This is the accepted version of the article: Bartolini Lucenti, Saverio; Madurell Malapeira, Joan; Martínez Navarro, Bien- venido; [et al.]. «A comparative study of the Early Pleistocene carnivore guild from Dmanisi (Georgia)». Journal of Human Evolution, Vol. 162 (January 2022), art.103108. DOI 10.1016/j.jhevol.2021.103108 This version is available at https://ddd.uab.cat/record/250226 under the terms of the license 1 A comparative study of the Early Pleistocene carnivore guild from Dmanisi (Georgia) 2 3 Abstract 4 The carnivore guild of the Early Pleistocene site of Dmanisi is among the most diverse of the Early 5 Pleistocene of the entire Old World. It includes 14 carnivoran taxa: Homotherium latidens, 6 Megantereon whitei, Panthera onca georgica, Acinonyx pardinensis, Lynx issiodorensis; 7 Pachycrocuta brevirostris; Canis (Xenocyon) lycaonoides, Canis borjgali, Vulpes alopecoides; 8 Ursus etruscus; Lutra sp., Martes sp., Meles sp. and Pannonictis sp. The analysis of this rich 9 carnivore guild was carried out under different methodological approachesin order to compare the 10 assemblage with other chronological coeval European, Asian and African sites from a 11 paleobiological perspective. To achieve the goal we used a permutational hierarchical method 12 called boostrapping cluster analysis based on taxonomic absence/presence matrices (at both generic 13 and specific level) and on ecological matrices (considering dietary preferences/hunting strategies of 14 each carnivoran) and carried out Mantels tests assessing magnitude of time, space, ecology and 15 taxonomy as source of difference between guilds. Our results suggest a close similarity among the 16 Dmanisi carnivore assemblage and other guilds recorded from European late Villafranchian sites 17 such as Pirro Nord, Venta Micena and Apollonia 1 and, in a lesser extent, to European 18 Epivillafranchian sites as Vallonnet, Untermassfeld or the Vallparadís Section. Early to Middle 19 Pleistocene Asian carnivore assemblages display several similarities with the Dmanisi guild mainly 20 in the record and diversity of felid and the canid ecomorphotypes. Eastern African sites such as 21 Olduvai and Omo, as well as South African sites display a lower similarity with the studied sample, 22 basically for the most diverse hyenid taphocoenoses. To sum up, the present study suggests a close 23 similarity between the Dmanisi carnivore guild and other European late Early Pleistocene 24 assemblages without close parallels with African or Asian assemblages. 25 26 Keywords: Dmanisi; Carnivorans; Early Pleistocene; Quaternary; Eurasia; Africa 1 27 2 28 29 1. Introduction 30 The Georgian archeo-paleontological site of Dmanisi is located in the Mashavera river valley, 31 the Lesser Caucasus, 85 km southwest of Tbilisi, (44° 210E, and 41° 90N). The whole stratigraphic 32 sequence of the fossil assemblage is chronologically constrained to 1.85–1.76 Ma (Gabunia et al., 33 2000; Vekua et al., 2002; Lordkipanidze et al., 2005, 2007; Ferring et al., 2011). 34 Dmanisi has one of the most extraordinary assemblage of fossil large mammals (including 35 hominins) of the late Early Pleistocene of Eurasia, coupled with an abundant record of Oldowan 36 tools (Lordkipanidze et al., 2005, 2007). This allows the study of potential interactions of the early 37 Homo erectus out of Africa in association with the other large mammal species, and to infer 38 paleoenvironmental details. The Dmanisi taphocoenosis is remarkably rich and diverse, with several 39 species that the hominins probably survived on. The favorable conditions recorded at Dmanisi, and 40 in general in the Pontocaspian region (Blain et al., 2104; Krijgsman et al., 2019) at the end 41 Gelasian–Calabrian transition, provided the optimal ecological conditions of the middle latitudes for 42 the faunal dispersal across Eurasia and Africa (Vekua, 1995; Lordkipanidze et al., 2007; Martínez- 43 Navarro, 2010; Espigares et al., 2013; Martínez-Navarro et al., 2014a; Capellini et al., 2019; Medin 44 et al., 2019; Bartolini-Lucenti et al., 2020). 45 The fossil record of Dmanisi includes 26 large mammal species, represented by a combination of 46 species evolved in Europe during the Villafranchian, including the elephant Mammuthus 47 meridionalis, or the large omnivorous carnivore Ursus etruscus (Medin et al., 2019) in association 48 with other taxa (mainly ungulates: bovids, cervids, equids, or rhinocerotids) that originated in Asia 49 (e.g., Bison georgicus, Gallogoral meneghinii, Praemegaceros obscurus), but also including 50 carnivorans, e.g. Canis (Xenocyon) lycaonoides or Canis borjgali (Bartolini-Lucenti et al., 2020; 51 2021). The Dmanisi fauna also includes taxa of African origin, e.g., Megantereon whitei which 52 dispersed into Europe at the beginning of the late Villafranchian (Vekua, 1995; Lordkipanidze et 53 al., 2007; Martínez-Navarro, 2010; Rook and Martínez-Navarro, 2010). 3 54 Fossil carnivore guilds, as all fossil associations, are affected by caveats of taphonomy (e.g., 55 time-averaging, accumulation agent, direct or indirect selection, preferred fossilization; etc.) that 56 might limit their record as opposed to their original biocoenoses. Moreover, bias arising from 57 taxonomic attribution should not be ruled out. This considered, here we offer a comparative analysis 58 of the carnivore guild of Dmanisi with those from other Early Pleistocene localities across the Old 59 World in order to detect similarities and differences among them and to infer paleobiogeographic, 60 paleoecological and biochronological implications. 61 62 2. Materials and methods 63 To assess the degree of similarity among Dmanisi carnivore guild and other guilds of different 64 localities of the Old World, we performed a cluster analysis on the generic and specific composition 65 matrices of these assemblages (Table 1 and Supplementary Online Material [SOM] Table S1), 66 calculating the Jaccard’s index (Jaccard, 1912) to further compare and estimate their similarity. The 67 thirty-seven localities selected to undertake the comparisons, are distributed across the Old World 68 (see Fig. 1) and span between the latest Pliocene and the beginning of the early Middle Pleistocene 69 (2.7–0.7 Ma). We tried to choose localities with accepted chronologies and with a consistent 70 number of carnivorans recorded. The compiled matrices of occurrences are based on published data 71 (Teilhard de Chardin, 1940; De Giuli and Masini, 1986; Arribas and Palmqvist, 1998; Sotnikova et 72 al., 2002; Palombo and Valli, 2003; Gaudzinski, 2004; Martinez-Navarro, 2004; Mazza et al., 2004; 73 Qiu et al., 2004; Delson et al., 2006; Lordkipanidze et al., 2007; O’Regan, 2007; Jin and Liu, 2009; 74 Martínez-Navarro et al., 2009; Sotnikova and Titov, 2009; Geraads et al., 2010; Rook and 75 Martínez-Navarro, 2010; Werdelin and Sanders, 2010; Geraads, 2011; 2016; Kahlke et al., 2011; 76 Petrucci et al., 2013; Rook et al., 2013; Wang et al., 2013; Madurell-Malapeira et al., 2010; 2012; 77 2014; Bartolini-Lucenti and Rook, 2016; 2018; Adams et al., 2016; Fourvel et al., 2017; Gkeme et 78 al., 2017; O’Regan and Steininger, 2017; Koufos, 2018; Sahnouni et al., 2018; Cohen et al., 2019; 79 Bartolini-Lucenti and Madurell-Malapeira, 2020; Bartolini-Lucenti et al., 2020; Jin et al., 2021) in 4 80 most cases revised and updated. The cluster analyses were carried out by applying a bootstrapping 81 cluster analysis (hereafter BCA). Bootstrapping cluster analysis is a partitioning methodology that 82 is useful to test the robustness of the clustering results. It was used in recent papers (Raia et al., 83 2009 and Carotenuto et al., 2016) to statistically identify Pleistocene communities of Eurasian large 84 mammals. This method performs an initial clusterization of the dataset by means of the unweighted 85 pair group method with arithmetic mean (UPGMA) algorithm (Sokal, 1958). In this way the 86 analysis generates a reference clustering of the selected localities on the basis of their taxonomic 87 similarity which can putatively have biochronological, taphonomical and paleoecological 88 implications. This reference clustering is the used to assess probability of each branching in the 89 reference via a permutational approach. Subsequently, a sample of the original data is randomly 90 chosen and used to obtain a new UPGMA cluster analysis, which yield a new clustering. This 91 operation is repeated a number of times. The obtained new clusterization are compared with the 92 reference one. Asimilarity index (G*) is computed between the reference and the sample clustering. 93 This index can range between 0 (if the two clusteringzations are totally different) and 1 (if the 94 clusters generated with the original and sampled data coincides). The G* is then compared with the 95 expected similarity value between reference and sample-based clustering (G°) under the null 96 hypothesis that the sampled dataset is a truly random sample of the original data. We repeated 1000 97 times the procedure of resampling the original dataset, performing as many the UPGMAs on 98 sampled data and computing similarity indices. In the end, if the probability that G* is higher or 99 equal to G° is higher than the significant level (P[G° ≤ G*]; α = 0.05), the partitioning levels of the 100 reference cluster analysis are sharp (see Pillar, 1999 for a detailed explanation of the algorithm). We 101 performed BCA by using the R package ‘pvclust’ v. 2.2-0 (Suzuki et al., 2019). 102 We also compared ecological divisions in the analyzed guild grouping carnivorans according to 103 the classic diet categories: hypocarnivores (less than 50% vertebrate meat in the diet), 104 mesocarnivores (between 50–70% of vertebrate meat in the diet) and hypercarnivores (>70% 105 vertebrate meat in the diet; Crusafont-Pairó and Truyols-Santonja, 1956; Van Valkenburgh, 1989). 5 106 To these, we added an additional class of diet for insectivores (taxa specialized to insect feeding). 107 Finally, to better characterize hypercarnivorous spectrum of adaptations we included hunting 108 strategies (e.g., ambush or cursorial predation) and dietary habits (e.g., piscivorous or bone-cracking 109 carnivorans), following the ecological categories proposed in literature (e.g., Van Valkenburgh, 110 1989; Van Valkenburgh and Koepfli, 1993; Popowics, 2003; Werdelin and Peigné, 2010; Ester, 111 2012; Flower and Schreve, 2014; Medin et al., 2019). In this ecological analysis of the carnivore 112 guild of Dmanisi, we decided to ordinate the relative abundances of the diet classes of each locality 113 according to the order provided by a BCA on the ecological composition of each site. 114 Eventually, we tested whether the differences emerged from the previous analyses were 115 statistically related to temporal and/or geographical factors. To this aim, we performed a Mantel test 116 to asses any relationship between the taxonomical and the ecological distance matrices as used 117 before in comparison to the geographic and age distance matrices measured between all the fossil 118 localities here studied. More specifically, the kind of analysis we performed is able to test the 119 relationship between a response matrix and a predictor matrix while taking into account the effect 120 of other predictor matrices by means of permutations. Hence, we first analyzed the correlation 121 between taxonomic and ecological distance while taking into account time and geographic 122 distances. Then deepened the analysis taking into account taxonomy and ecological preference 123 singularly. We tested the relationship between the taxonomic distance and the geographic distance 124 by accounting for the effect of the temporal distance and, then, we repeated the same analyses, this 125 time by accounting for the effect of geographic distance on the relationship between taxonomic and 126 temporal distance matrices. We proceeded in the same way with the ecological distance matrix. In 127 addition, since we expected that the African sites would have accrued the contribution of geography 128 on determining taxonomical distances between fossil sites, we repeated the mantel tests this time by 129 using Eurasian fossil sites in order to test effect of both space and time in assessing the taxonomical 130 and ecological proximity of Dmanisi to the other Eurasian sites. We carried out the Mantel tests 131 using the R package ‘ecodist’ v. 2.0.1 (Goslee and Urban, 2007). 6 132 133 3. Results 134 Dmanisi carnivore guild is among the most diverse ones of the contemporary Old-World 135 assemblages analyzed in the present work (Table 1 and SOM Table S1). The carnivore guild from 136 Dmanisi is composed of five species of Felidae, i.e., Homotherium latidens, Megantereon whitei, 137 Panthera onca georgica, Acinonyx pardinensis, Lynx issiodorensis; one Hyaenidae (Pachycrocuta 138 brevirostris); three species of Canidae, i.e., Canis (Xenocyon) lycaonoides, Canis borjgali, Vulpes 139 alopecoides; one Ursidae (Ursus etruscus); and four Mustelidae, i.e., unstudied species of the 140 genera Lutra, Martes, Meles and Pannonictis (Fig. 2; Vekua, 1995; Lordkipanidze et al., 2007; 141 Hemmer et al., 2010; Medin et al., 2019; Bartolini-Lucenti and Madurell-Malapeira, 2020; 142 Bartolini-Lucenti et al., 2020, 2021). Compared to selected almost coeval localities from Africa and 143 East Asia (Fig. 2), the carnivoran association at Dmanisi is characterized by a lower diversity of 144 Hyaenidae (only a single species contrary to the two species of Olivola and Jinyuan Cave Lower 145 Fauna and the four of Kromdraai Member 2) but a considerably higher diversity of Mustelidae. 146 Felidae is instead the most abundant family in all the considered localities, with the exception of 147 Kromdraai Member 2, with the remarkable absence of the genus Lynx in Africa, a genus very 148 common in Eurasian localities. Additionally, homotherinii and smilodontinii felids are normally 149 recorded in Eurasian localities being metailurinii, the genus Panthera and Acinonyx more common 150 in African ones. Ursidae are very abundant in Eurasian localities and also present Northafrican 151 ones, but it is absent in subsaharian localities. Canidae diversity in Dmanisi is similar to those from 152 Asia and Europe, rather than to the African one. Peculiarity exclusive of the African site is the 153 presence of Hespertidae and Viverridae in the assemblage, unlike the other localities. A more 154 extensive analysis of the guild composition of Dmanisi is here undertaken. 155 156 3.1. The Dmanisi carnivore guild structure: Comparison with European, Asian and African guilds 7 157 To assess the degree of similarity and/or differences between Old World localities we performed 158 a BCA on taxonomic matrices of absence/presence. The results of the BCA on the genera of the 159 carnivore guilds of numerous localities across the Old World are reported in the SOM S1 (SOM 160 Figures S1–S2, whereas that performed on the species are shown in Figures 3 and 4. The two 161 different analyses show similar patterns (i.e., distinction between African and Eurasian localities, 162 distintion between older and younger guilds, the position of Dmanisi). Of the two, the analysis on 163 the matrix of absence/presence based on species is, as expected, more resolved (Fig. 3), although its 164 Jaccard indices are in general lower than those of genera (see Fig. 4, SOM Fig. S2 and SOM Tables 165 S2–S3). Similarly to the genera-based analysis (SOM Fig. S1), the dendrogram of the BCA on the 166 species matrix shows two distinct groups: one composed only of African localities (hereafter the 167 ‘African cluster’; identified at node A1) and a second one is made up of the Eurasian ones 168 (hereafter the ‘Eurasian cluster’, sprouting from node B1). These two groups are well supported by 169 high values of percentage p-values (both 98%) and evidently separated also on the basis of Jaccard 170 similarity indices (Fig. 4). In the African cluster, the pattern of similarities yielded in the genera- 171 based hierarchical analysis (SOM Fig. S1) or in the respective Jaccard indices (SOM Fig. S2) are 172 more or less confirmed. For instance, the grouping of the South African and of the North African 173 sites, of the members of the Shungura Formation of Omo (Fig. 3). The Jaccard values of Omo 174 members are the highest of the African cluster (Fig. 4) and much higher in comparison to those of 175 SOM Figure S2, testifying to a considerably higher similarity when species are taken into 176 consideration (Fig. 4). In contrast to the genera-based hierarchical analysis, the site of Ahl al 177 Oughlam is rooted at the base of the African cluster and groups with the other North African site of 178 Tighennif. Both these sites have diverse and numerous peculiar taxa, supporting the distance in 179 comparison to other African sites. This is confirmed also by their Jaccard indices (Fig. 4). For 180 instance, in Tighennif there is the first record of Lyncodontini in the African continent (Pannonictis 181 hoffstetteri) whereas some taxa recorded in Ahl al Oughlam show affinity with Eurasian species like 182 the raccoon dog Nyctereutes abdeslami, the scimitar-toothed cat Homotherium latidens, Lynx gr. 8 183 issiodorensis and Ursus etruscus, although most of the species is African related. After Ahl al 184 Oughlam-Tighennif group, the following group to branch out (node A6) is that of the South African 185 localities. Of these four, Member 2 of Kromdraai is the first to separated from the others. This is 186 possibly due to the occurrence of taxa like Aonyx capensis, Civettictis civetta, Mellivora capensis, 187 Propoecilogale bolti and primitive wild of the Canis (Xenocyon) falconeri group, absent in the 188 other sites. The same can be said for Sterkfontein Member 4, which shares more affinity with 189 Drimolen and Coopers D, but also possess exclusive taxa absent in the latter sites, e.g., 190 Pachycrocuta and Canis brevirostris. The support to node A8 is among the highest of the analysis 191 (percentage p-value = 98%). Despite their clear similarity and fairly numerous shared species (eight 192 taxa), Drimolen and Coopers possess several carnivorans that are present in one site but absent in 193 the other one. For instance, Drimolen lacks Acinonyx jubatus, Herpestes ichneumon, Lupulella 194 mesomelas, Panthera leo, but on the other side it records Dinofelis barlowi, Lycaenops silberbergi, 195 Vulpes chama. These differences are testified to also by the rather low Jaccard indices. The last 196 group in the African cluster is that composed of East African localites (node A3; Fig. 3). As 197 discussed before, the members of Shungura Formation of Omo form a well-supported and identified 198 subgroup (node A12). The Tanzanian Olduvai beds and Kenyan Koobi Fora members are arranged 199 together, apparently not strictily following a chronological order. They are joined by some elements 200 like the felids Panthera leo and Panthera pardus and the hyenid Crocuta crocuta and Hyaena 201 hyaena and the viverrid Pseudocivetta ingens, which are present in almost every site. Despite these 202 shared taxa, each site shows a number of species that differ from the other. This limited similarity is 203 reflected in the Jaccard values, which are high only for Olduvai Bed II and KBS member (that share 204 5 taxa out of eight, Olduvai Bed II, and of six, KBS, respectively). 205 The opposite group is entirely composed of Eurasian localities (Fig. 3). This Eurasian cluster is 206 further subdivided in two groups: one comprises all Chinese localities (node B2), whereas European 207 localities fall together (node B3). In turn, those groups, the Chinese and the European clusters, are 208 made up by two subclusters. The first of the two Chinese subclusters (node B4) is composed of 9 209 three localities close in age to one another, the upper levels of Jinyuan Cave, Zhoukoudian Loc. 1 210 and Gongwangling. The closer affinity between the latter two sites (node B5), is due to four shared 211 species although both record several taxa absent in the other (six of GWL absent in ZKD1 and five 212 of ZDK1 not present in GWL) as testified to by the fairly low percentage of the p-value (= 58%). 213 The second group of the Chinese cluster includes the oldest Asian localities on the analysis and it is 214 fairly well supported (percentage p-value = 92%). The first to branch out is Zhoukoudian Loc. 18. 215 The limited record of this site makes difficult to narrow down its affinity to other localities. The 216 latter are grouped in two small clusters, both well supported (node B12 = 95% and node B14 = 217 99%). Longguopo and and Renzidong form a group sharing almost all their carnivorans. The other 218 group is composed of three sites. Jinyuan Cave Lower Fauna lies at the base of the subcluster and is 219 distinguished from Longdan and Nihewan (Xiashagou), which cluster together (Fig. 3). The Jaccard 220 indices are high for the three sites (> 0.60; Fig. 4), testifying to their affinities. Of the last three 221 sites, Jinyuan Cave Lower Fauna is distinct from Longdan and Nihewan for the presence of Martes 222 andressoni. Moreover, there are some taxa shared by Jinyuan Cave Lower Fauna and Longdan and 223 others present only in Nihewan and Jinyuan Cave Lower Fauna. With Longdan, Jinyuan Cave 224 Lower Fauna shares Panthera leo and Canis (Xenocyon) dubius. With Nihewan these shared taxa 225 include Nyctereutes sinensis and Ursus gr. etruscus. Longdan and Nihewan share thirteen taxa and 226 have a high Jaccard similarity index (Fig. 4). The differences, in addition to those already discussed 227 above, Nihewan uniquely records Acinonyx pardinensis, Canis (Xenocyon) gr. lycaonoides, Eucyon 228 minor, Lutra licenti and Ursus etruscus, whereas Longdan records the presence of Canis 229 (Xenocyon) dubius and Ursus gr. thibetanus. 230 Opposite to this Chinese cluster, there is a large well-supported European cluster (node B3; Fig. 231 3), which is again divided in two subclusters (nodes B7 and B11). The smaller one (node B7), in 232 terms of number of sites included in it, is composed of six localities in two subgroups (nodes B9 233 and B10). This group includes localities dated between ca 2.5 and 1.7 Ma. The Casa Frata lies at the 234 base of the cluster composed of the Italian localities of Poggio Rosso and Olivola (Fig. 3), in a 10 235 group of geographically and spatially close localities. Jaccard similarity index for Casa Frata is 236 fairly higher of Olivola and Poggio Rosso (Fig. 4) because it lacks several elements of the latter 237 sites (e.g., Canis etruscus, Chasmaporthetes lunensis, and Felis sylvestris), while possessing other 238 taxa like Canis (Xenocyon) falconeri and Vulpes alopecoides absent in the latter sites. Poggio Rosso 239 and Olivola have high Jaccard indices (the highest of the group of node B7), thanks for the fourteen 240 shared taxa. The second small subgroup (node B9) shows Khapry Faunal Complex at the base of 241 the grouping of the French sites of St. Vallier and Senèze (node B13). Three sites share iconic 242 Gelasian species like derived Nyctereutes gr. megamastoides and Pliocrocuta perrieri, as well as 243 others e.g., Acinonyx pardinensis and Homotherium latidens. The similarity between Senèze and St. 244 Vallier is undoubtful, although at a lower degree compared to the group of Olivola and Poggio 245 Rosso, as testified to by the Jaccard indices (Fig. 4). The second subcluster is the one that includes 246 also Dmanisi, together with all younger sites (node B11; Fig. 3). At its base stems ‘Ubeidiya. Its 247 guild shares with the other late Early Pleistocene sites the presence of some Eurasian elements like 248 Canis mosbachensis, C. (Xenocyon) lycaonoides, V. alopecoides, Martellictis ardea and U. 249 etruscus. On the other side, ‘Ubeidiya guild is characterized by the occurrence of African taxa like 250 Crocuta crocuta and Vormela peregusna. Within the subcluster, there are two groups formed 251 respectively by late Villafranchian (node B19) and by Epivillafranchian sites (node B18). 252 Interestingly, these groups have the highest Jaccard index of all (Fig. 4). Within the late 253 Villafranchian cluster, Pirro Nord lies at the base. Pirro Nord is the most diverse guild of the late 254 Villafranchian group (with fourteen species). It shares almost all the taxa with the remaining sites 255 but has some exclusive taxa like: Puma pardoides, Pannonictis nestii and Mustela palerminea. 256 Dmanisi is closely associated with Venta Micena and Apollonia 1 (as it is visible from some of the 257 highest values of Jaccard similarity index, Fig. 4). Indeed, there are only few differences among the 258 guilds of the three sites, despite the difference in age: the badger of Venta Micena and Apollonia 1 259 belongs to the lineage of the extant Meles meles, whereas the specimens from Dmanisi are yet 260 unpublished and not studied in detail. Furthermore, no Acinonyx was found in the Spanish and 11 261 Greek sites, whereas A. pardinensis is present in Dmanisi. The Epivillafranchian sites are grouped 262 in couples with the two chronological groups of the Vallparadís Section grouped together (node 263 B21), whereas on the other side (node B20), Vallonnet is grouped with Untermassfeld for their 264 almost identical guild (Fig. 3). This is also confirmed by the indices of Fig. 4. 265 If we analyze in more detail the Jaccard similarity indices reported in Fig. 4, we see that the 266 African localities have very low indices (between 0.0 and 0.2; SOM Table S3) when compared to 267 Eurasian localities, justifying their clusterization in two different groups. Within the African group, 268 the East African sites of Olduvai Bed 2, KBS and Upper Burgi of Koobi Fora have fairly high 269 indices (0.20–0.56; SOM Table S3) when compared to other African localities, with the exception 270 of Drimolen which have particularly low indices, close to zero, as opposed to the East African sites. 271 Indeed, Drimolen is one of the most different of all African localities, possessesing very low values 272 of Jaccard index, apart with Coopers D and Sterkfortein Member 4. Other guilds with fairly low 273 similarity indices are those of the North African Ahl al Oughlam and Tighennif (below 0.26). 274 Whereas, close to several East African guilds, e.g., Olduvai Bed 2, KBS and Upper Burgi, members 275 of Shungura Formation at Omo differ markedly in Jaccard indices from North and South African 276 ones (indices between 0.17 and 0.0). In comparison to the other guilds analyzed, Ahl al Oughlam 277 and Tighennif are the African localities with a limited similarity with Eurasian guild, as expected 278 considering the position and the connection between the Mediterranean Africa and the Levatine 279 corridor or the Iberian Peninsula. Indeed, the third highest Jaccard index of Tighennif is with 280 ‘Ubeidiya (still rather low: 0.22), whereas Ahl al Oughlam has a slightly high values (between 281 0.207 and 0.176) with St. Vallier and with 2.0 Ma-old localities of China (e.g., Nihewan, 282 Longgupo) and Europe (e.g., Olivola). The other African guilds, both East and South African, do 283 not show indices higher than 0.15 (SOM Table S3). The Eurasian group is characterized by sensibly 284 higher Jaccard indices (Fig. 4). As explained above this group is further subdivided into two 285 subclusters, a Chinese and a European one. The Jaccard indices confirm the pattern yielded by the 286 BCA, particularly showing that the cluster of Epivillafranchian localities, among which Dmanisi is 12 287 also included in spite of its older age (Fig. 4), has the highest Jaccard values of all the compared 288 guilds (above 0.56; SOM Table S3), with Venta Micena showing two greatest indices respectively 289 with Apollonia-1 (0.909) and with Dmanisi (0.818; SOM Table S3). Apart from the clear division 290 in two groups and the high values of the European Epivillafranchian guilds, Jaccard indices allow 291 additional comparisons. Of the Chinese group, the site younger in age (i.e., Gongwangling, 292 Zhoukoudian Loc. 1 and Jinyuan Cave Upper Fauna) are those with the lowest Jaccard values, 293 because of their taxonomic difference with the rest of the Chinese guilds. This is particularly 294 evident with Jinyuan Upper Cave (maximum Jaccard index with a Chinese locality = 0.27), 295 although this site shows similar values (0.214–0.231; SOM Table S3) with European sites, e.g., 296 Venta Micena, EVT 10/12 and Dmanisi. A comparable degree of affinity withWestern sites is 297 shown also by Xiashagou locality of Nihewan basin, which have modestly high values of Poggio 298 Rosso, Dmanisi and Olivola but also with the younger Apollonia-1. Within the European group, the 299 Jaccard indices are on average fairly high (average of the Jaccard indices = 0.373). The sites with 300 lowest Jaccard indices are Khapry FC and Senèze. The former has even an index of zero in 301 comparison to ‘Ubeidiya (Fig. 4; SOM Table S3). On the contrary, in comparison to Chinese 302 localities Khapry FC have modestly high values (between 0.150 and 0.238, SOM Table S3), apart 303 from Gongwangling, Zhoukoudian Loc. 1 and Loc. 18. Indeed, in respect to Chinese guilds, Senèze 304 is more different compared to Khapry FC and so is St. Vallier, whose Jaccard indices are below 305 0.125 (only Nihewan is above this value). The localities of Casa Frata, Olivola and Poggio Rosso 306 have intermediate values (roughly enclosed between 0.20 and 0.50) in comparison to the early 307 Gelasian and the later Calabrian localities of Western Europe. Apart from the high values of the late 308 Calabrian group that includes Dmanisi, ‘Ubeidiya interestly possesses higher Jaccards with guilds 309 of Dmanisi, Venta Micena and Apollonia-1 than in comparison to other localities of Asia and 310 Europe (Fig.4; SOM Table S3). As priviledged subject of this research, the Jaccard indices of 311 Dmanisi confirm its prominent affinity with younger sites (Figs. 3–4; SOM Table S3), but also a 13 312 certain affinity to the Italian sites of Olivola, slightly older, and Casa Frata, slightly younger of the 313 Georgian one. 314 315 3.2. Ecological comparison between guild elements 316 The comparison between the guild structures, in terms of ecological preferences is reported in 317 Figure 5. This figure shows the relative abundance of each preference in the locality but also a 318 clusterizazion of the site on the basis of their ecological composition. The BCA on these 319 parameters, because indirectly influenced by taxonomy, yields interesting results. The Asian 320 localities are separated from the others. Apart from Renzidong, which is completely separated from 321 all the other sites, the other localities are included in a group at the bottom of Figure 5, together 322 with the eastern European locality of Khapry FC. Despite this apparently large-scale geographic 323 clustering, the other sites, the European and African ones, are arranged in different clusters. Some 324 of the pattern seen in the BCA based on taxonomy are respected in Figure 5. Among these, the 325 relationship of Dmanisi with Pirro, Venta Micena and Apollonia 1. If analyzed in detail, the 326 carnivorans of the guild from Dmanisi spot out in their ecological preferences: although more than 327 70% of its taxa were plausibly hypercarnivorous, as meso- and hypocarnivorous species account for 328 30% of the guild, within this large group, different diet preferences and hunting strategies can be 329 observed. Of these hypercarnivorous taxa, surely the most abundant are the cursorial ones, 330 carnivores basing their hunting strategy on long pursuits such as the Eurasian hunting dog C. 331 (Xenocyon) lycaonoides and the cheetah Acinonyx pardinensis, evidencing the prevalence of open 332 environments at Dmanisi. These taxa comprise almost a third of the whole carnivore guild (Fig. 5). 333 The second most important component of the carnivore guild are the hypercarnivorous taxa that 334 ambush their prey (Fig. 5). In Dmanisi those are the felids L. issiodorensis, M. whitei and P. onca 335 georgica. Piscivorous and bone-cracker carnivores are limited to one taxon for each group in 336 Dmanisi, respectively Lutra and Pachycrocuta. The small-sized hypercarnivores are represented by 337 Martes sp. whilst mesocarnivorous and omnivorous species include respectively V. alopecoides, 14 338 and Meles and Ursus. In comparison to African guilds, Dmanisi lacks insectivorous taxa (Fig. 5), 339 e.g., of the families Herpestidae or even Canidae (as in the case of Olduvai Bed I). In turns, African 340 localities lack small-sized hypercarnivores common in Eurasian context (mainly represented by 341 mustelids of the genera Martes and Mustela; Fig. 5). In these two elements, Dmanisi is closer to 342 Eurasian sites. Moreover, generally African localities have a larger proportion of bone-cracking 343 Hyaenidae in their guild compared to Dmanisi and Eurasian sites. Hypocarnivorous taxa with 344 adaptations to nearly herbivorous diet has only been recorded in Eurasia: Ailuropoda in eastern 345 Asian and Ursus deningeri in the late Villafranchian deposits of Europe. Ursus etruscus from 346 Dmanisi is related to the latter yet it does not show herbivorous-related adaptations in its 347 dentognathic features (Medin et al., 2019). As a whole, the most similar guilds are those from 348 Eurasia as the BCA of Figure 5 clearly shows. In terms of proportions between these ecological 349 divisions are ‘Ubeidiya and Nihewan Basin, although the latter has fewer omnivorous species 350 compared to Dmanisi and in the former the percentage of ambush predators is larger compared to 351 the Georgian one. 352 353 3.3. Mantel test 354 The Mantel tests provided significant results when considering both the whole record and the 355 reduced sample of Eurasian sites. The first set of Mantel tests, at Old-World scale, yielded 356 significant relationships either between the taxonomic and ecological distance matrices, when 357 accounting for the effect of time and space (r = 0.243; p(r<0) = 0.002; p(r>0) = 0.999; p(r=0) = 0.002). 358 When excluding ecology, the test yield significant result either between the taxonomic and the 359 temporal distance matrices when accounting of the effect of space (r = 0.163; p(r<0) = 0.008; p(r>0) = 360 0.993; p(r=0) = 0.008) either between the taxonomic and geographic distance matrices when 361 accounting for time (r = 0.551; p(r<0) = 0.001; p(r>0) = 1.000; p(r=0) = 0.001). If the ecology distance 362 matrix is used as response matrix in comparison to temporal (with space distance accounted for) 363 and then spatial (with time accounted for) are predictor matrices, we obtain interesting results. 15 364 When space is the main predictor matrix, we obtain significant correlation between it and ecology 365 distance (r = 0.467; p(r<0) = 0.001; p(r>0) = 1.000; p(r=0) = 0.001), as in the case of the taxonomy 366 distance matrix. In the case of the ecology and age distance matrices accounting for the geographic 367 distance, the relationship is low and non-significant (r = 0.088; p(r<0) = 0.113; p(r>0) = 0.888; p(r=0) = 368 0.240). 369 Comparable results are obtained when we performed by ruling out the African fossil sites. There is 370 significant relationships either between taxonomic and temporal distances when accounting for 371 geography (r = 0.405; p(r<0) = 0.001; p(r>0) = 1.000; p(r=0) = 0.001); between taxonomic and spatial 372 distances when accounting for time (r = 0.581; p(r<0) = 0.001; p(r>0) = 1.000; p(r=0) = 0.001); and 373 lastly, between ecological and geographic distance matrices when accounting for time (r = 0.399; 374 p(r<0) = 0.001; p(r>0) = 1.000; p(r=0) = 0.001). When testing relationship between ecological and 375 temporal distances considering the effect of space differences, we find no significant relationship 376 between them (r = 0.131; p(r<0) = 0.066; p(r>0) = 0.935; p(r=0) = 0.104). 377 378 4. Discussion 379 4.1 Carnivore guild of Dmanisi in the framework of Old World carnivore guilds 380 The association of carnivorans recovered from Dmanisi represent one of most diverse guilds of the 381 Old World, with fourteen taxa (ten recognized at a specific level and four only at a generic one) and 382 seven different ecological categories (the maximum among the studied guilds). While analyzing 383 similarities and differences between carnivore guilds, some issues must be kept in mind. A first idea 384 one might get dealing with large scale analyses of sites (as well as guilds) might be that their 385 clusterization was driven by biases in the taxonomic attributions made by authors describing those 386 associations or those carrying out the analyses. Undoubtedly, such biases arise for different reasons 387 (competency of the researcher; different interpretations, etc.) and are greater at a specific level, 388 rather than at a genus one. The correspondence between the patterns resulted from the analyses 389 genera and the species here carried out (Fig. 3; SOM Fig. S1) rules out the possibity that the 16 390 affinities yielded by the BCA could be explained by biases derived by questionable attributions of 391 taxa, especially at a specific level. Those similarities/differences are therefore related rather to 392 geography, age, taxonomic and ecological composition of the guild. More complex and 393 considerably difficult to assess, is the impact of taphonomic processes in the fossil assemblages, and 394 particularly for the carnivore guild. Indeed, taphonomy may affect greatly the interpretation derived 395 by comparisons between different guilds, both in terms of taxonomy and of ecology. For instance, 396 certain elements like carnivorans preying on small prey are generally small-sized themselves (e.g., 397 herpestids, mustelids, viverrids) and could be underrepresented (or absent) in many sites due to a 398 selective biotic or abiotic agent of accumulation, unsuitable or unfavorable condition of burial 399 and/or fossilization, etc. Likewise, competition among large-sized carnivorans of similar ecology 400 might dwindle the chances of or even prevent the accumulation and the eventual fossilization of 401 some of them. Borne that in mind, the analyses of the considered guilds yielded interesting results. 402 Our study clearly points out the affinity of the carnivorans association of Dmanisi to the younger, 403 late Villafranchian ones of Western Europe in comparison to any other of the Old World. Among 404 these, particularly, Apollonia 1, Pirro Nord and Venta Micena, on the basis of both ecological 405 preferences of their carnivorans and their taxonomy (Figs. 3–5). Furthermore, the 406 similarities/differences resulted from the analyses sheds light on larger-scale patterns of dispersal 407 and affinity that took place during the Early Pleistocene. Although Eurasian sites shares circa one 408 thirds of the species with the African localities, this was not enough to homogenize the composition 409 carnivore guilds of the two continents, neither taxonomically nor ecologically, as explained by the 410 Mantel tests. This supports the importance of areas connecting different continent, e.g., the 411 Levantine corridor and the Pontocaspian region, in explaining association in the Old World. On the 412 one hand, it is an indirect proof that geographical corridors between the two continents acted as a 413 filter for many taxa and as ecological barriers limiting the dispersal of taxa in certain areas (e.g., the 414 presence of ursids in subsaharian Africa). On the other, many of the shared taxa are recorded in 415 these areas of contact. Although explained by geographic distance, the ecological differences are 17 416 poorly related to chronological difference, as visible from the Mantel test. The contribution of both 417 geography and time still explains the taxonomical differences when considering Eurasian sites only, 418 and likewise the ecological differences are still related to geographic distances rather than to 419 temporal ones. 420 421 4.2 The Dmanisi carnivorans in the Old World context 422 The carnivoran taxa that compose the diverse carnivore guild of Dmanisi can be ascribed to three 423 groups based on their paleogeographic origin. 424 425 Species previously recorded in European guilds Most of the species recorded in Dmanisi were 426 previously identified in the middle-late Villafranchian guilds of Europe as H. latidens, A. 427 pardinensis, L. issiodorensis, U. etruscus and V. alopecoides (Rook and Martínez-Navarro, 2010; 428 Madurell-Malapeira et al., 2010; Bartolini-Lucenti and Madurell-Malapeira, 2020 and references 429 therein). Within these typical Villafranchian species, Homotherium latidens was a pursuit predator 430 adapted to open environments with an estimated body mass of 150–250 kg (Hemmer, 2004). It is 431 recorded since the beginning of the Villafranchian (ca. 3.2 Ma) in localities like Perrier-Les 432 Etouaires, Saint Vallier, Senèze, Upper Valdarno or Incarcal complex (Viret, 1954; Ballesio, 1963; 433 Rook et al., 2013; Alba et al., 2016) being also widely recorded in Africa and Asia during the Plio- 434 Pleistocene (Werdelin and Peigné, 2010; Jin et al., 2021). Concerning the social behaviour of 435 Homotherium, or if this species hunts in packs as extant lions, a lot of literature is centered on the 436 sociality of sabertoothed cats, especially Smilodon (Carbone et al., 2009). Nowadays, no direct 437 evidence supports the sociality in Homotherium, however is reasonable to think that this species 438 displays a certain degree of social behavior (Antón, 2013). 439 The European record of the giant cheetah A. pardinensis is noticeably scarce. Although, it is 440 present in several European sites since the beginning of the Villafranchian as Montopoli, Saint 441 Vallier, Villarroya, Pantalla or Olivola, normally it is only recorded on the basis of partial or 18 442 fragmentary remains (Viret, 1954; Cherin et al., 2014a; Madurell-Malapeira and Vinuesa, 2016). 443 The record of a complete forelimb of this species in Dmanisi (Hemmer et al., 2011), show that A. 444 pardinensis is stoutly built solitary hunter (ca. 60–130 kg), which preferred open habitats and can 445 be a potential fresh meat supplier for other scavenger species, including hominins (Hemmer, 2004; 446 Hemmer et al., 2011). The genus is known to occur in Africa since 4 Ma (Werdelin and Peigné, 447 2010), whereas this species is also recorded in Asian Pleistocene taphocoenoses under different 448 taxonomical adscriptions (Qiu et al., 2004). 449 Lynx-like felids traditionally included in Lynx issiodorensis are recorded in Europe since the 450 beginning of the Pliocene in sites like Serrat d’en Vaquer or Cuevas de Alzamora (Werdelin, 1981; 451 Montoya et al., 2001). However, the taxonomic status of these early representatives of the Lynx 452 lineage remains controversial as a consequence of the scanty and poorly diagnostic record. Since 453 the beginning of the Villafranchian, true lynxes were recorded in several sites such as: Perrier-Les 454 Etouaires, Saint Vallier, Olivola and Pantalla (Viret, 1954; Ficcarelli and Torre, 1977; Kurtén, 455 1978; Cherin et al., 2013; Mecozzi et al., 2021). According to some authors L. issiodorensis was the 456 ancestor of both Lynx lynx and Lynx pardinus (Werdelin, 1981; Boscaini et al., 2015; 2016; 457 Mecozzi et al., 2021), with L. pardinus first recorded in eastern Iberian Peninsula at ca. 1.6 Ma 458 (Boscaini et al., 2015). The Dmanisi lynx together with the Upper Valdarno record probably 459 corresponds to the last representatives of this large species, larger than an extant boreal lynx and 460 with an estimated body mass of 23.5 Kg (Mecozzi et al., 2021) enabling it to hunt medium to large 461 species of ungulates. 462 In a recent work focused on the ursid remains from Dmanisi, Medin et al. (2019) provided some 463 inferences on the dietary behavior and evolutionary history of U. etruscus, recorded since the 464 Middle Villafranchian (ca. 2.6 Ma) in European localities like Saint Vallier, Upper Valdarno or 465 Kuruksay (Ristori 1897; Viret, 1954; Sotnikova, 1978). We favor the traditional point of view of 466 Kurtén (1958) who first hypothesized a phylogenetic line for cave bears started with the Early 467 Pleistocene U. etruscus, followed by the Middle Pleistocene Ursus deningeri and finally in the Late 19 468 Pleistocene for Ursus spelaeus s.l. Considering the profound effects of the Pleistocene climatic 469 shifts and their impact on the members of this lineage, it is possible that this pioneering hypothesis 470 of gradual evolution from the omnivorous etruscoid forms to the hyperherbivorous speleoid forms 471 needs to be improved. Especially considering the severity of climatic oscillations at the beginning 472 of the Early-Middle Pleistocene Transition (ca. 1.2 Ma), the sudden increase in size and the start of 473 cave dependence of these forms around that time, appears to fit more with rapid diversification rates 474 of the punctuated equilibria. However, the results of Medin et al. (2019) on the sample from 475 Dmanisi show the existence of a remarkable sexual dimorphism in the early forms of this lineage. 476 Moreover, the authors point evidence that U. etruscus inhabited mixed environments of woodlands 477 and/or wooded grasslands, where it fed on a broad variety of food items with a substantial intake of 478 meat and/or fish (Medin et al., 2017). 479 Regarding medium-to small sized carnivores, few remains of a small canid from Dmanisi can be 480 ascribed to Vulpes alopecoides. The taxonomy and phylogenetic relationships among European 481 Plio-Pleistocene foxes were a matter of considerable debate the last decades (Bartolini-Lucenti and 482 Madurell-Malapeira, 2020 and references therein). In paleontological literature, several different 483 species were described from Early Pleistocene deposits across Europe. These were V. alopecoides, 484 Vulpes praeglacialis and Vulpes praecorsac. One of the main criteria used to distinguish them was 485 size. In a recent revision of the Early Pleistocene fox record across Europe, Bartolini-Lucenti and 486 Madurell-Malapeira (2020) considered that the observed morphological and morphometric 487 variability of fossil foxes from this chronological time span can be accommodated in a single 488 species. Considering the priority of the name V. alopecoides, V. praeglacialis and V. praecorsac 489 should be junior subjective synonyms of V. alopecoides. Its origin seems to be related to the Near 490 Eastern forms of the late Early Pliocene (Bartolini-Lucenti and Madurell-Malapeira, 2021) and it is 491 plausibly related to the extant red fox, Vulpes vulpes (Bartolini-Lucenti and Madurell-Malapeira, 492 2020). The pan-European species V. alopecoides was most probably replaced at the beginning of 20 493 the Middle Pleistocene (ca. 0.6 Ma) by the red fox (Bonifay, 1971; Madurell-Malapeira et al., 494 2021). 495 Although not yet studied in detail or fully characterized, the mustelids recovered from Dmanisi 496 confidently belong to some of the genera widespread in Eurasia during the Late Pliocene and Early 497 Pleistocene. For instance, Pannonictis spp. are known from the Earliest Pleistocene deposits of 498 Hungary (Ros-Montoya et al., 2021: Fig. 1). The most probable origin of the genus is the European 499 continent (Bartolini-Lucenti, 2018; Rook et al., 2018). Whereas Meles was recorded in other 500 European localities like Saint Vallier, Vatera and Lesvos islands (Viret, 1951; 1954; Madurell- 501 Malapeira et al., 2011; Koufos, 2014). The evolutionary history of Lutra is more complicated and 502 fragmentary (Cherin and Rook, 2014; Cherin, 2017). The earliest European record is that of Lutra 503 bravardi from Perrier-Les Etouaries yet the validity of this taxon was questioned by several 504 scholars as its type was lost (Willemsen, 1992; Cherin and Rook, 2014; Cherin, 2017). A certain 505 attribution is that of Lutra fatimazohrae from the Earliest Pleistocene of Morocco (Geraads, 1997). 506 Yet as the author of the original description asserts, this species cannot be regarded as the ancestor 507 of subsequent species (e.g., Lutra simplicidens) and so, for the moment an African origin for these 508 taxa could be excluded (Geraads, 1997). Although its origin is still clouded, Lutra and other similar 509 taxa (Cherin et al., 2016; Cherin, 2017), became widespread during the Late Villafranchian. 510 511 Species previously recorded in Asian guilds Although early, disputed, findings of Canis species are 512 recorded from 3 Ma deposits (Vialette; Lacombat et al., 2006) of Western Europe, the center of 513 radiation and dispersion in the Old World is most certainly eastern Eurasia (Sotnikova and Rook, 514 2010). The subsequent records of Canis in Europe seem to testify these waves of dispersion of 515 different canid taxa from Asia. For instances, Coste San Giacomo, Fonelas P1, Gerakarou, Olivola, 516 Pantalla, Senèze and Upper Valdarno record the arrival of C. etruscus and Canis senezensis (Torre, 517 1967; Martin, 1973; Garrido and Arribas, 2008; Cherin et al., 2014b). Canis arnensis is slightly 518 younger and at present limited to Italy (Bartolini-Lucenti and Rook, 2016). The record in Dmanisi 21 519 of a new species of medium-sized canid, Canis borjgali, represents a novelty in the scenario of 520 Eurasian canids (Bartolini-Lucenti et al., 2020). As also noted by other scholars (e.g., Sotnikova and 521 Rook, 2010), the cranial features possessed by C. borjgali are more derived compared to the coeval 522 canids of Europe (i.e., Canis etruscus, C. arnensis) and Asia (e.g., Canis chihliensis; Bartolini- 523 Lucenti et al., 2020). Considering the similarity, C. borjgali can be regarded as the ancestor of the 524 Canis mosbachensis lineage, including Canis orcensis (Bartolini-Lucenti et al., 2017; 2020; 525 Martínez-Navarro et al., 2021) and, thus, strictly related to the wolf-crown group (i.e., Canis lupus, 526 Canis latrans, Canis lupaster, Canis anthus, Canis aureus; Sotnikova and Rook, 2010; 527 Gopalakrishnan et al., 2017; Bartolini-Lucenti et al., 2020). Moreover, the early record of this 528 derived form anticipates the turnover that can be called the ‘late Early Pleistocene canid revolution’, 529 when the diversity of medium- and large-sized Canidae dropped to two taxa (C. mosbachensis and 530 Canis [Xenocyon] lycaonoides) that replaced more primitive forms across Europe, probably 531 remaining in isolated refugia (Sotnikova, 2001; Rook and Martinez-Navarro, 2010; Sotnikova and 532 Rook, 2010; Petrucci et al., 2013; Bartolini-Lucenti et al., 2017). 533 Whereas other large-sized and hypercarnivorous species are known across Eurasia and Africa 534 during the Calabrian [e.g., C. (Xenocyon) falconeri and C. (Xenocyon) africanus; see Rook, 1994), 535 the specimen of Canis (Xenocyon) lycaonoides from Dmanisi is more derived compared to these 536 (Bartolini-Lucenti et al., 2021), clearly resembling the later forms from Venta Micena, Pirro Nord, 537 Vallparadís Estació and Untermassfeld (Sotnikova, 2001). Like in the case of C. borjgali, the 538 discovery a specimen of C. (Xenocyon) lycaonoides in Dmanisi confirms the relevancy of the 539 Georgian sites in the scenario of dispersion of Canidae, and other species across the entire Old 540 World. Unlike the medium-sized forms, C. (Xenocyon) lycaonoides also managed to disperse into 541 Africa (as testified by the record of Olduvai Bed II) and also to reach North America (Cripple Creek 542 Sump, Alaska; Tedford et al., 2009), becoming one of the extinct canids with the widest range of 543 distribution, larger than extant C. lupus. 544 22 545 Species previously recorded in African guilds Only three of the species identified in the Dmanisi 546 sample are putative possible African immigrants, namely: M. whitei, Pa. brevirostris and P. onca 547 georgica. Similarly to the genus Homotherium, the genus Megantereon was recorded in Europe 548 since the beginning of the Villafranchian (3.2 Ma; Rook and Martínez-Navarro, 2010). 549 Megantereon was clearly an ambush-predator with a powerful forelimb which might have preferred 550 closed and forested habitats. Its estimated body mass ranged between 80–150 Kg (Hemmer, 2004). 551 Some recent research in Asian taphocoenosis suggest Megantereon might not be well-adapted to 552 closed-humid forested environment but rather than more open-habitats (Zhu et al., in press). The 553 species diversity of the genus Megantereon have been highly debated in the past (Martínez-Navarro 554 and Palmqvist, 1995, 1996; Werdelin and Lewis, 2002; Palmqvist et al., 2007; Lewis and Werdelin, 555 2010;), from authors favoring the existence of only two main species in the Old World (Palmqvist 556 et al., 2007) while other support the existence of a higher diversity (Werdelin and Lewis, 2002). 557 Despite the debate, it seems to be widely accepted that in Europe in Early-Middle Villafranchian 558 times the most abundant species is M. cultridens and that, later on, M. whitei replaced this taxon. 559 Megantereon whitei was an African taxa that dispersed into Europe in the Calabrian, and Dmanisi is 560 indeed one of the first European records. This species was subsequently recorded in Venta Micena, 561 Pirro Nord, Argentario Apollonia 1, and lately at Untermassfeld and Vallparadís Section EVT12 562 (Martínez-Navarro and Palmqvist, 1996; Sardella et al., 2008; Petrucci et al., 2013; Hemmer, 2001; 563 Madurell-Malapeira et al., 2010, 2017). Megantereon whitei was an efficient hunter, which 564 probably used its powerful forelimbs to subdue its prey, in order to bite, suffocate and bleed out the 565 prey with a powerful bite with its sharp and long canines (around 10 cm length). Some authors have 566 pointed out that because of its canines and specialized check teeth, M. whitei was only able to eat 567 the soft parts of the prey, leaving a big part of the carcasses for scavengers, like the large-sized 568 hyena Pa. brevirostris and hominins (Martínez-Navarro and Palmqvist, 1995, 1996; Palmqvist et 569 al., 2007). 23 570 The origin and earliest records of the genus Pachycrocuta has also been strongly debated in the 571 last decades. Palmqvist et al. (2011) favored an African origin of this genus, while other authors 572 such as Werdelin (1999) and Liu et al. (2021) convincingly supported an Asian origin of this genus. 573 The latter scholars point out that the earliest records in Asia and Africa are practically simultaneous: 574 ca. 4.0–3.5 Ma (Turner and Antón, 1996). Moreover, these scholars argue that the earlier Asian 575 forms display more reliable morphological characters in comparison with the African ones 576 (Werdelin, 1999; Liu et al., 2021). In either case Pa. brevirostris dispersed into Eurasia around 2.0 577 Ma, where it became one of the most common elements in Eurasian faunas, often the primary agent 578 responsible for the accumulations. Its ubiquitous Eurasian presence lastet until the late Early 579 Pleistocene in Europe, until 0.86 Ma (Martínez-Navarro, 2004, 2010; Madurell-Malapeira et al., 580 2010, 2017), and until the Middle Pleistocene in Asia, where it was extensively recorded at the 581 Zhoukoudian 1 site (Pei, 1934; Palmqvist et al., 2011; Espigares et al., 2013; Madurell-Malapeira et 582 al., 2017; Liu et al., 2021). The impact of this species in the European Pleistocene taphocoenosis 583 and their relationships of putative competition with early hominins raised also hotly debates the last 584 decades (see Espigares et al., 2013; Madurell-Malapeira et al., 2017). 585 Finally, the evolutionary story of Old-World jaguars is far from being resolved. Most of the 586 authors hypothesized an African origin of this species (Hemmer et al., 2010; Argant and Argant, 587 2011) linking their first appearance in Europe at ca. 1.8 Ma with the record of a large pantherine 588 form in the Late Pliocene of Laetoli Upper Beds (ca. 3.7 Ma) ascribed to Barry (1987) to Panthera 589 leo aff. gombaszoegnesis. Other authors on the contrary included the first pantherine material of 590 Laetoli to Panthera leo (Turner, 1990). Eventually, in a more parsimonious hypothesis, Werdelin 591 and Peigné (2010) pointed out that African pantherine specimens older than 2 Ma are not diagnostic 592 at species level. From 2 Ma onwards, all the African material can be ascribed to P. leo and P. 593 pardus (Werdelin and Peigné, 2010). We agree with the latter authors supporting the idea that there 594 is no a direct link among the African specimens and European P. gombaszoegensis s.l. Since 1.8 Ma 595 the early forms of P. gombaszoegensis (or P. toscana according to other authors) were common 24 596 elements of the Late Villafranchian faunas of Europe, as testified to by the records of Olivola, 597 Upper Valdarno, Tegelen, Pirro Nord, Untermassfeld and until the Middle Pleistocene of Chateau 598 (Hemmer, 2001; Argant and Argant, 2011; Hemmer et al., 2010: Jiangzuo and Liu, 2020). Hemmer 599 et al., (2010) analyzed the remains of jaguar from Dmanisi recently pointing several differences 600 with the European species P. gombaszoegensis and erecting a new sub-species P. o. georgica that in 601 the former authors opinion is the direct ancestor of the American species P. o. augusta (Hemmer et 602 al., 2010; Argant and Argant, 2011; Jiangzuo and Liu, 2020). In any case, Hemmer and coauthors 603 (2010) describe the jaguar from Dmanisi as an ambush-predator with an estimated body mass of ca. 604 100 kg, which preferred riparian landscapes in forested and open grassland areas. 605 606 4.3. The role of humans in Dmanisi and the Early Pleistocene Old World carnivore context 607 Dmanisi is characterized by the presence of an abundant lithic artifacts collection of the Oldowan 608 complex (Ferring et al., 2011) and, as it has been described in this study, by a relevant record of 609 fossil vertebrates and a high diversity of large carnivorans species. This abundance of carnivorans is 610 also similar to other Eurasian Early Pleistocene archeo-paleontological sites, such those from Orce, 611 Barranco León and Fuente Nueva 3 , dated 1.4 and 1.3 Ma, respectively (Martínez-Navarro et al., 612 1997, 2010; Espigares et al., 2013, 2019; Toro et al., 2013; Titton et al., 2021), Vallonnet in France, 613 (1.1 Ma; Moullè et al., 2006), or at the site of ‘Ubeidiya in Israel (1.6–1.2 Ma), where a very 614 primitive Acheulian lithic collection together with a rich carnivore assemblage was described 615 (Ballesio, 1986; Bar-Yosef and Goren-Inbar, 1993; Martínez-Navarro et al., 2009). Likewise, 616 carnivorans are well represented at the African sites older than 1.7–1.8 Ma, such as the South 617 African caves, i.e., the lower levels of Swartkrans or Sterkfontein Member 5 (Brian, 1980; 618 O’Regan, 2007; O'Regan and Steininger, 2017), or those from eastern Africa, such as Olduvai Bed 619 I, Lokalalei, or Shungura E–G, and in North Africa at Ain Hanech (see Sahnouni and van der Made, 620 2009, and references there in). 25 621 This scenario totally changes when the archeo-paleontological assemblages are dominated by the 622 presence of developed Acheulian tools. In this context, carnivorans are normally not recorded or, if 623 they are present, they are very scarce and their diversity is very reduced. This evidence is testified 624 to by African records younger than 1.7 Ma, as in the Eastern African sites of Melka Wakena 625 (Ethiopia, Hovers et al., 2021), Buia and Engel Ela-Ramud (Eritrea), dated close to 1.0 Ma 626 (Martinez-Navarro et al., 2004; Martínez-Navarro et al., 2016), or at the Middle Pleistocene site of 627 Asbole, (Ethiopia, ca 0.6 Ma; Geraads et al., 2004), and in North Africa at the site of Wadi Sarrat in 628 Tunisia (0.7 Ma; Martínez-Navarro et al., 2014b). An analogous pattern can be seen in sites out of 629 Africa characterized by developed Acheulian lithic industries and dated to the Early-Middle 630 Pleistocene transition, as those of Gesher Benot Ya’akob (Israel, 0.7–0.8 Ma; Martínez-Navarro and 631 Rabinovich, 2011), and in most of the Middle Pleistocene European Acheulian assemblages, such la 632 Solana del Zamborino in Spain, (0.4 Ma; Álvarez-Posada et al., 2017), or at Venosa Notarchirico in 633 Italy (ca. 0.5 Ma; Piperno, 2000). Nevertheless, an exception to this distribution may be found in 634 the Acheulian sites of Tighennif in Algeria (ca. 1.0 Ma; Geraads, 2016), where carnivores are well 635 represented (Geraads, 2016). 636 This evidence shows that when the lithic assemblages correspond to the Oldowan complex 637 carnivorans are abundant, whereas in Acheulian complexes carnivorans are less represented. How 638 may be interpreted this correlation? Martínez-Navarro (2018) suggested that this can be related to 639 the evolution of human behavior from scavenging to hunting. Humans with Oldowan tools probably 640 found challenging to compete with large carnivorans as active hunter and behaved more frequently 641 as opportunistic scavengers, trying to steal parts or the entire prey’ s carrion hunted down by other 642 large carnivorans or simply found dead. This competition has been described especially with the 643 large hyenas (Epigares et al., 2013). In this scenario, humans were just another species with also 644 carnivore habits in the faunal assemblages of the Early Pleistocene, in strict ecological competition 645 with the large carnivore guild. On the contrary, the limited presence of large carnivorans in 26 646 Acheulian assemblages may indicate that a different and radical change in humans behavior have 647 taken place, from scavengers to active hunters. 648 Dmanisi fits with this interpretation, representing a typical Oldowan site with abundant large 649 carnivorans dominating the scenario. 650 651 5. Conclusions 652 We analyzed the composition of the Dmanisi carnivore guild under different perspectives. We 653 carried out a permutation cluster analysis (BCA; Fig. 3; SOM Fig. S1) on the generic and species 654 composition matrices comparing the studied sample with coeval European, Asian and African 655 Pleistocene sites. Later we tested the different ecomorphological diet categories present at Dmanisi 656 with the same key Old-World sites in order to decipher changes in the environment and carnivore 657 guild composition. Finally, we analyzed the origin and lineage history of the taxa of carnivorans 658 present at Dmanisi. 659 The results of the deep analyses of the Dmanisi carnivore assemblage show that it is consistently 660 most similar to the Late Villafranchian European sites of Pirro Nord, Venta Micena and Apollonia 661 1, and to a lesser extent with the Epivillafranchian localities of Le Vallonnet, Untermassfeld and 662 Vallparadís Section. All these localities, both Late Villafranchian and Epivillafranchian, are 663 characterized by a large diversity of felids, including species like H. latidens, P. onca- 664 gombaszoegensis and M. whitei, a reduced diversity of hyenids, with only Pa. brevirostris present, 665 and pursuit-predator canids as the medium-sized C. borjgali-mosbachensis and the large C. 666 (Xenocyon) lycaonoides. Such guild composition, characterized by new and more derived elements 667 like M. whitei, C. gr. mosbachensis and C. (X.) lycaonoides, which replaced the older and more 668 primitive taxa, respectively M. cultridens, C. etruscus and C. (X.) falconeri, has its First Appearance 669 Datum in Dmanisi. These results support evidence from other taxa (bovids, e.g., Soergelia minor 670 and Bison georgicus; equids, Equus altidens; Bernor et al., 2021) and fits also with the Eurasian 671 dispersion undertaken by the genus Homo at the end of the Gelasian-beginning of the Calabrian, 27 672 whose earliest western European record are the very same sites most closely affine to Dmanisi, 673 Pirro Nord and Orce Site complex. 674 675 Author contribution 676 SBL and JMM conceived the paper with the help of LR and BMN; D.L. and M.B. provided the 677 Dmanisi material; SBL created the dataset, performed the analyses and the comparisons; SBL and 678 JMM wrote the paper with the contributions from LR and BMN; OC, MB, DL, LP discussed the 679 results. 680 681 Acknowledgments 682 The Italian Embassy in Georgia offered continuous support to LR and SBL while working in 683 Tbilisi. The Italian Ministry for Foreign Affairs (DGPCC-V) is acknowledged for financially 684 supporting Italian paleontological research in Georgia. This study has been funded by the Spanish 685 Agencia Estatal de Investigación (grants CGL2016-80975-P to BMN, and CGL2017-82654-P to 686 JM-M, AEI/FEDER-UE) and the Generalitat de Catalunya (CERCA Program GENCAT 2017SGR 687 859 to BMN; SGR 416 GRC to JM-M, AGAUR, Generalitat de Catalunya), and University of 688 Florence (Fonti di Ateneo, to LR). SBL thanks Francesco Carotenuto for his support and fruitful 689 discussion on the analyses. The authors thanks Hannah O’Regan and an anonymous reviewer for 690 their useful insights on the first draft of the manuscript. 691 692 References 693 Adams, J.W., Rovinsky, D.S., Herries, A.I.R., Menter, C.G., 2016. Macromammalian faunas, 694 biochronology and palaeoecology of the Early Pleistocene Main Quarry hominin-bearing 695 deposits of the Drimolen palaeocave system, South Africa. PeerJ 4, e1941. 28 696 Alba, D.M., Madurell-Malapeira, J., Delson, E., Vinuesa, V., Susanna, I., Espigares, M.P., Ros- 697 Montoya, S., Martínez-Navarro, B., 2016. First record of macaques from the Early Pleistocene 698 of Incarcal (NE Iberian Peninsula). J. Hum. Evol. 96, 139–144. 699 Geraads, D., Alemseged, Z., Reed, D., Wynn, J., Roman, D. C., 2004. The Pleistocene fauna (other 700 than Primates) from Asbole, lower Awash Valley, Ethiopia, and its environmental and 701 biochronological implications. Geobios 37, 697–718. 702 Álvarez-Posada, C., Parés, J.M., Sala, R., Viseras, C., Pla-Pueyo, S., 2017. New 703 magnetostratigraphic evidence for the age of Acheulean tools at the archaeo-palaeontological 704 site Solana del Zamborino. Sci. Rep. 7, 13495. 705 Antón, M., 2013. Sabertooth. Indiana University Press, Bloomington. 706 Argant, A., Argant, J., 2011. The Panthera gombaszoegensis story: The contribution of the Château 707 breccia (Saône-et-Loire, Burgundy, France). Quaternaire 4, 247–269. 708 Arribas, A., Palmqvist, P., 1998. Taphonomy and palaeoecology of an assemblage of large 709 mammals: Hyaenid activity in the Lower Pleistocene site at Venta Micena (Orce, Guadix-Baza 710 Basin, Granada, Spain). Geobios 31, 3–47. 711 Ballesio, R., 1963. Monographie d'un Machairodus du gisement villafranchien de Senèze: 712 Homotherium crenatidens Fabrini. Trav. Lab. Géol. Lyon 9, 1–129. 713 Ballesio, R., 1986. Les Carnivores du gisement Pléistocène d’Oubéidiyeh. In: Tchernov, E. (Ed.), 714 Les Mammifères du Pléistocène Inférieur, de la Vallée du Jourdain à Oubéidiyeh. Association 715 Paléorient Edition, Paris, pp. 63–92. 716 Barry, J.C., 1987. The large carnivores from the Laetoli region of Tanzania. In: Leakey, M.D., 717 Harris, J.M. (Eds.), Laetoli: A Pliocene Site in Northern Tanzania. Clarendon Press, Oxford, 718 pp. 238–258. 719 Bartolini-Lucenti, S., 2018. Revising the species “Mustela” ardea Gervais, 1848–1852 (Mammalia, 720 Mustelidae): Martellictis gen. nov. and the systematics of the fossil “Galictinae” of Eurasia. C. 721 R. Palevol 17, 522–535. 29 722 Bartolini-Lucenti, S., Rook, L., 2016. A review on the Late Villafranchian medium-sized canid 723 Canis arnensis based on the evidence from Poggio Rosso (Tuscany, Italy). Quat. Sci. Rev. 151, 724 58–71. 725 Bartolini-Lucenti, S., Rook, L., 2018. The fossil record of the genus Canis (Canidae, Carnivora, 726 Mammalia) from the Upper Valdarno: A critical revision in the frame of the Early and Middle 727 Pleistocene Canids of Eurasia. Alp. Mar. Quat. 31, 169–172. 728 Bartolini-Lucenti, S., Alba, D.M., Rook, L., Moyá-Solá, S., Madurell-Malapeira, J., 2017. Latest 729 Early Pleistocene wolf-like canids from the Iberian Peninsula. Quat. Sci. Rev. 162, 12–25. 730 Bartolini-Lucenti, S., Bukhsianidze, M., Martínez-Navarro, B., Lordkipanidze, D., 2020. The wolf 731 from Dmanisi and Augmented Reality: Review, implications, and opportunities. Front. Earth 732 Sci. 8, 131. 733 Bartolini-Lucenti, S., Madurell-Malapeira, J., 2020. Unraveling the fossil record of foxes: An 734 updated review on the Plio-Pleistocene Vulpes spp. from Europe. Quat. Sci. Rev. 236, 106296. 735 Bartolini-Lucenti, S., Madurell-Malapeira, J., 2021. Ancestral foxes at the gates of Europe: The 736 Pliocene fox from Çalta-1 (Turkey) and their relationships with Asian and European Plio- 737 Pleistocene foxes. C. R. Palevol 20, 619–626. 738 Bartolini-Lucenti, S., Madurell-Malapeira, J., Martínez-Navarro, B., Lordkipanidze, D., Rook, L., 739 2021. The early hunting dog from Dmanisi with comments on the social behaviour in Canidae 740 and hominins. Sci. Rep. 11, 13501. 741 Bar-Yosef, O., Goren-Inbar, N., 1993. The Lithic Assemblages of the Site of Ubeidiya, Jordan 742 Valley. Qedem 34, 1-266. 743 Bernor, R.L., Cirilli, O, Bushkhianidze, M., Lordkipanidze, D., Rook, L., 2021. The Dmanisi 744 Equus: Systematics, Biogeography and Paleoecology. J. Hum. Evol. 21, 103051. 745 Blain, H.A., Agusti, J., Lordkipanidze, D., Rook, L., Delfino, M., 2014. Paleoclimatic and 746 paleoenvironmental context of the Early Pleistocene hominins from Dmanisi (Georgia, Lesser 747 Caucasus) inferred from the herpetofaunal assemblage. Quat. Sci. Rev. 105, 136–150. 30 748 Brain, C.K., 1981. The Hunters or the Hunted? An Introduction to African Cave Taphonomy. 749 University of Chicago Press, Chicago. 750 Boscaini, A., Madurell-Malapeira, J., Llenas, M., Martínez-Navarro, B., 2015. The origin of the 751 critically endangered Iberian lynx: Speciation, diet and adaptive changes. Quat. Sci. Rev. 123, 752 247–253. 753 Boscaini, A., Alba, D.M., Beltrán, J.F., Moya-Sola, S., Madurell-Malapeira, J., 2016. Latest Early 754 Pleistocene remains of Lynx pardinus (Carnivora, Felidae) from the Iberian Peninsula: 755 Taxonomy and evolutionary implications. Quat. Sci. Rev. 143, 96–106. 756 Bonifay, M.-F., 1971. Carnivores quaternaires du Sud-Est de la France. Bull. Mus. Natl. Hist. Nat. 757 21, 1–377. 758 Capellini, E., Welker, F., Pandolfi, L., Ramos-Madrigal, J., Samodova, D., Rüther, P.L., Fotakis, 759 A.K., Lyon, D., Moreno-Mayar, V., Bukhsianidze, M., Rokownikow, R., Christensen, J., 760 Mackie, M., Ginolhac, A., Ferring, R., Tappen, M., Palkopoulou, E., Dickinson, M.R., 761 Stafford, T.W., Chan, Y.L., Götherström, A., Senthvel, K.S.S., Heintzman, P.D., Kapp, J.D., 762 Kirillova, I., Moddley, Y., Agustí, J., Kahlke, R.D., Kiladze, G., Martínez-Navarro, B., Liu, S., 763 Sandoval Velasco, M., Sinding, M.H., Kelstrup, C.D., Allentoft, M.E., Orlando, L., Penkman, 764 K., Shapiro, B., Rook, L., Dalén, L., Gilbert, T.P., Olsen, J.V., Lordkipanidze, D., Willerslev, 765 E., 2019. Early Pleistocene enamel proteome from Dmanisi resolves Stephanorhinus 766 phylogeny. Nature 574, 103–107. 767 Carbone, C., Maddox, T., Funston, P.J., Mills, M.G., Grether, G.F., Van Valkenburgh, B., 2009. 768 Parallels between playbacks and Pleistocene tar seeps suggest sociality in an extinct sabretooth 769 cat, Smilodon. Biol. Lett. 5, 81–85. 770 Cherin, M., 2017. New material of Lutra simplicidens (Carnivora, Mustelidae, Lutrinae), a key 771 taxon for understanding the evolution of European otters. Riv. It. Paleontol. Strat. 123, 433– 772 441. 31 773 Cherin, M., Iurino, D.A., Sardella, R., 2013. New well-preserved material of Lynx issiodorensis 774 valdarnensis (Felidae, Mammalia) from the Early Pleistocene of Pantalla (central Italy). Boll. 775 Soc. Paleontol. It. 52, 103–111. 776 Cherin, M., Rook, L., 2014. First report of Lutra simplicidens (Carnivora, Mustelidae, Lutrinae) in 777 the Early Pleistocene of the Upper Valdarno (Italy) and the origin of European otters. Italian It. 778 J. Geosci. 133, 200–203. 779 Cherin, M., Iurino, D.A., Sardella, R., Rook, L., 2014a. Acinonyx pardinensis (Carnivora, Felidae) 780 from the Early Pleistocene of Pantalla (Italy): Predatory behavior and ecological role of the 781 giant Plio-Pleistocene cheetah. Quat. Sci. Rev. 87, 82–97. 782 Cherin, M., Bertè, D.F., Rook, L., Sardella, R., 2014b. Re-defining Canis etruscus (Canidae, 783 Mammalia): A new look into the evolutionary history of Early Pleistocene dogs resulting from 784 the outstanding fossil record from Pantalla (Italy). J. Mamm. Evol. 21, 95–110. 785 Cohen, B.F., O'Regan, H.J., Steininger, C.M., 2019. Mongoose Manor: Herpestidae remains from 786 the Early Pleistocene Cooper's D locality in the Cradle of Humankind, Gauteng, South Africa. 787 Palaeontol. Afr. 53, 97–113. 788 Crusafont-Pairó, M., Truyols-Santonja, J., 1956. A Biometric Study of the Evolution of Fissiped 789 Carnivores. Evolution 10, 314–332. 790 De Giuli, C., Masini, F., Torre, D., 1986. Late Villafranchian faunas in Italy: the Casa Frata Local 791 Fauna (Upper Valdarno, Tuscany). Palaeontograp. It. 74, 1–9. 792 Delson, E., Faure, M., Guérin, C., Aprile, L., Argant, J., Blackwell, B.A.B, Debard, E., Harcourt- 793 Smith, W., Martin-Suarez, E., Monguillon, A., Parenti, F., Pastre, J.-F., Sen, S., Skinner A.R., 794 Swisher C.C. III, Valli, A.M., 2006. Franco-American renewed research at the Late 795 Villafranchian locality of Senèze (Haute-Loire, France). Cour. Forsch.-Inst. Senckenberg 256, 796 275–290. 32 797 Espigares, M.P., Martínez-Navarro, B., Palmqvist, P., Ros-Montoya, S., Toro, I., Agustí, J., Sala, 798 R., 2013. Homo vs. Pachycrocuta: Earliest evidence of competition for an elephant carcass 799 between scavengers at Fuente Nueva-3 (Orce, Spain). Quat. Int. 295, 113–125. 800 Espigares, M.P., Palmqvist, P., Guerra-Merchán, A., Ros-Montoya, S., García-Aguilar, J.M., 801 Rodríguez-Gómez, G., Serrano, F., Martínez-Navarro, B., 2019. The earliest cut marks of 802 Europe: A discussion on hominin subsistence patterns in the Orce sites (Baza basin, SE Spain). 803 Sci. Rep. 9, 15408. 804 Estes, R.D., 2012. The Behavior Guide to African Mammals: Including Hoofed Mammals, 805 Carnivores, Primates. University of California Press, Berkley. 806 Ferring, R., Oms, O., Agustí, J., Berna, F., Nioradze, M., Sheila, T., Tappen, M., Vekua, A., 807 Zhvania, D., Lordkipanidze, D., 2011. Earliest human occupations at Dmanisi (Georgian 808 Caucasus) dated to 1.85-1.78 Ma. Proc. Natl. Acad. Sci. USA 108, 10432–10436. 809 Ficcarelli, G., Torre, D., 1977. Phyletic relationships between Lynx gr. issiodorensis and Lynx 810 pardina. Boll. Soc. Paleontol. It. 16, 197–202. 811 Flower, L.O.H., Schreve, D.C., 2014. An investigation of palaeodietary variability in European 812 Pleistocene canids. Quat. Sci. Rev. 96, 188–203. 813 Fourvel, J.B., Thackeray, J.F., Brink, J.S., O’Regan, H., Braga, J., 2018. Taphonomic 814 interpretations of a new Plio-Pleistocene hominin-bearing assemblage at Kromdraai (Gauteng, 815 South Africa). Quat. Sci. Rev. 190, 81–97. 816 Gabunia, L., Vekua, A., Lordkipanidze, D., Swisher, C.C., Ferring, R., Justus, A., Nioradze, M., 817 Tvalchrelidze, M., Antón, S., Bosinski, G., Jöris, O., de Lumley, M.A., Majsuradze, G., 818 Mouskhelishvili, A., 2000. Earliest Pleistocene cranial remains from Dmanisi, Republic of 819 Georgia: Taxonomy, geological setting, and age. Science 288, 1019–1025. 820 Gaudzinski, S., 2004. Subsistence patterns of Early Pleistocene hominids in the Levant— 821 taphonomic evidence from the 'Ubeidiya Formation (Israel). J. Archaeol. Sci. 31, 65–75. 33 822 Gkeme, A.G., Koufos, G.D., Kostopoulos, D.S., 2017. The Early Pleistocene stenonoid horse from 823 Libakos (Western Macedonia, Greece): Biochronological and palaeoecological implications 824 and dispersal events. In: Drina, H., Antonarakou, A., Dermitzakis, M.D., Anastasakis, G., 825 Kontakiotis, G., Karakitsios, V., Tsourou, T., Louvari, M. (Eds.), Proceedings of the Abstracts 826 of 15th Congress of the Regional Committee on Mediterranean Neogene Stratigraphy: 827 Exploring a ‘Physical Laboratory: The Mediterranean Basin’. Athens, Greece, 2017. National 828 and Kapodistrian University of Athens, Athens, p. 67. 829 Geraads, D., 1997. Carnivores du Pliocène terminal de Ahl al Oughlam (Casablanca, 830 Maroc). Geobios 30, 127–164. 831 Geraads, D., 2011. A revision of the fossil Canidae (Mammalia) of north-western Africa. 832 Palaeontology 54, 429–446. 833 Geraads, D., 2016. Pleistocene Carnivora (Mammalia) from Tighennif (Ternifine), Algeria. Geobios 834 49, 445–458.Geraads, D., Raynal, J.P., Sbihi-Alaoui, F.Z., 2010. Mammalian faunas from the 835 Pliocene and Pleistocene of Casablanca (Morocco). Hist. Biol. 22, 275–285. 836 Geraads, D., Alemseged, Z., Reed, D., Wynn, J., Roman, D. C., 2004. The Pleistocene fauna (other 837 than Primates) from Asbole, lower Awash Valley, Ethiopia, and its environmental and 838 biochronological implications. Geobios 37, 697–718.Garrido, G., Arribas, A., 2008. Canis 839 accitanus nov. sp., a new small dog (Canidae, Carnivora, Mammalia) from the Fonelas P-1 840 Plio-Pleistocene site (Guadix basin, Granada, Spain). Geobios 41, 751–761. 841 Gopalakrishnan, S., Castruita, J.A.S., Sinding, M.H.S., Kuderna, L.F., Räikkönen, J., Petersen, B., 842 Sicheritz-Ponten, T., Larson, G., Orlando, L., Marques-Bonet, T., Hansen, A.J., 2017. The wolf 843 reference genome sequence (Canis lupus lupus) and its implications for Canis spp. population 844 genomics. BMC Genomics 18, 1–11. 845 Goslee, S.C., Urban, D.L., 2007. The ecodist package for dissimilarity-based analysis of ecological 846 data. J. Stat. Software 22, 1–19. 34 847 Hemmer, H., 2004. Notes on the ecological role of European cats (Mammalia: Felidae) of the last 848 two million years. In: Baquedano-Pérez, E., Rubio, S. (Eds.), Miscelánea en Homenaje a 849 Emiliano Aguirre. Museo Arqueológico Regional, Madrid, pp. 214–232. 850 Hemmer, H., Kahlke, R.-D., Vekua, A.K., 2010. Panthera onca georgica ssp. nov. from the Early 851 Pleistocene of Dmanisi (Republic of Georgia) and the phylogeography of jaguars (Mammalia, 852 Carnivora, Felidae) Neues Jahrb. Geol. Palaontol. Abh. 257, 115–127. 853 Hemmer, H., Kahlke, R.-D., Vekua, A.K., 2011. The cheetah Acinonyx pardinensis (Croizet et 854 Jobert, 1828) s.l. at the hominin site of Dmanisi (Georgia) – A potential prime meat supplier in 855 Early Pleistocene ecosystems. Quat. Sci. Rev. 30, 2703–2714. 856 Hovers, E., Gossa, T., Asrat, A., Niespolo, E.M., Resom, A., Renne, P.R., Ekshtain, R., Herzlinger, 857 G., Ketema, N., Martínez-Navarro, B., 2021. The expansion of the Acheulian to the 858 Southeastern Ethiopian Highlands: Insights from the new early Pleistocene site-complex of 859 Melka Wakena. Quat. Sci. Rev. 253, 106763 860 Kahlke, R.D., García, N., Kostopoulos, D.S., Lacombat, F., Lister, A.M., Mazza, P.P., Spassov, N., 861 Titov, V.V., 2011. Western Palaearctic palaeoenvironmental conditions during the Early and 862 early Middle Pleistocene inferred from large mammal communities, and implications for 863 hominin dispersal in Europe. Quat. Sci. Rev. 30, 1368–1395. 864 Koufos, G.D., 2014. The Villafranchian carnivoran guild of Greece: Implications for the fauna, 865 biochronology and paleoecology. Integr. Zool. 9, 444–460. 866 Koufos, G.D., 2018. New material and revision of the Carnivora, Mammalia from the lower 867 Pleistocene locality Apollonia 1, Greece. Quaternary 1, 1–6. 868 Krijgsman, W., Tesakov, A., Yanina, T., Lazarev, S., Danukalova, G., Van Baak, C.G.C., Agustí, 869 J., Alçiçek, M.C., Aliyeva E., Bista D., Bruch, A., Büyükmeriç, Y., Bukhsianidze, M., Flecker, 870 R., Frolov P., Hoyle, T.M., Jorissen, E.L., Kirscher, U., Koriche, S.A., Kroonenberg, S.B., 871 Lordkipanidze, D., Oms, O., Rausch, L., Singarayer, J., Stoica, M., van de Velde, S., Titov, 35 872 V.V., Wesselingh, F.P., 2019. Quaternary time scales for the Pontocaspian domain: Interbasinal 873 connectivity and faunal evolution. Earth-Sci. Rev. 188, 1–40. 874 Kurtén, B., 1978. The lynx from Etouaries, Lynx issiodorensis (Croizet and Jobert), Late Pliocene. 875 Ann. Zool. Fennici 15, 314–322. 876 Jiangzuo, Q., Liu, J., 2020. First record of the Eurasian jaguar in southern Asia and a review of 877 dental differences between pantherine cats. J. Quat. Sci. 35, 817–830. 878 Jin, C., Liu, J.Y., 2009. Paleolithic Site–The Renzidong Cave, Fanchang, Anhui Province. China 879 Scientific Book, Beijing, pp. 1–439. 880 Jin, C., Wang, Y., Liu, J., Ge, J., Zhao, B., Liu, J., Jiangzuo, Q., 2021. Late Cenozoic mammalian 881 faunal evolution from Jinyuan Cave at Luotuo Hill from Dalian, Northeast China. Quat. Int. 882 577, 15–28. 883 Lacombat, F., Abbazzi, L., Ferretti, M.P., Martínez-Navarro, B., Moullé, P.E., Palombo, M.R., 884 Rook, L., Turner, A., Valli, A.M.F., 2008. New data on the Early Villafranchian fauna from 885 Vialette (Haute-Loire, France) based on the collection of the Crozatier Museum (Le Puy-en- 886 Velay, Haute-Loire, France). Quat. Int. 179, 64–71. 887 Lewis, M.E., Werdelin, L., 2010. Carnivoran dispersal out of Africa during the early Pleistocene: 888 Relevance for hominins? In: Fleagle, J.G., Shea, J., Grine, F., Baden, A., Leakey, R. (Eds.), Out 889 of Africa I: The First Hominin Colonization of Eurasia. Springer, Dordrecht, pp. 13–26. 890 Liu, J., Liu, J., Zhang, H., Wagner, J., Jiangzuo, Q., Song, Y., Jin, C., 2021. The giant short-faced 891 hyena Pachycrocuta brevirostris (Mammalia, Carnivora, Hyaenidae) from Northeast Asia: A 892 reinterpretation of subspecies differentiation and intercontinental dispersal. Quat. Int. 577, 29– 893 51. 894 Lordkipanidze, D., Vekua, A., Ferring, R., Rightmire, G.P., Agustí, J., Kiladze, G., 895 Mouskhelishvili, A., Nioradze, M., Ponce de León, M.S., Tappen, M., Zollikofer, C.P.E., 2005. 896 The earliest toothless hominin skull. Nature 434, 717–718. 36 897 Lordkipanidze, D., Jashashvili, T., Vekua, A., Ponce de León, M.S., Zollikofer, C.P.E., Rightmire, 898 G.P., Pontzer, H., Ferring, R., Oms, O., Tappen, M., Bukhsianidze, M., Agustí, J., Kahlke, R., 899 Kiladze, G., Martínez-Navarro, B., Mouskhelishvili, A., Nioradze, M., Rook, L., 2007. 900 Postcranial evidence from early Homo from Dmanisi, Georgia. Nature 449, 305–310. 901 Madurell-Malapeira, J., Minwer-Barakat, R., Alba, D.M., Garcés, M., Gómez, M., Aurell-Garrido, 902 J., Moyá-Solá, S., Berástegui, X., 2010. The Vallparadís section (Terrassa, Iberian Peninsula) 903 and the latest Villafranchian faunas of Europe. Quat. Sci. Rev. 29, 3972–3982. 904 Madurell-Malapeira, J., Alba, D.M., Marmi, J., Aurell, J., Moyá-Solá, S., 2011. The taxonomic 905 status of European Plio-Pleistocene badgers. J. Vert. Paleontol. 31, 885–894. 906 Madurell-Malapeira, J., Alba, D.M., Minwer-Barakat, R., Aurell-Garrido, J., Moyá-Solá, S., 2012. 907 Early human dispersals into the Iberian Peninsula: A comment on Martínez et al. (2010) and 908 Garcia et al. (2011). J. Hum. Evol. 62, 169–173. 909 Madurell-Malapeira, J., Ros-Montoya, S., Espigares, M.P., Alba, D.M., Aurell-Garrido, J., 2014. 910 Villafranchian large mammals from the Iberian Peninsula: Paleobiogeography, paleoecology 911 and dispersal events. J. Iber. Geol. 40, 167–178. 912 Madurell-Malapeira, J., Vinuesa, V., 2016. El yacimiento villafranchiense inicial de Villarroya 913 (Logroño): Colecciones del Institut Català de Paleontologia Miquel Crusafont. In: Alberdi, 914 M.T., Azanza, B., Cervantes, E. (Eds.), Villarroya, Yacimiento Clave de la Paleontología 915 Riojana. Instituto de Estudios Riojanos, Logroño, pp. 183–216. 916 Madurell-Malapeira, J., Alba, D.M., Espigares, M.P., Vinuesa, V., Palmqvist, P., Martínez-Navarro, 917 B., Moyá-Solá, S., 2017. Were large carnivorans and great climatic shifts limiting factors for 918 hominin dispersals? Evidence of the activity of Pachycrocuta brevirostris during the Mid- 919 Pleistocene Revolution in the Vallparadís Section (Vallès-Penedès Basin, Iberian 920 Peninsula). Quat. Int. 431, 42–52. 37 921 Madurell-Malapeira, J., Bartolini-Lucenti, S., Alba, D.M., 2021. Middle Pleistocene fox from the 922 Vallparadís Section (Vallès-Penedès Basin, NE Iberian Peninsula) and the earliest records of 923 the extant red fox. Riv. It. Paleontol. Strat. 127, 179–187. 924 Martin, R., 1973. Trois nouvelles especes de Caninae (Canidae, Carnivora) des gisements Plio- 925 Villafranchiens d'Europe. Doc. Lab. Geol. Fac. Sci. Lyon 57, 87–96. 926 Martínez-Navarro, B., 2004. Hippos, pigs, bovids, saber-toothed tigers, monkeys, and hominids: 927 Dispersals through the Levantine corridor during late Pliocene and early Pleistocene times. In: 928 Goren-Inbar, N., Speth, J.D. (Eds.), Human Paleoecology in the Levantine Corridor. Oxbow 929 Books, Oxford, pp. 37–52. 930 Martínez-Navarro, B., 2010. Early Pleistocene faunas of Eurasia and hominin dispersals. In: 931 Fleagle, J.G., Shea, J.J., Grine, F.E., Baden, A.L., Leakey, R.E. (Eds.), Out of Africa I: The 932 First Hominin Colonization of Eurasia. Springer, Dordrecht, pp. 207–224. 933 Martínez-Navarro, B., 2018. Oldowan scavengers vs. acheulian hunters: What does the faunal 934 record say?. Glob. J. Archaeol. Anthropol. 6, 555679. 935 Martínez-Navarro, B., Palmqvist, P., 1995. Presence of the African machairodont Megantereon 936 whitei (Broom, 1937) (Felidae, Carnivora, Mammalia) in the Lower Pleistocene site of Venta 937 Micena (Orce, Granada, Spain), with some considerations on the origin, evolution and dispersal 938 of the genus. J. Archaeol. Sci. 22, 569–582. 939 Martínez-Navarro, B., Palmqvist, P., 1996. Presence of the African saber-toothed felid 940 Megantereon whitei (Broom, 1937) (Mammalia, Carnivora, Machairodontinae) in Apollonia-1 941 (Mygdonia Basin, Macedonia, Greece). J. Archaeol. Sci. 23, 869–872. 942 Martínez-Navarro, B., Turq, A., Agustí, J., Oms, O., 1997. Fuente Nueva-3 (Orce, Granada, Spain) 943 and the first human occupation of Europe. J. Hum. Evol. 33, 611–620. 944 Martínez-Navarro, B., Rook, L., Segid, A., Yosief, D., Ferretti, M.P., Shoshani, J., Tecle, T.M., 945 Libsekal, Y., 2004. The Large Fossil Mammals from Buya (Eritrea): Systematics, 946 Biochronology and Paleoenvironments. Riv. It. Paleontol. Strat. 110, 61–88. 38 947 Martinez-Navarro, B., Belmaker, M., Bar-Yosef, O., 2009. The large carnivores from ‘Ubeidiya 948 (Early Pleistocene, Israel): Biochronological and biogeographical implications. J. Hum. 949 Evol. 56, 514–524. 950 Martínez-Navarro, B., Palmqvist, P., Madurell, J., Ros-Montoya, S., Espigares, M.P., Torregrosa, 951 V., Pérez-Claros, J.A., 2010. La fauna de grandes mamíferos de Fuente Nueva-3 y Barranco 952 León-5: Estado de la cuestión. Ocupaciones humanas en el Pleistoceno inferior y medio de la 953 cuenca de Guadix-Baza. Consejería de Cultura, Junta de Andalucía, Arqueología Monografías, 954 Granada, pp. 197-236. 955 Martínez-Navarro, B., Rabinovich, R., 2011. The fossil Bovidae (Artiodactyla, Mammalia) from 956 Gesher Benot Ya’akov, Israel: Out of Africa during the Early-Middle Pleistocene transition. J. 957 Hum. Evol. 60, 375–386. 958 Martínez-Navarro, B., Espigares, M.P., Pastó, I., Ros-Montoya, S., Palmqvist, P., 2014a. Europe: 959 Early Homo Fossil Records. In: Smith, J. (Ed.), Encyclopedia of Global Archaeology. Springer, 960 New York, pp. 2561–2570. 961 Martínez Navarro, B., Karoui-Yaakoub, N., Oms, O., Amri, L., López-García, J. M., Zerai, K., 962 Blain, H.-A., Mtimet, M.-S., Espigares, M.-P., Ben Haj Ali, N., Ros-Montoya, S., Boughdiri, 963 M., Agustí, J., Khayati-Ammar, H., Maalaoui, K., Om El Khir, M., Sala, R., Othmani, A., 964 Hawas, R., Gómez-Merino, G., Solè, A., Carbonell, E., Palmqvist, P., 2014b. The early Middle 965 Pleistocene archaeopaleontological site of Wadi Sarrat (Tunisia) and the earliest record of Bos 966 primigenius. Quat. Sci. Rev. 90, 37–46. 967 Martínez-Navarro, B., Carbonell, E., Medin, T., Oms, O., Rodriguez-Álvarez, X.-P., Ros-Montoya, 968 S., Araia, D., Tsehale, L., Libsekal, Y., 2016. The Engel Ela-Ramud Basin: A new Plio- 969 Pleistocene archaeo-paleontological site in Eritrea. Proc. Eur. Soc. Hum. Evol. 5, 154. 970 Martínez-Navarro, B., Bartolini-Lucenti, S., Palmqvist, P., Ros-Montoya, S., Madurell-Malapeira, 971 J., Espigares, M.P., 2021. A new species of dog from the Early Pleistocene site of Venta 972 Micena (Orce, Baza Basin, Spain). C. R. Palevol 20, 297–314. 39 973 Mazza, P.P.A., Bertini, A., Magri, M., 2004. The late Pliocene site of Poggio Rosso (central Italy): 974 Taphonomy and paleoenvironment. Palaios 19, 227–248. 975 Mecozzi, B., Sardella, R., Boscaini, A., Cherin, M., Costeur, L., Madurell-Malapeira, J., Pavia, M., 976 Profico, A., Iurino, D.A., 2021. The tale of a short-tailed cat: New outstanding Late Pleistocene 977 fossils of Lynx pardinus from southern Italy. Quat. Sci. Rev. 262, 106840. 978 Medin, T., Martínez-Navarro, B., Madurell-Malapeira, J., Figueirido, B., Kopaliani, G., Rivals, F., 979 Kiladze, G., Palmqvist, P., Lordkipanidze, D., 2019. The bears from Dmanisi and the first 980 dispersal of early Homo out of Africa. Sci. Rep. 9, 17752. 981 Medin, T., Martínez-Navarro, B., Rivals, F., Madurell-Malapeira, J., Ros-Montoya, S., Espigares, 982 M.P., Figueirido, B., Rook, L., Palmqvist, P., 2017. Late Villafranchian Ursus etruscus and 983 other large carnivorans from the Orce sites (Guadix-Baza basin, Andalusia, southern Spain): 984 Taxonomy, biochronology, paleobiology, and ecogeographical context. Quat. Int. 431, 20–41. 985 Montoya, P., Sendra, J., 2001. Una mandíbula de Lynx issiodorensis (Croicet y Jobert, 1828) 986 (Carnivora, Mammalia) en el Plioceno Inferior de Cuevas del Almanzora (Almería, 987 España). Revista Española de Paleontología 10, 125–131. 988 Moullé, P.E., Lacombat, F., Echassoux, A., 2006. Contribution of the large mammals of Vallonnet 989 cave (Roquebrune-Cap-Martin, Alpes-Maritimes, France) to the knowledge of biochronological 990 frame of the second half of the Lower Pleistocene in Europe. L’Anthropologie 110, 837-849. 991 O'Regan, H.J., 2007. Revision of the Carnivora from Member 5, Sterkfontein, South Africa, based 992 on a re-assessment of the published material and site stratigraphy. Ann. Transvaal Mus. 44, 993 209–214. 994 O'Regan, H.J., Steininger, C.M., 2017. Felidae from Cooper's Cave, South Africa (Mammalia: 995 Carnivora). Geodiversitas 39, 315–332. 996 Palmqvist, P., Martínez-Navarro, B., Pérez-Claros, J.A., Torregrosa, V., Figueirido, B., Jiménez- 997 Arenas, J.M., Espigares, M.P., Ros-Montoya, S., De Renzi, M., 2011. The giant hyena 40 998 Pachycrocuta brevirostris: Modelling the bone-cracking behavior of an extinct carnivore. Quat. 999 Int. 243, 61–79. 1000 Palmqvist, P., Torregrosa, V., Pérez-Claros, J.A., Martínez-Navarro, B., Turner, A., 2007. A re- 1001 evaluation of the diversity of Megantereon (Mammalia, Carnivora, Machairodontinae) and the 1002 problem of species identification in extinct carnivores. J. Vert. Paleontol. 27, 160–175. 1003 Palombo, M.R., Valli, A.M.F., 2003. Remarks on the biochronology of mammalian faunal 1004 complexes from the Pliocene to the Middle Pleistocene in France. Geol. Romana 37, 145–163. 1005 Pei, W.C., 1934. On the Carnivora from locality 1 of Choukoutien. Paleontologia Sinica 8, 76–80. 1006 Petrucci, M., Cipullo, A., Martinez-Navarro, B., Rook, L., Sardella, R., 2013. The Late 1007 Villafranchian (Early Pleistocene) carnivores (Carnivora, Mammalia) from Pirro Nord 1008 (Italy). Palaeontograp. Abt. A 298, 113–145. 1009 Piperno, M., 2000. Notarchirico. Un Sito del Pleistocene Medio Iniziale nel Bacino di Venosa 1010 (Poliedrica). Osanna Edizioni, Venosa. 1011 Popowics, T.E., 2003. Postcanine dental form in the Mustelidae and Viverridae (Carnivora: 1012 Mammalia). J. Morphol. 256, 322–341. 1013 Qiu, Z., Deng, T., Wang, B., 2004. Early Pleistocene mammalian fauna from Longdan, Dongxiang, 1014 Gansu, China. Paleontol. Sinica 191, 1–198. 1015 Ristori, M., 1897. L’Orso pliocenico di Valdarno ed Olivola in Val di Magra. Palaeontograp. It. 3, 1016 15–76. 1017 Rook, L., 1994. The Plio-Pleistocene Old World Canis (Xenocyon) ex gr. falconeri. Boll. Soc. 1018 Paleontol. It. 33, 71–82. 1019 Rook, L., Martínez-Navarro, B., 2010. Villafranchian: The long story of a Plio-Pleistocene 1020 European large mammal biochronologic unit. Quat. Int. 243, 61–79. 1021 Rook, L., Croitor, R., Delfino, M., Ferreti, M., Gallai, G., Pavia, M., 2013. The Upper Valdarno 1022 Plio-Pleistocene vertebrate record: An historical overview, with notes on palaeobiology and 1023 stratigraphic significance of some important taxa. Ital. J. Geosci. 132, 104–125. 41 1024 Rook, L., Bartolini-Lucenti, S., Tuveri, C., Arca, M., 2018. Mustelids (Carnivora, Mammalia) from 1025 Monte Tuttavista fissure fillings (Early and Middle Pleistocene; Orosei, Sardinia): Taxonomy 1026 and evolution of the insular Sardinian Galictini. Quat. Sci. Rev. 197, 209–223. 1027 Ros-Montoya, S., Bartolini-Lucenti, S., Espigares, M.R., Palmqvist, P., Martínez-Navarro, B., 1028 2021. First review of the Lycodontini material (Mustelidae, Carnivora, Mammalia) from the 1029 Lower Pleistocene archaeo-palaeontological sites of Orce (Southern Spain). Riv. It. Paleontol. 1030 Strat. 127, 33–47. 1031 Sahnouni, M., Made, J.v.d., 2009. The Oldowan in North Africa within a biochronological 1032 framework. In: Schick, K., Toth, N. (Eds.), The Cutting Edge: New Approaches to the 1033 Archaeology of Human Origins. Stone Age Institute Press, Bloomington, pp. 179–210. 1034 Sahnouni, M., Parés, J. M., Duval, M., Cáceres, I., Harichane, Z., Made, J. v. d., Pérez-González A., 1035 Abdessadok S., Kandi N., Derradji A., Medig M., Boulaghraif K., Semaw, S., 2018. 1.9- 1036 million-and 2.4-million-year-old artifacts and stone tool–cutmarked bones from Ain Boucherit, 1037 Algeria. Science 362, 1297–1301. 1038 Sardella, R., Petrucci, M., Rook, L., 2008. The African species Megantereon whitei from the Early 1039 Pleistocene of Monte Argentario (South Tuscany, Central Italy). C. R. Palevol 7, 601–606. 1040 Sotnikova, M.V., 1978. Upper Pliocene carnivora of central Asia. Int. Geol. Rev. 20, 335–338. 1041 Sotnikova, M., 2001. Remains of Canidae from the Lower Pleistocene site of Untermassfeld. In: 1042 Kahlke, R.-D. (Ed.), Das Pleistozän von Untermaßfeld bei Meiningen (Thüringen). 1043 Monographien des Römisch-Germanischen Zentralmuseums 40. Habelt, Bonn, pp. 607–632. 1044 Sotnikova, M., Baigusheva, V.S., Titov, V., 2002. Carnivores of the Khapry faunal assemblage and 1045 their stratigraphic implications. Strat. Geol. Correl. 10, 375–390. 1046 Sotnikova, M., Titov, V., 2009. Carnivora of the Tamanian faunal unit (the Azov Sea area). Quat. 1047 Int. 201, 43–52. 1048 Sotnikova, M., Rook, L., 2010. Dispersal of the Canini (Mammalia, Canidae: Caninae) across 1049 Eurasia during the late Miocene to early Pleistocene. Quat. Int. 212, 86–97. 42 1050 Tedford, R.H., Wang, X., Taylor, B.E., 2009. Phylogenetic systematics of the North American 1051 fossil caninae (Carnivora: Canidae). Bull. Am. Mus. Nat. Hist. 325, 1–218. 1052 Teilhard de Chardin, P., 1940. The fossils from locality 18 near Peking. Palaeontol. Sinica 9, 1–101. 1053 Titton, S., Oms, O., Barsky, D., Bargalló, A., Serrano-Ramos, A., Solano-García, J.A., Sánchez- 1054 Bandera, C., Yravedra, J., Blain, H.A., Toro-Moyano, I., Jiménez-Arenas, J.M., Sala-Ramos, 1055 R., 2021. Oldowan stone knapping and percussive activities on a raw material reservoir deposit 1056 1.4 million years ago at Barranco León (Orce, Spain). Anthropol. Archaeol. Sci. 13, 108. 1057 Toro-Moyano, I., Martínez-Navarro, B., Agustí, J., Souday, C., Bermúdez de Castro, J.M., 1058 Martinón-Torres, M., Fajardo, B., Duval, M., Falguères, C., Oms, O., Parés, J.M., Anadón, P., 1059 Julià, R., García-Aguilar, J.M., Moigne, A.M., Espigares, M.P., Ros-Montoya, S., Palmqvist, 1060 P., 2013. The oldest human fossil in Europe, from Orce (Spain). J. Hum. Evol. 65, 1–9. 1061 Torre, D., 1967. I cani villafranchiani della Toscana. Palaeontogr. Ital. 64, 113–138. 1062 Turner, A., 1990. The evolution of the guild of terrestrial carnivores during the Plio-Pleistocene in 1063 Africa. Geobios 23, 349–368. 1064 Van Valkenburgh, B., 1989. Carnivore dental adaptations and diet: A study of trophic diversity 1065 within guilds. In: Gittleman, J.L. (Ed.), Carnivore Behavior, Ecology, and Evolution. Cornell 1066 University Press, Ithaca, pp. 410–436. 1067 Van Valkenburgh, B., Koepfli, K.-P., 1993. Cranial and dental adaptations to predation in canids. 1068 Symp. Zool. Soc. London 65, 15–37. 1069 Vekua, A., 1995. Die Wirbeltierfauna des Villafranchian von Dmanisi und ihre biostratigraphische 1070 Bedeutung. Jahrb. Roem. Germ. 42, 77–180. 1071 Vekua, A., Lordkipanidze, D., Rightmire, G.P., Agustí, J., Ferring, R., Maisuradze, G., 1072 Mouskhelishvili, A., Nioradze, M., Ponce de Leon, M., Tappen, M., Tvalchrelidze, M., 1073 Zollikofer, C., 2002. A new skull of early Homo from Dmanisi, Georgia. Science 297, 85–89. 1074 Viret, J., 1951. Meles thorali n. sp. du loess villafranchien de Saint-Vallier (Drôme). Eclog. Geol. 1075 Helv. 43, 274–287. 43 1076 Viret, J., 1954. Le loess a bancs durcis de Saint-Vallier (Drôme) et sa faune de mammifères 1077 villafranchiens. Nouv. Arch. Mus. Hist. Nat. Lyon 4, 1–195. 1078 Wang, X., Flynn, L.J., Fortelius, M. (Eds.), 2013. Fossil Mammals of Asia: Neogene 1079 Biostratigraphy and Chronology. Columbia University Press, New York, pp. 1–732. 1080 Werdelin, L., 1981, The evolution of lynxes. Ann. Zool. Fennici 18, 37–71. 1081 Werdelin, L., 1999. Pachycrocuta (hyaenids) from the Pliocene of east Africa. Paläontolog. Zeitsch. 1082 73, 157–165. 1083 Werdelin, L., Lewis, M.E., 2002. Species identification in Megantereon: A reply to Palmqvist. J. 1084 Paleontol. 76, 931–933. 1085 Werdelin, L., Peigné, S., 2010. Carnivora. In: Werdelin, L., Sanders, W.J. (Eds.), Cenozoic 1086 Mammals of Africa. University of California Press, Berkeley, pp. 603–657. 1087 Werdelin, L., Sanders, W.J. (Eds.), 2010. Cenozoic Mammals of Africa. University of California 1088 Press, Berkeley, pp. 1–986. 1089 Willemsen, G.F., 1992. A revision of the Pliocene and Quaternary Lutrinae from Europe. Scrip. 1090 Geol. 101, 1-115. 1091 Zhu, M., Jiangzuo, Q., Qin, D., Jin, C., Sun, C., Wang, Y., Liu, J., In press. First discovery of 1092 Megantereon skull from southern China. Hist. Biol. 1093 https://doi.org/10.1080/08912963.2020.1867981 1094 1095 1096 Figure Captions 1097 Figure 1. Map of the Old World showing the considered localities in the present paper. Localities 1098 are grouped in chronological cluster: circles are localities dated 2.70–2.01 Ma; squares, sites 1099 between 2.00–1.20 Ma; triangles are localities between 1.19–0.70 Ma. Red star is the location of 1100 Dmanisi. Numbers are the same as in Table 1. Numbers: 1 = Dmanisi (Georgia); 2 = Ahl al 1101 Oughlam (Morocco); 3 = Tighennif (Algeria); 4 = Upper Burgi Member, Koobi Fora Formation 44 1102 (Kenya); 5 = KBS Member, Koobi Fora Formation (Kenya); 6 = Okote Member, Koobi Fora 1103 Formation (Kenya); 7 = Olduvai Bed I (Tanzania); 8 = Olduvai Bed II (Tanzania); 9 = Omo 1104 Members C–F, Shungura Formation (Ethiopia); 10 = Omo Members G–K, Shungura Formation 1105 (Ethiopia); 11 = Sterkfontein Member 4 (South Africa); 12 = Kromdraai Member 2 (South Africa); 1106 13 = Drimolen (South Africa); 14 = Coopers D (South Africa); 15 = Saint Vallier (France); 16 = 1107 Senèze (France); 17 = Olivola (Italy); 18 = Poggio Rosso (Italy); 19 = Casa Frata (Italy); 20 = 1108 Venta Micena (Spain); 21 = Pirro Nord (Italy); 22 = Apollonia 1 (Greece); 23 = Vallonnet (France); 1109 24 = EVT10/12, Vallparadís Estacio level 10–12 (Spain); 25 = EVT7/CGRD7, Vallparadís Estacio 1110 level 7/Cal Guardiola 7 (Spain); 26 = Untermassfeld (Germany); 27 = Ubeidiya (Israel); 28 = 1111 Khapry Faunal Complex (European Russia); 29 = Longdan locality (China); 30 = Zhoukoudian 1112 Locality 18 (China); 31 = Renzidong Cave (China); 32 = Xiashagou, Classic Nihewan (China); 33 1113 = Jinyuan Cave Lower fauna (China); 34 = Longgupo (China); 35 = Gongwangling (China); 36 = 1114 Jinyuan Cave Upper fauna (China); 37 = Zhoukoudian Locality 1 (China). 1115 1116 Figure 2. Pie charts of carnivore guild composition in Dmanisi and three roughly coeval localities 1117 of Europe (Olivola, Italy), Asia (Jinyuan Cave Lower Fauna; China) and Africa (Kromdraai 1118 Member 2; South Africa) with the relative percentage of the families there recorded. 1119 1120 1121 1122 Figure 3. Dendrogram resulting from the bootstrapping cluster analysis (BCA) based on the 1123 presence/absence species group matrix of 36 sites across the Old World (Table 1). Abbreviations: 1124 AaO = Ahl al Oughlam (Morocco); APL = Apollonia 1 (Greece); CF = Casa Frata (Italy); COOPD 1125 = Coopers D; DRIM = Drimolen (South Africa); EVT10/12 = Vallparadís Estacio level 10–12 1126 (Spain); EVT7/CGRD7 = Vallparadís Estacio level 7/Cal Guardiola 7 (Spain); GWC = 1127 Gongwangling (China); JYCLow = Jinyuan Cave Lower fauna (China); JYCUpp = Jinyuan Cave 45 1128 Upper fauna (China); KBS = KBS Member, Koobi Fora Formation (Kenya); KhFC = Khapry 1129 Faunal Complex (European Russia); KRMD2 = Kromdraai Member 2 (South Africa); LGP = 1130 Longgupo (China); Longdan = Longdan locality (China); Nihew = Xiashagou, Classic Nihewan 1131 (China); Okot = Okote Member, Koobi Fora Formation (Kenya); Old 1 = Olduvai Bed I 1132 (Tanzania); Old 2 = Olduvai Bed II (Tanzania); OLIV = Olivola (Italy); OMO C-F = Members C– 1133 F, Shungura Formation (Ethiopia); OMO G-K = Members G–K, Shungura Formation (Ethiopia); 1134 Pirro = Pirro Nord (Italy); PR = Poggio Rosso (Italy); RZD = Renzidong Cave (China); Sen = 1135 Senéze (France); St.Val = St. Vallier (France); STK Mb4 = Sterkfontein Member 4 (South Africa); 1136 TIGH = Tighennif (Algeria); U Burgi = Upper Burgi Member, Koobi Fora Formation (Kenya); 1137 UBEI = ‘Ubeidiya (Israel); UMD = Untermassfeld (Germany); Vallt = Vallonnet (France); VM = 1138 Venta Micena (Spain); ZKD1 = Zhoukoudian Locality 1 (China); ZKD18 = Zhoukoudian Locality 1139 18 (China). 1140 1141 Figure 4. Heatmap of the Jaccard similarity index calculated on the presence/absence matrix of the 1142 species of the thirty-six localities analyzed here (Table 1). Dendrograms are the same as in Figure 1143 4. The color code is expressed by the legend on the top left corner of the graph. This color legend 1144 also reports the number of each Jaccard index (light cyan histogram). Abbreviations: AaO = Ahl al 1145 Oughlam (Morocco); APL = Apollonia 1 (Greece); CF = Casa Frata (Italy); COOPD = Coopers D; 1146 DRIM = Drimolen (South Africa); EVT10/12 = Vallparadís Estacio level 10–12 (Spain); 1147 EVT7/CGRD7 = Vallparadís Estacio level 7/Cal Guardiola 7 (Spain); GWC = Gongwangling 1148 (China); JYCLow = Jinyuan Cave Lower fauna (China); JYCUpp = Jinyuan Cave Upper fauna 1149 (China); KBS = KBS Member, Koobi Fora Formation (Kenya); KhFC = Khapry Faunal Complex 1150 (European Russia); KRMD2 = Kromdraai Member 2 (South Africa); LGP = Longgupo (China); 1151 Longdan = Longdan locality (China); Nihew = Xiashagou, Classic Nihewan (China); Okot = Okote 1152 Member, Koobi Fora Formation (Kenya); Old 1 = Olduvai Bed I (Tanzania); Old 2 = Olduvai Bed 1153 II (Tanzania); OLIV = Olivola (Italy); OMO C-F = Members C–F, Shungura Formation (Ethiopia); 46 1154 OMO G-K = Members G–K, Shungura Formation (Ethiopia); Pirro = Pirro Nord (Italy); PR = 1155 Poggio Rosso (Italy); RZD = Renzidong Cave (China); Sen = Senéze (France); St.Val = St. Vallier 1156 (France); STK Mb4 = Sterkfontein Member 4 (South Africa); TIGH = Tighennif (Algeria); U Burgi 1157 = Upper Burgi Member, Koobi Fora Formation (Kenya); UBEI = ‘Ubeidiya (Israel); UMD = 1158 Untermassfeld (Germany); Vallt = Vallonnet (France); VM = Venta Micena (Spain); ZKD1 = 1159 Zhoukoudian Locality 1 (China); ZKD18 = Zhoukoudian Locality 18 (China). 1160 1161 1162 Figure 5. Histograms of the relative proportions of the ecological preference of the carnivorans 1163 in the analyzed guilds of the Old World, arranged according to a BCA based on ecological matrix 1164 of the sites. Abbreviations: AMB = hypercarnivorous ambush predators; BoCr = hypercarnivorous 1165 species with bone-cracking adaptations; CURS = hypercarnivorous cursorial predators; HERB = 1166 hypocarnivorous adaptations towards herbivory; HSmP = small-sized hypercarnivorous species; 1167 INSE = insectivorous species; MESO = mesocarnivores; OMNI = omnivorous, hypocarnivorous 1168 and opportunistic feeders; PISC = hypercarnivores adapted to a piscivorous diet; AaO = Ahl al 1169 Oughlam (Morocco); APL = Apollonia 1 (Greece); CF = Casa Frata (Italy); COOPD = Coopers D; 1170 DRIM = Drimolen (South Africa); EVT10/12 = Vallparadís Estacio level 10–12 (Spain); 1171 EVT7/CGRD7 = Vallparadís Estacio level 7/Cal Guardiola 7 (Spain); GWC = Gongwangling 1172 (China); JYCLow = Jinyuan Cave Lower fauna (China); JYCUpp = Jinyuan Cave Upper fauna 1173 (China); KBS = KBS Member, Koobi Fora Formation (Kenya); KhFC = Khapry Faunal Complex 1174 (European Russia); KRMD2 = Kromdraai Member 2 (South Africa); LGP = Longgupo (China); 1175 Longdan = Longdan locality (China); Nihew = Xiashagou, Classic Nihewan (China); Okot = Okote 1176 Member, Koobi Fora Formation (Kenya); Old 1 = Olduvai Bed I (Tanzania); Old 2 = Olduvai Bed 1177 II (Tanzania); OLIV = Olivola (Italy); OMO C-F = Members C–F, Shungura Formation (Ethiopia); 1178 OMO G-K = Members G–K, Shungura Formation (Ethiopia); Pirro = Pirro Nord (Italy); PR = 1179 Poggio Rosso (Italy); RZD = Renzidong Cave (China); Sen = Senéze (France); St.Val = St. Vallier 47 1180 (France); STK Mb4 = Sterkfontein Member 4 (South Africa); TIGH = Tighennif (Algeria); U Burgi 1181 = Upper Burgi Member, Koobi Fora Formation (Kenya); UBEI = ‘Ubeidiya (Israel); UMD = 1182 Untermassfeld (Germany); Vallt = Vallonnet (France); VM = Venta Micena (Spain); ZKD1 = 1183 Zhoukoudian Locality 1 (China); ZKD18 = Zhoukoudian Locality 18 (China). 1184 48 Table 1 Presence of species of carnivorans in selected European, Asian and African sites from the Early to Middle Pleistocene with their ecological characterization. Species E. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 Crossarchus transvaalensis OMNI X Galerella debilis MESO X Helogale hirtula-kitafe MESO X X Herpestes gr. ichneumon MESO X X X X X Ichneumia albicaudata INSE X X Mungos minutus INSE X Suricata suricatta INSE X X X Civettictis gr. civetta OMNI X Cynictis penicillata INSE X X Genetta genetta MESO X X Megaviverra gr. pleistocenica MESO X X Pseudocivetta ingens MESO X X X X X X Viverra leakey MESO X X Aonyx capensis HSmP X Arctonyx minor OMNI X X Arctonyx collaris OMNI X Baranogale helbingi HSmP X Eirictis gr. pachygnatha HSmP X X X Lutra gr. simplicidens PISC X X X X Lutra licenti PISC X Martellictis ardea MESO X X X Martes crassa-andressoni HSmP X X Meles gr. thorali-meles OMNI X X X X X X X X Meles gr. chiai-teilhardi OMNI X X X X X X X X Mellivora gr capensis MESO X X X X Mustela altaica HSmP X X Mustela palerminea HSmP X Oriensictis melina PISC X Pannonictis hoffstetteri MESO X Pannonictis nestii MESO X X Poecillictis lybica MESO X X Prepoecilogale bolti MESO X X Torolutra ougandensis PISC X Vormela peregusna HSmP X Ontocetus emmonsi PISC X Acinonyx gr. jubatus CURS X X X X X X Acinonyx gr. pardinensis CURS X X X X X X X X X Caracal caracal AMB X X X Dinofelis petteri-aronoki-piveteau AMB X X X X X Dinofelis barlowi AMB X X X Felis gr. sylvetris AMB X X X X X X X X X X X X X Homotherium latidens CURS X X X X X X X X X X X X X X X X X X X X Lynx gr. issiodorensis-pardinus AMB X X X X X X X X X X X X X X X X X X X X Megantereon cultridens AMB X X X Megantereon gr. falconeri AMB X X X X X X Megantereon whitei AMB X X X X X X X X X X X X X X Panthera leo CURS X X X X X X X X X X X Panthera gr. 'onca' AMB X X X X X X X X X Panthera pardus AMB X X X X X X X X X X X X X X Puma pardoides AMB X X X X Sivapanthera linxiaensis CURS X X X X Chasmaporthetes gr. kani- progressus CURS X X X Chasmaporthetes gr. lunensis CURS X X X X Chasmaporthetes gr. nitidula CURS X X X X X Crocuta gr.crocuta BoCr X X X X X X X X X X X X X X Hyaena gr. hyaena BoCr X X X X X X X Hyaenictitherium? barbatum CURS X Lycaenops silberbergi CURS X X Pachycrocuta brevirostris BoCr X X X X X X X X X X X X X X X X X X X X X X Parahyaena brunnea BoCr X X X Pliocrocuta perrieri BoCr X X X X X Proteles gr. cristatus INSE X X Canis arnensis MESO X X Canis brevirostris MESO X Canis gr. chihliensis CURS X X X X X Canis gr. etruscus MESO X X X X X X X X Canis gr. mosbachensis CURS X X X X X X X X X X X X Canis (Xenocyon) gr. dubius CURS X X X X Canis (Xenocyon) gr. falconeri CURS X X X Canis (Xenocyon) gr. lycaonoides CURS X X X X X X X X X X X X Cuon alpinus CURS X Eucyon minor HSmP X Lupulella gr. adustus OMNI X Lupulella gr. mesomelas MESO X X X X X X Lycaon pictus CURS X Nyctereutes gr. megamastoides OMNI X X X X X X X X X Protootocyon reckii INSE X Vulpes alopecoides MESO X X X X X X X X X X X Vulpes chikushanensis OMNI X X X X X Vulpes gr. chama MESO X X Vulpes gr. rueppelli OMNI X X Ailuropoda gr. melanoleuca HERB X X X Ursus gr. etruscus OMNI X X X X X X X X X X X X X X X X Ursus deningeri HERB X X X X X Ursus gr. thibetanus OMNI X X X X Total number of species per site 10 21 11 9 6 7 15 8 7 4 12 15 11 16 14 9 11 8 8 10 14 11 13 10 11 11 11 7 16 8 14 19 14 13 10 6 9 Abbreviations: E. = ecological preferences; AMB = hypercarnivorous ambush predators; BoCr = hypercarnivorous species with bone- cracking adaptations; CURS = hypercarnivorous cursorial predators; HERB = hypocarnivorous adaptations towards herbivory; HSmP = small-sized hypercarnivorous species; INSE = insectivorous species; MESO = mesocarnivores; OMNI = omnivorous, hypocarnivorous and opportunistic feeders; PISC = hypercarnivores adapted to a piscivorous diet; 1 = Dmanisi (Georgia; ca 1.8 Ma); 2 = Ahl al Oughlam (Morocco; ca 2.5 Ma); 3 = Tighennif (Algeria; ca 1.0 Ma); 4 = Upper Burgi Member, Koobi Fora Formation (Kenya; ca 2.0-1.9 Ma); 5 = KBS Member, Koobi Fora Formation (Kenya; ca 1.9-1.7 Ma); 6 = Okote Member, Koobi Fora Formation (Kenya; ca 1.5-1.3 Ma); 7 = Olduvai Bed I (Tanzania; ca 1.8 Ma); 8 = Olduvai Bed II (Tanzania; ca 1.2 Ma); 9 = Omo Members C–F, Shungura Formation (Ethiopia; ca 2.6-2.3 Ma); 10 = Omo Members G–K, Shungura Formation (Ethiopia; ca 2.0-1.5 Ma); 11 = Sterkfontein Member 4 (South Africa; ca 2.1 Ma); 12 = Kromdraai Member 2 (South Africa; ca 2.0 Ma); 13 = Drimolen (South Africa; ca 2.0-1.5 Ma); 14 = Coopers D (South Africa; ca 1.9 Ma); 15 = Saint Vallier (France; ca 2.4 Ma); 16 = Senèze (France; ca 2.2 Ma); 17 = Olivola (Italy; ca 2.0 Ma); 18 = Poggio Rosso (Italy; ca 1.9 Ma); 19 = Casa Frata (Italy; ca 1.7-1.6 Ma); 20 = Venta Micena (Spain; ca 1.5 Ma); 21 = Pirro Nord (Italy; ca 1.4 Ma); 22 = Apollonia 1 (Greece; ca 1.2-0.9 Ma); 23 = Vallonnet (France; ca 1.0 Ma); 24 = EVT10/12, Vallparadís Estacio level 10–12 (Spain; ca 1.1 Ma); 25 = EVT7/CGRD7, Vallparadís Estacio level 7/Cal Guardiola 7 (Spain; ca 0.9 Ma); 26 = Untermassfeld (Germany; ca 1.0 Ma); 27 = Ubeidiya (Israel; ca 1.4 Ma); 28 = Khapry Faunal Complex (European Russia; ca 2.5-2.0 Ma); 29 = Longdan locality (China; ca 2.4-2.2 Ma); 30 = Zhoukoudian Locality 18 (China; ca 2.2 Ma); 31 = Renzidong Cave (China; ca 2.0 Ma); 32 = Xiashagou, Classic Nihewan (China; ca 1.8-1.6 Ma); 33 = Jinyuan Cave Lower fauna (China; ca 2.0-1.8 Ma); 34 = Longguopo (China; ca 1.9 Ma); 35 = Gongwangling (China; ca 1.1 Ma); 36 = Jinyuan Cave Upper fauna (China; ca 1.1 Ma); 37 = Zhoukoudian Locality 1 (China; ca 1.0 Ma).