2015). The origin of 813 the critically endangered Iberian lynx: Speciation, diet and adaptive changes. Quat. 814 Sci. Rev., 123, 247–253. 815 Boscaini, A., Alba, D. M., Beltrán, J. F., Moya-Sola, S., Madurell-Malapeira, J. (2016). 816 Latest Early Pleistocene remains of Lynx pardinus (Carnivora, Felidae) from the 817 Iberian Peninsula: Taxonomy and evolutionary implications. Quat. Sci. Rev., 143, 818 96–106. 819 Boule, M. (1919). Les Grottes de Grimaldi (Baoussé-Roussé). Géologie et Paléontologie. 820 Imprimerie de Monaco, 1, 271–279. 821 Boule, M., Villeneuve, L. de (1927). La Grotte de l'Observatoire à Monaco. Archives de 822 l'Institut de Paléontologie Humaine, 1, 113 p 32 823 Capasso Barbato, L., Cassoli, P. F., Minieri, M. R., Petronio, C., Sardella, R., Scarano, M. 824 (1992). Note preliminari sulla fauna pleistocenica di Ingarano (Apricena, Foggia). 825 Boll. Soc. Paleontol. Ital., 31, 325–334. 826 Capasso Barbato, L., Di Stefano, G., Petronio, C., Sardella, R. (1998). Pleistocene 827 mammal faunas from Ponte Molle (Rome). Quat. Int., 47, 73–75. 828 Cassoli, P. F., Tagliacozzo, A. (1994a). I macromammiferi dei livelli tardo pleistocenici 829 delle Arene Candide (Savona, Italia): considerazioni paleontologiche e 830 archeozoologiche. Quaternaria Nova, 4(10). 831 Cassoli, P. F., Tagliacozzo, A. (1994b). Considerazioni paleontologiche, paleoecologiche 832 e archeozoologiche sui macromammiferi e gli uccelli dei livelli del Pleistocene 833 superiore del Riparo di Fumane (VR) scavi 1988–91. Boll. Mus. Civ. Stor. Nat. 834 Verona, 18, 349–445 835 Červený, J., Koubek, P. (2000). Variabilty of body and skull dimensions of the lynx (Lynx 836 lynx) in the Czech Republic. Lynx (Praha) 31, 5-12. 837 Cherin, M., Iurino, D. A., Sardella, R. (2013). New well-preserved material of Lynx 838 issiodorensis valdarnensis (Felidae, Mammalia) from the Early Pleistocene of 839 Pantalla (central Italy). Boll. Soc. Paleontol. Ital., 52, 103–111. 840 Cherin, M., Iurino, D. A., Zanatta, M., Fernandez, V., Paciaroni, A., Petrillo, C., Rettori, R., 841 Sardella, R. (2018). Synchrotron radiation reveals the identity of the large felid from 842 Monte Argentario (Early Pleistocene, Italy). Sci. Rep., 8, 1-10. 843 Cherin M., D’Allestro V., Masini F. (2019) - New bovid remains from the Early Pleistocene 844 of Umbria (Italy) and a reappraisal of Leptobos merlai. J. Mamm. Evol., 26, 201–224. 845 Cherin M, Alba DA, Crotti M, Menconero S, Moullé PE, Sorbelli L, Madurell-Malapeira J. 846 (2020). The post-Jaramillo persistence of Sus strozzii (Suidae, Mammalia) in Europe: 847 New evidence from the Vallparadís Section (NE Iberian Peninsula) and other coeval 848 sites. Quat. Sci. Rev., 233, 106234. doi: 10.1016/j.quascirev.2020.106234. 849 Cipullo, A. (2010). L’evoluzione del genere Lynx dal Pliocene al Pleistocene medio 850 nell’area mediterranea. Phd Dissertation, University of Florence, pp. 142. 851 Clot, A., Besson, J. P. (1974). Nouveaux restes osseux de Lynx dans les Pyrénées. 852 Bulletin de la Société d’Histoire naturelle de Toulouse. 110, 157–169. 853 Clavero, M., Delibes, M. (2013). Using historical accounts to set conservation baselines: 854 the case of Lynx species in Spain. Biodivers. Conserv., 22, 1691–1702. 33 855 Curcio, M. T., Contoli, L., Di Canzio, E., Kotsakis, T. (2005). Preliminary analysis of the 856 first lower molar variability in Late Pleistocene and living populations of Terricola savii 857 (Arvicolidae, Rodentia). Geologia Alpina, 2, 91–98. 858 de Beaufort, F. (1965). Lynx des Pyrénées, Felis (L.) lynx lynx (L.). Mammalia, 29(4), 598– 859 601. 860 Del Campana, D. (1925). Ancora sulla Lynx issiodorensis Cr. et Job. del Pliocene Italiano. 861 Riv. Ital. Paleont., 31, 9–15. 862 Depéret C. (1890). Les animaux pliocènes du Roussillon. Mémoires de la Société 863 Géologique de France, 3, 1–194. 864 Dubois, A., Stehlin, H.G. (1933). La Grotte de Cotencher, station moustérienne. 865 Birkhäuser, Basel. 866 Fabrini, E. (1896). La lince del Pliocene italiano. Palaeontogr. Ital., 2, 1–24. 867 Fernández, N., Palomares, F., Delibes, M. (2002). The use of breeding dens and kitten 868 development in the Iberian lynx (Lynx pardinus). J. Zool., 258, 1–5. 869 Fernández, N., Delibes, M., Palomares, F. (2006). Landscape evaluation in conservation: 870 molecular sampling and habitat modeling for the Iberian lynx. Ecological Applications, 871 16, 1037–1049. 872 Ferreras, P., Rodríguez, A., Palomares, F., Delibes, M. (2010). Iberian lynx: the uncertain 873 future of a critically endangered cat. In: Macdonald, D.M., Loveridge, A.J. (Eds.), 874 Biology and Conservation of Wild Felids. Oxford University Press Inc., New York, pp. 875 507–520. 876 Ficcarelli, G., Torre, D. (1977). Phyletic relationships between Lynx group issiodorensis 877 and Lynx pardina. Boll. Soc. Paleontol. Ital., 16, 197–202. 878 Fosse, P., Fourvel, J.B., Madelaine, S. (2020). Le lynx Lynx pardinus spelaeus Boule, 879 1910 du Pléistocène moyen de la grotte de l'Escale (Bouches-du-Rhône, France): 880 données paléontologiques et taphonomiques. Paleo. Revue d'archéologie 881 préhistorique, 30(2), 108-137. 882 Gagnière, S. (1926). Sur un maxillaire inférieur de Lynx de l'Age du Bronze. Bibliographie, 883 n°47, et Hugues, A., n°7, pp. 65–67. 884 García, N., Arsuaga, J.-L. (1998). The carnivore remains from the hominid-bearing 885 Trinchera-Galería, Sierra de Atapuerca, Middle Pleistocene site (Spain). Geobios, 886 31, 659–674. 887 García-Perea, R. (1991). Variabilidad morfologica del genero Lynx Kerr, 1792 (Carnivore, 888 Felidae). Madrid: University of Complutense. 34 889 García-Perea, R. (1996). Patterns of postnatal development in skulls of lynxes, genus 890 Lynx (Mammalia: Carnivora). J. Morphol., 229, 241–254. 891 García-Perea, R. (1997). Phylogeny and Conservation of Iberian Lynxes. Cat News, 27, 892 23–24. 893 García-Perea, R., Gisbert, J., Fernando, P. (1985). Review of the Biometrical and 894 Morphological Features of the Skull of the Iberian Lynx, Lynx pardina (Temminck, 895 1827). Säugetierkundliche Mitteilungen, 32, 249–259. 896 Garrido, G., Arribas, A. (2008). Generalidades Sobre Los Carnívoros Del Villafranquiense 897 Superior En Relación Con El Registro Fósil De Fonelas P-1. Cuadernos del Museo 898 Geominero, 10, 1–62. 899 Ghezzo, E., Boscaini, A., Madurell-Malapeira, J., Rook, L. (2015). Lynx remains from the 900 Pleistocene of Valdemino cave (Savona, Northwestern Italy), and the oldest 901 occurrence of Lynx spelaeus (Carnivora, Felidae). Rendiconti Lincei, 26, 87–95. 902 Ginsburg L. (1998). Le gisement de vertébrés pliocènes de Çalta, Ankara, Turquie. 5. 903 Carnivores. Geodiversitas, 20, 379–396. 904 Gomerčić, T., Sindičić, M., Gomerčić, M.D., Gužvica, G., Frković, A., Pavlović, D., Kusak, 905 J., Galov, A., Huber, D. (2010) Cranial morphometry of the Eurasian lynx (Lynx lynx 906 L.) from Croatia. Veterinarski Arhiv, 80, 393–410. 907 Head, M.J., Gibbard, P.L. (2005). Early–Middle Pleistocene transitions: an overview and 908 recommendation for the defining boundary. In: Head, M.J., Gibbard, P.L. (Eds.), 909 Early–Middle Pleistocene Transitions: The Land-Ocean Evidence. Special 910 Publications 247. Geological Society, London, pp. 1–18. 911 Hemmer, H. (1979). Fossil history of living Felidae. Carnivore, 2, 58–61. 912 Hemmer, H. (2001). Die Feliden aus dem Epivillafranchium von Untermaßfeld. Teil 3. In: 913 Kahlke, R.D. (Ed.), Das Pleistozän von Untermaßfeld bei Meiningen (Thüringen). 914 Monographien des Römisch-Germanischen Zentralmuseums Mainz, 40, pp. 699– 915 782. 916 Heptner, V. G., Sludskii, A. A. (1992). Mammals of the Soviet Union, Carnivora (hyaenas 917 and cats), Vol. II, Part 2. E. J. Brill, Leiden, the Netherlands. 918 Iannucci, A., Sardella, R., Strani, F., Mecozzi, B. (2020) Size shifts in late Middle 919 Pleistocene to Early Holocene Sus scrofa (Suidae, Mammalia) from Apulia (southern 920 Italy): ecomorphological adaptations? Hystrix, doi:10.4404/hystrix-00258-2019. 35 921 Iurino, D. A. (2014). Body size reduction and tooth agenesis in Late Pleistocene Meles 922 meles (Carnivora, Mammalia) from Ingarano (Southern Italy). Riv. It. Pal. Strat., 120, 923 109–118. 924 Iurino, D. A., Profico, A., Cherin, M., Veneziano, A., Costeur, L., Sardella, R. (2015). A lynx 925 natural brain endocast from Ingarano (Southern Italy; Late Pleistocene): 926 Taphonomic, Morphometric and Phylogenetic approaches. Hystrix, 26, 110-117. 927 Iurino, D. A., Conti, J., Mecozzi, B., Sardella, R. (2020). Braincase with natural endocast of 928 a juvenile Rhinocerotinae from the late Middle Pleistocene site of Melpignano 929 (Apulia, southern Italy). Front. Earth Sci., 8:94. doi: 10.3389/feart.2020.00094. 930 Johnson, W. E., Godoy, J. A., Palomares, F., Delibes, M., Fernandes, M., Revilla, E., 931 O’Brien, S. J. (2004). Phylogenetic and phylogeographic analysis of Iberian lynx 932 populations. J. Hered., 95, 19–28. 933 Johnson, W. E., Eizirik, E., Pecon-Slattery, J., Murphy, W. J., Antunes, A., Teeling, E., 934 O’Brien, S. J., (2006). The late Miocene radiation of modern Felidae: a genetic 935 assessment. Science, 311, 73–77. 936 Kahlke, R.D. (1999). The History of the Origin, Evolution and Dispersal of the Late 937 Pleistocene Mammuthus-Coelodonta Faunal Complex in Eurasia (Large Mam- mals). 938 Fenske Companies, Rapid City. 939 Kahlke, R.D., García, N., Kostopoulos, D.S., Lacombat, F., Lister, A.M., Mazza, P.P.A., 940 Spassov, N., Titov, V.V. (2011). Western Palaearctic palaeoenvironmental conditions 941 during the Early and early Middle Pleistocene inferred from large mammal 942 communities, and implications for hominin dispersal in Europe. Quat. Sci. Rev., 30, 943 1368–1395. 944 Kazlauskas, J., Matuzevicius A. (1970). On the helminthic fauna of the lynx (Felis lynx) in 945 Lithuania. Acta Parasitol. Lithuanica, 10, 113–115. 946 Koufos, G. (1992). The Pleistocene carnivores of the Mygdonia basin (Macedonia, Greece). 947 Annales de Paléontologie, 78(4), 205-257. 948 Koufos, G., Kostopoulos, D. (1997). New carnivore material from the Plio-Pleistocene of 949 Macedonia (Greece) with the description of a new canid. Münchner 950 Geowissenschaften Abhlungen, 34, 33-63. 951 Kurtén, B. (1957). A note on the systematic and evolutionary relationship of Felis teilhardi 952 Pei. Vertebrata Palasiatica, 1(2), 123-128. 953 Kurtén, B. (1963). Return of a lost structure in the evolution of the felid dentition. Soc. Sci. 954 Fenn. Comment. Biol., 26, 3–11. 36 955 Kurtén, B. (1968). Pleistocene Mammals of Europe. Aldine, Chicago. 956 Kurtén, B. (1978). The Lynx from Etouaires, Lynx issiodorensis (Croizet & Jobert), late 957 Pliocene. Ann. Zool. Fenn., 15, 314–322. 958 Kurtén, B., Granqvist, E. (1987). Fossil pardel lynx (Lynx pardina spelaea Boule) from a 959 cave in southern France. Ann. Zool. Fenn., 24, 39–43. 960 Kurtén, B., Werdelin, L. (1984). The relationships of Lynx shansius Teilhard. Ann. Zool. 961 Fenn., 21, 129–133. 962 Larivière, S., Walton, L.R. (1997). Lynx rufus. Mamm. Species, 563, 1–8. 963 Lavoie M, Renard A, Larivière S. (2019). Lynx canadensis (Carnivora: Felidae). Mamm. 964 Species, 51, 136–154. 965 López-García, J. M., Luzi, E., Berto, C., Peretto, C., Arzarello, M. (2015). Chronological 966 context of the first hominin occurrence in southern Europe: the Allophaiomys ruffoi 967 (Arvicolinae, Rodentia, Mammalia) from Pirro 13 (Pirro Nord, Apulia, southwestern 968 Italy). Quat. Sci. Rev., 107, 260–266. 969 Macdonald, D. W., Loveridge, A. J., Nowell, K. (2010). Dramatis personae: an introduction 970 to wild felids. In: Macdonald, D.W., Loveridge, A.J. (Eds.), Biology and Conservation 971 of Wild Felids. Oxford University Press, New York, pp. 3–58. 972 Magri, D., Palombo, M. R. (2013). Early to middle Pleistocene dynamics of plant and 973 mammal communities in South west Europe. Quat. Int., 288, 63–72. 974 Miracle, P. T. (1991). Carnivore dens or carnivore hunts? A review of Upper Pleistocene 975 mammalian assemblages in Croatia and Slovenia. Rad Hrvatske Akademije Znanosti 976 i Umjetnosti, 458, 193–219. 977 Miracle, P. T. (1995). Broad-spectrum adaptations re-examined: Hunter-Gatherer 978 Responses to Late-Glacial Environmental Changes in the Eastern Adriatic. PhD 979 dissertation, University of Michigan. 980 Miracle, P. T., Lenardić, J. M., Brajković, D. (2010). Last glacial climates, “Refugia”, and 981 faunal change in Southeastern Europe: Mammalian assemblages from Veternica, 982 Velika pećina, and Vindija caves (Croatia). Quat. Int., 212, 137–148. 983 Montoya, P., Alberdi, M. T., Blázquez, A. M., Barbadillo, L. J., Fumanal, M. P., Van der 984 Made, J., Marín, J. M., Molina, A., Morales, J., Murelaga, X., Peñalver, E., Robles, F., 985 Bustos, A. R., Sánchez, A., Sanchiz, B., Soria, D., Szyndlar, Z. (1999). La fauna del 986 Pleistoceno inferior de la Sierra de Quibas (Abanilla, Murcia). Estud. Geol., 55, 127– 987 162. 37 988 Montoya, P., Morales, J., Sendra, J. (2001a). Una mandíbula de Lynx issiodorensis 989 (Croizet y Jobert, 1828) (Carnivora, Mammalia) en el Plioceno inferior de Cuevas de 990 Almanzora (Almería, España). Rev. Esp. Paleontol., núm. ext., 125–131. 991 Montoya, P., Alberdi, M. T., Barbadillo, L. J., van der Made, J., Morales, J., Murelaga, X., 992 Peñalver, E., Robles, F., Ruiz Bustos, A., Sánchez, A., Sanchiz, B., Soria, D., 993 Szyndlar, Z. (2001b). Une faune très diversifiée du Pléistocène inférieur de la Sierra 994 de Quibas (province du Murcia, Espagne). C. R. Acad. Sci. Paris, 332, 387–393. 995 Morales, J., Soria, D., Montoya, P., Pérez, B., Salesa, M. J. (2003). Caracal depereti nov. 996 sp. y Felis aff. silvestris (Felidae, Mammalia) del Pleistoceno Inferior de Layna (Soria, 997 España). Estud. Geol. 59, 229–247. 998 Moullé, P. E., Lacombat, F., Echassoux, A. (2006). Apport des grands mammifères de la 999 grotte du Vallonnet (Roquebrune-Cap-Martin, Alpes-Maritimes, France) à la 1000 connaissance du cadre biochronologique de la seconde moitié du Pléistocène 1001 inférieur d’Europe. L’Anthropologie, 110, 837–849. 1002 Moyà-Solà, S., Agustí, J., Gibert, J., Pons-Moyà, J. (1981). El yacimiento cuaternario de 1003 Venta Micena (España) y su importancia dentro de las asociaciones faunísticas del 1004 Pleistoceno inferior europeo. Paleontol. i Evol., 16, 39–53. 1005 Napoleone, G., Albianelli, A., Azzaroli, A., Bertini, A., Magi, M., Mazzini, M. (2003). 1006 Calibration of the upper Valdarno basin to the plio-pleistocene for correlating the 1007 Apennine continental sequences. Il Quaternario, 16, 131–166. 1008 Nomade, S., Pastre, J. F., Guillou, H., Faure, M., Guérin, C., Delson, E., Debard, E., 1009 Voinchet, P., Messager, E. (2014). 40Ar/39Ar constraints on some French landmark 1010 late Pliocene to early Pleistocene large mammalian paleofaunas: paleoenvironmental 1011 and paleoecological implications. Quat. Geochronol., 21, 2–15. 1012 Ognev, S. I. (1935). Mammals of the U.S.S.R. and adjacent countries. III. Fissipedia and 1013 Pinnepedia. Israel Prog. Sci. Transl., Jerusalem, and Nat. Sci. Found., Washington, 1014 D.C., 641 pp. 1015 Palomares, F. (2001). Vegetation structure and prey abundance requirements of the 1016 Iberian lynx: implications for the design of reserves and corridors. J. Appl. Ecol., 38, 1017 9–18. 1018 Palomares, F., Delibes, M., Ferreras, P., Fedriani, J.M., Calzada, J., Revilla, E. (2000). 1019 Iberian lynx in a fragmented landscape: pre-dispersal, dispersal, and post-dispersal 1020 habitats. Conserv. Biol., 14, 809–818. 38 1021 Palombo, M. R., Sardella, R., Novelli, M. (2008). Carnivora dispersal in Western 1022 Mediterranean during the last 2.6 Ma. Quat. Int., 179, 176–189. 1023 Palombo, M.R., Valli, A.M.F. (2003). Remarks on the biochronology of mammalian faunal 1024 complexes from the Pliocene to the Middle Pleistocene in France. Geol. Romana, 37, 1025 145–163. 1026 Petronio, C., Sardella, R. (1998). Remarks on the stratigraphy and biochronology of the 1027 Late Pleistocene deposit of Ingarano (Apulia, Southern Italy). Riv. It. Pal. Strat.,104, 1028 287–294. 1029 Petronio C., Billia E., Capasso Barbato L., Di Stefano G., Mussi M., Parry S.J., Sardella R., 1030 Voltaggio M. (1996). The Late Pleistocene fauna from Ingarano (Gargano, Italy): 1031 biochronological, palaecological, palaeoethnological and geochronological 1032 implications. Boll. Soc. Pal. It., 34, 333–339. 1033 Petrucci, M., Cipullo, A., Martínez-Navarro, B., Rook, L. (2013). The late Villafranchian 1034 (Early Pleistocene) carnivores (Carnivora, Mammalia) from Pirro Nord (Italy). 1035 Palaeontogr. Abt. A, 298, 113–145. 1036 R Core Team (2019). R: A Language and Environment for Statistical Computing. R 1037 Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org. 1038 Rodríguez, A., Calzada, J. (2015). Lynx pardinus. The IUCN Red List of Threatened 1039 Species 2015: e.T12520A50655794. doi: 10.2305/IUCN.UK.2015- 1040 2.RLTS.T12520A50655794.en. 1041 Rodríguez, R., Ramírez, O., Valdiosera, C.E., García, N., Alda, F., Madurell- Malapeira, J., 1042 Marmi, J., Doadrio, I., Willerslev, E., Götherström, A., Arsuaga, J.L., Thomas, M.G., 1043 Lalueza-Fox, C., Dalén, L. (2011). 50,000 years of genetic uniformity in the critically 1044 endangered Iberian lynx. Mol. Ecol., 20, 3785–3795. 1045 Rodríguez-Hidalgo, A., Sanz, M., Daura, J., Sánchez-Marco, A. (2020). Taphonomic 1046 criteria for identifying Iberian lynx dens in quaternary deposits. Sci. Rep., 10, 7225. 1047 doi: 10.1038/s41598-020-63908-6. 1048 Rodríguez-Varela, R., Tagliacozzo, A., Ureña, I., García, N., Crégut-Bonnoure, E., 1049 Mannino, M. A., Arsuaga, J.L., Valdiosera, C. (2015). Ancient DNA evidence of 1050 Iberian lynx palaeoendemism. Quat. Sci. Rev., 112, 172–180. 1051 Rodríguez-Varela, R., García, N., Nores, C., Álvarez-Lao, D., Barnett, R., Arsuaga, J. L., 1052 Valdiosera, C. (2016). Ancient DNA reveals past existence of Eurasian lynx in Spain. 1053 J. Zool., 298, 94–102. 39 1054 Rustioni, M., Sardella, R., Rook, L. (1995). Note sulla distribuzione e sulla tassonomia del 1055 genere Lynx in Italia. Padusa, 1, 359–364. 1056 Sardella, R., Bertè, D. F., Iurino, D. A., Cherin, M., Tagliacozzo, A. (2014). The wolf from 1057 Grotta Romanelli (Apulia, Italy) and its implications in the evolutionary history of 1058 Canis lupus in the Late Pleistocene of Southern Italy. Quat. Int., 328–329, 179–195. 1059 Sardella, R., Bellucci, L., Bona, F., Cherin, M., Iurino, D.A., Rook, L. (2018). Before and 1060 after the earliest Homo dispersal in Europe: Evidence from the early Pleistocene 1061 sites of the Italian Peninsula. CR Palevol, 17(4-5), pp.287-295. 1062 Simón, M. A, editor. (2012). Ten years conserving the Iberian lynx. Consejería de 1063 Agricoltura, Pesca y Medio Ambiente. Sevilla: Junta de Andalucía. 1064 Simón, M. A., Gil-Sánchez, J. M., Ruiz, G., Garrote, G., McCain, E. B., Fernández, L., 1065 López-Parra, M., Rojas, E., Arenas-Rojas, R., Rey, T. D., García-Tardío, M., López, 1066 G. (2012). Reverse of the decline of the endangered Iberian Lynx. Conserv. Biol., 26, 1067 731–736. 1068 Sommer, R.S., Benecke, N. (2006) Late Pleistocene and Holocene development of the 1069 felid fauna (Felidae) of Europe: a review. J. Zool., 269, 7–19. 1070 Strani, F., DeMiguel, D., Alba, D.M., Moyà-Solà, S., Bellucci, L., Sardella, R., Madurell- 1071 Malapeira, J. (2019). The effects of the “0.9 Ma event” on the Mediterranean 1072 ecosystems during the Early-Middle Pleistocene transition as revealed by dental 1073 wear patterns of fossil ungulates. Quat. Sci. Rev., 210, 80–89. 1074 Sunquist, M., Sunquist, F. (2002). Wild Cats of the World. Chicago University Press, 1075 Chicago. 1076 Testu, A. (2006). Etude paléontologique et biostratigraphique des Felidae et Hyaenidae 1077 pléistocènes de l’Europe méditerranéenne. Phd Dissertation, Université de 1078 Perpignan, pp 317. 1079 Tumlison, R. (1987). Felis lynx linnaeus, 1758. Mamm. Species, 269, 1–8. 1080 Van den Brink, F. H. (1971). Le lynx pardelle en France. Bulletin de la Société d'Étude des 1081 Sciences naturelles de Nîmes, 51, 109–117. 1082 Van Valkenburgh, B. (1990). Skeletal and dental predictors of body mass in carnivores. In: 1083 Damuth, J., MacFadden, B.J. (Eds.), Body Size in Mammalian Paleobiology: 1084 Estimation and Biological Implications. Cambridge University Press, Cambridge etc, 1085 pp. 181–205. 40 1086 Vigne, J. D. (1996). La faune des vertébrés. Le Laouret et la Montagne d’Alaric à la fin de 1087 l’ȃge du Bronze. Un hameau abandonné entre Floure et Monze (Aude) (éd Gasco J), 1088 pp. 197–239. 1089 Vigne, J. D., Pascal, M. (2003). Le Lynx pardelle Lynx pardinus (Temnick, 1827). Evolution 1090 de la Faune de Vertébrés en France Métropolitaine depuis le début de l’Holocéne. 1091 Invasions et Extinction (Eds.) Pascal, M., Lorvelec, O., Vigne, J. D., Keith, P. & 1092 Clergeau, P.), pp. 87–89. 1093 Viret, J. (1954). Le loess à bancs durcis de Saint-Vallier (Drȏme) et sa faune de 1094 mammifères villafranchiens. Nouv. Arch. Mus. Hist. Nat. Lyon 4, 1–200. 1095 von den Driesch, A. (1976) A guide to the measurement of animal bones from 1096 archaeological sites. Peabody Mus Bull, 1, 1–137. 1097 Voelcker, J. (1930). Felis issiodorensis Croizet von Mauer a. d. Elsenz. Sitzungs ber. 1098 Heidelberger Akad. Wiss., Math.–Nat. wiss. Kl.Jhg., 12, 1-9. 1099 Werdelin, L. (1981). The evolution of lynxes. Ann. Zool. Fenn., 18, 37–71. 1100 Werdelin, L. (1987). Supernumerary teeth in Lynx lynx and the irreversibility of evolution. J. 1101 Zool., 211, 259–266. 1102 Wiig, Ø., Andersen, T. (1986). Sexual size dimorphism in the skull of Norwegian lynx. Acta 1103 Theriol, 31(12), pp.147-394. 1104 1105 1106 Figure captions 1107 1108 Figure 1. Location of the paleontological site of Ingarano (Foggia, Italy). 1109 1110 Figure 2. Crania of Lynx pardinus from Ingarano (Italy). MGPT-PU 135415 (A1-A3), 1111 ING75 (B1-B3), ING76 (C1-C3), M693 (D1-D3). Crania figured in dorsal (1), ventral (2) 1112 and lateral (3) views. Scale bar 3 cm. 1113 1114 Figure 3. 3D model of the complete skull of Lynx pardinus MGPT-PU 135415 from 1115 Ingarano (Italy), in anterior (A), posterior (B), right lateral (C) and left lateral (D) views. 1116 1117 Figure 4. Hemimandibles of Lynx pardinus from Ingarano (Italy). Hemimandibles 1118 associated with the cranium MGPT-PU 135415 (A1-A4) figured in labial (1, 3) and buccal 41 1119 (2, 4) views. Right hemimandibles: ING68 (B1-B2), ING69 (C1-C2), INGND870 (D1-D2), 1120 IN15NS (E1-E2), INGND869 (F1-F2), ING72 (G1-G2), ING73 (H1-H2). Left 1121 hemimandibles: INGND312 (I1-I2), ING61 (J1-J2), INGND868 (K1-K2), INGND455 (L1- 1122 L2), INGND144 (M1-M2). Hemimandibles figured in lingual (1) and buccal (2) views. Scale 1123 bar 3 cm. 1124 1125 Figure 5. First lower molars (M1) of Lynx pardinus from Ingarano (Italy). Right M1: ING68 1126 (A1-A2), ING69 (B1-B2), INGND870 (C1-C2), IN15NS (D1-D2), INGND869 (E1-E2), 1127 ING72 (F1-F2), ING73 (G1-G2), MGPUTV35415 (H1-H2), ING63 (I1-I2), ING70 (J1-J2), 1128 ING79 (K1-K2), INGND369 (L1-L2), INGND372 (M1-M2), INGNDF318 (N1-N2). Left M1: 1129 INGND312 (O1-O2), ING61 (P1-P2), INGND868 (Q1-Q2); INGND319 (R1-R2), INGND867 1130 (S1-S2), INGND145 (T1-T2), ING62 (U1-U2), INGND144 (V1-V2), INGND373 (W1-W2), 1131 ING71 (X1-X2), FG107 (Y1-Y2), MGPT-PU V35415 (Z1-Z2), FG109 (AA1-AA2). Teeth 1132 figured in lingual (1) and buccal (2) views. Scale bar 1 cm. 1133 1134 Figure 6. Comparative cranial morphology of Lynx pardinus from Ingarano (MGPT-PU 1135 135415, A1-A3; ING75, B1-B3; M693, C1-C2); L. pardinus from Grotta del Principe (D1- 1136 D2); L. pardinus from Grotte de l’Observatoire (E1-E3); extant L. pardinus MNB 8865 (F1- 1137 F3); extant Lynx lynx MC 85 (G1-G3); Lynx issiodorensis SABAP_UMB 337653 from 1138 Pantalla (H1-H3); L. issiodorensis MNB Ol 1 from Olivola (I1-I3). The white arrows indicate 1139 the temporal ridges, the black arrows indicate the sagittal crest, and the zoom-boxes show 1140 the arrangement of the jugular and hypoglossal foramina. The crania are normalized. 1141 1142 Figure 7. Schematic representation of the diagnostic characters of Lynx issiodorensis, 1143 extant Lynx lynx, Lynx pardinus from Ingarano (as a representative of fossil L. pardinus), 1144 and extant L. pardinus. A, Length of carnassial teeth; B, Frequency of diagnostic 1145 characters of carnassial teeth in the analyzed samples; C, Shape of the temporal ridges 1146 and sagittal crest; D, Average body mass; E, Arrangement of the jugular and hypoglossal 1147 foramina. 1148 1149 Figure 8. Estimated body masses (kg) for Lynx issiodorensis, Lynx sp., Lynx pardinus s.l., 1150 and Lynx lynx from various Plio-Pleistocene sites of Europe. Estimates are based on 1151 condylobasal length, following Van Valkenburgh (1990). The body mass range for extant 1152 species is shown on the right. Details are in Supplementary table 8. 42 1153 1154 Figure 9. Map of the Plio-Pleistocene records of European lynxes. Early Pliocene: 1 1155 Almanzora, 2 La Gloria 4, 3 La Calera; Late Pliocene: 4 Serrat d’en Vaquer, 5 Les 1156 Etouaires, 6 La Côte d’Ardé, 7 Pardines, 8 Garfagnana, 9 Layna, 10 Çalta; Early 1157 Pleistocene: 11 Olteţ River Valley, 12 La Puebla de Valverde, 13 Fonelas P-1, 14 1158 Villarroya, 15 Saint Vallier, 16 Olivola, 17 Pantalla, 18 Upper Valdarno, 19 Figline, 20 1159 Matassino, 21 Pirro Nord, 22 Untermassfeld, 23 Apollonia, 24 Tourkovounia, 25 1160 Argentario, 26 Avenc Marcel, 27 Sierra de Quibas, 28 Cueva Victoria, 29 Cal Guardiola, 1161 30 Vallparadís Estació, 31 Vallonnet, 32 Somssich Hill, 33 Ubeidiya, 34 Venta Micena, 35 1162 Trinchera Dolina 6, 36 Sainzelles, 37 Soleilhac, 38 Grotte de la Sartenette; Middle 1163 Pleistocene: 39 Trinchera Galeria, 40 Sima de Los Huesos, 41 Aldene - Lower levels, 42 1164 Caune de l’Arago, 43 L'Escale, 44 Lazaret, 45 Lunel-Viel, 46 Mosbach, 47 Mauer, 48 1165 Ponte Molle, 49 Valdemino - lower levels, 50 Villacastín; Late Pleistocene: 51 Abri 1166 Cornille, 52 Aldene - upper level, 53 Baume-Longue, 54 Campfiel, 55 Grotte Balauzìere, 1167 56 Grotte de l'Observatoire, 57 Grotta de Reihac, 58 Grotte Vaufrey, 59 Hortus, 60 La 1168 Crouzade, 61 La Salpetriére, 62 Moula Guercy-Baume, 63 Orgnac 3, 64 Portel Ouest, 65 1169 Puech-Margal, 66 Grotta del Principe, 67 Grotta della Masseria del Monte, 68 Monte 1170 Tignoso, 69 Valle Radice, 70 Abric Romaní, 71 Almada, 72 Cau d’en Borràs, 73 Cau del 1171 Duc, 74 Cova de l’Arbreda, 75 Cova Negra, 76 Cova Toll, 77 Cova Toixoneres, 78 Cueva 1172 de Chaves, 79 Cueva del Puerto, 80 Devil's Tower, 81 El Escarche, 82 El Muscle, 83 1173 Ermittia, 84 Erralla, 85 Gorham's Cave, 86 Pena de Estebanvela, 87 Turó del Moro, 88 1174 Windmill Hill, 89 Alargo do Casais, 90 Caldeirao, 91 Casa do Moira, 92 Columbeira, 93 1175 Escoural, 94 Furinha, 95 Prado des Salemas, 96 Arene Candide, 97 Buca della Iena, 98 1176 Grotta della Madonna, 99 Grotta delle Striare, 100 Grotta dei Colombi, 101 Grotta di Equi, 1177 102 Grotta Polesini, 103 Grotta Sant’Agostino, 104 Grotta Tina di Camerota, 105 1178 Melpignano, 106 Riparo Fumane, 107 San Sidero, 108 Col d’Aran, 109 Groutte de Pène, 1179 110 Mostayous, 111 Réseau du Cébéri, 112 Willendorf, 113 Beeston Tor, 114 Lynx Cave, 1180 115 Neale's Cave, 116 Sewell's Cave, 117 Cueva de Los Casares, 118 Pagolusieta, 119 1181 Cueva de Santimamine, 120 Kostenki 21, 121 Grotte du Bourrouilla, 122 Valdemino - 1182 upper level, 123 Ingarano. 1183 1184 Figure 10. Reconstruction sequence of the head appearance of Lynx pardinus MGPT-PU 1185 135415 from Ingarano (Italy). Artwork by D.A. Iurino. 1186 43 1187 Table 1. Measurements of the Ingarano sample (see also Supplementary tables 1-3). 44