This is the accepted version of the journal article: Macaluso, Loredana; Bertini, Adele; Carnevale, Giorgio; [et al.]. «A combined palaeomodelling approach reveals the role as selective refugia of the Mediter- ranean peninsulas». Palaeogeography, Palaeoclimatology, Palaeoecology, Vol. 625 (September 2023), art. 111699. DOI 10.1016/j.palaeo.2023.111699 This version is available at https://ddd.uab.cat/record/280759 under the terms of the license 1 A combined palaeomodelling approach reveals the role as selective refugia of the 2 Mediterranean peninsulas 3 4 Loredana Macaluso1,2*, Adele Bertini3, Giorgio Carnevale1, Jussi T. Eronen4, Edoardo Martinetto1, 5 Juha Saarinen5, Andrea Villa6, Flavia Capasso7 and Massimo Delfino1,6 6 1 Dipartimento di Scienze della Terra, Università di Torino, Via Valperga Caluso 35, I-10125 Torino, Italy 7 2 Zentralmagazin Naturwissenschaftlicher Sammlungen, Domplatz 4, 06108 Halle (Saale), Germany 8 3 Dipartimento di Scienze della Terra, Università di Firenze, Via G. La Pira 4, I-50121 Firenze, Italy 9 4 Ecosystems and Environment Research Programme and Helsinki Institute of Sustainability Science (HELSUS) 10 Faculty of Biological and Environmental Sciences, P.O. Box 65 (Viikinkaari 1), 00014 University of Helsinki, 11 Finland 12 5 Department of Geosciences and Geography, P.O.Box 64, FIN-00014 University of Helsinki, Finland 13 6 Institut Català de Paleontologia Miquel Crusafont, Edifici ICTA-ICP, c/ Columnes, s/n, Campus de la UAB, 14 08193 Cerdanyola del Vallès, Barcelona, Spain 15 7 Dipartimento di Economia e Statistica “Cognetti De Martiis”, Università di Torino, Lungo Dora Siena 100A, 16 10153, Torino, Italy 17 *corresponding author, loredana.macaluso@unito.it 18 Declaration of interests. The authors declare that they have no known competing financial interests or personal 19 relationships that could have appeared to influence the work reported in this paper. 20 Data supporting statement: Data supporting the results are available in the Supplementary Material of this study. 21 Running title: Climate and extirpation of herpetofauna 22 Keywords: Ecological Niche Modelling, fossils, herpetofauna, palaeobotany, hypsodonty of mammals, 23 conservation palaeobiology 24 Author contributions. LM, MD, AV and GC conceived the study and collected data. LM and FC performed the 25 ENM analyses, AB and EM provided palaeobotanical data, JE and JS performed the hypsodonty analyses. LM, 26 MD and AB wrote the first draft of the manuscript, and all authors contributed substantially to revisions. 1 27 Abstract 28 The northern Mediterranean reptiles and amphibians show contrasting biogeographic 29 histories during the Plio-Pleistocene. The influence of European climate changes on the 30 evolution of the biogeographic ranges of taxa with a rich fossil record is determined herein 31 combining different proxies to obtain well-supported palaeoclimatic scenarios. Ecological 32 Niche Modelling through the ensemble modelling approach was used to reconstruct the 33 climatic niche of selected reptile and amphibian genera, combining modern occurrences with 34 seven independent bioclimatic variables. The obtained model was then projected on past 35 scenarios using fossil data to crosscheck the validity of the climatic explanation for the 36 selective extirpations. Palaeobotanical proxies and the hypsodonty of mammals were used to 37 reconstruct the humidity pattern of the last million years in the studied area. The climate of 38 the Italian Peninsula during the Last Glacial Maximum was proved herein to be unsuitable for 39 many reptile taxa, which persisted in this area until the Early or Late Pleistocene, whereas 40 they survived till the present day in the Iberian and/or Balkans peninsulas. On the contrary, 41 all the proxies agree that Italy was comparably more humid than the other peninsulas during 42 the Pliocene and Pleistocene, therefore allowing the prolonged survival of amphibian taxa 43 (including the still living Salamandrina and Speleomantes), which became extinct or 44 extirpated elsewhere in Europe. An intensification of the conservation efforts concerning 45 amphibians, particularly sensitive to the global climate change, is recommended in the Italian 46 Peninsula, which acted as an amphibian refugium for Europe due to its peculiar climatic 47 history. 48 1. Introduction 49 The latitudinal patterns of biodiversity of European organisms are currently well known, 50 whereas the longitudinal patterns have only been marginally explored (Atkinson et al., 2007). 2 51 In the southern part of Europe, this is mainly reflected in a generally similar role as 52 Quaternary biodiversity refugia attributed to the three major Mediterranean peninsulas (from 53 West to East: the Iberian, the Italian, and the Balkan peninsulas). It is generally assumed that 54 these three peninsulas, together with the adjoining Anatolian one, acted as refugia during the 55 Pleistocene global cooling (see Schmitt et al., 2021 for a recent biogeographic analysis of the 56 Italian Peninsula). However, comparative analyses investigating how this happened in the 57 different European peninsulas are far from being exhaustive (see different contributions and 58 literature cited in Weiss & Ferrand, 2007). 59 The Italian geographic province (sensu Lanza & Corti, 1996, which includes not only the 60 political region of Italy but also part of southeastern France, Corsica, and the Istria region) 61 currently has the most diversified amphibian fauna of the Mediterranean Region and 62 comprises the highest number of endemic species of Europe: about 50% of the amphibian 63 species of this province is endemic (Sindaco et al., 2006; Speybroeck et al., 2020). 64 Furthermore, the presence of three endemic genera of urodeles (Euproctus, Salamandrina 65 and Speleomantes) is of exceptional relevance. Such high rate of endemism contrasts with the 66 particularly low rate of endemic reptiles (17%; see fig. 5 in Sillero et al., 2014), with only one 67 endemic genus and comparatively few endemic species, mainly restricted to islands (Bologna 68 & Mazzotti, 2006; Sindaco et al., 2006; Speybroeck et al., 2020). 69 Why does the Italian geographic province comprise so many amphibian endemisms and such 70 a low number of endemic reptiles? And did other European Mediterranean peninsulas 71 undergo different dynamics that ultimately led to different patterns of diversity? Did climate 72 have a major role in shaping these diverging patterns? Answering to such questions is 73 important not only to understand the geographic distribution and the evolutionary history of 74 these taxa but also for conservation purposes, as time is running out and conservation efforts 3 75 are limited and often need to be prioritised. A first attempt to answer such questions has been 76 done by the recent work of Macaluso et al. (2021a), in which the multivariate approach of 77 Ecological Niche Modelling (ENM) was used to evaluate the possible climatic influence on 78 the extirpation of the Italian endemic genus Salamandrina. Climate changes during the last 79 3.3 million years were possibly responsible for its extirpation from most of Europe, whereas 80 it still survives in Italy, during all the chosen time bins. The present work is conceived as a 81 prosecution and strengthening of the conclusions of Macaluso et al. (2021a), applying a 82 similar approach to European taxa of amphibians and reptiles that showed significant 83 reductions or shifts of their geographic ranges during the Neogene and Quaternary to better 84 answer the questions presented at the beginning of this paragraph and further understand the 85 possible different roles of the three southern European peninsulas. After a synthesis of the 86 distribution of fossil remains from Miocene to Late Pleistocene, climate is evaluated as 87 possible driver for the selective extirpations of different herp taxa through ENM. Given the 88 uncertainty of palaeoclimatic reconstructions, two different palaeontological approaches are 89 used in tandem with ENM to support or contradict the results of the models. First, the 90 palaeobotanical record of Europe is used to qualitatively find possible analogies between 91 floral composition, vegetation structure, and herpetological distributional patterns, as fossil 92 plants are largely considered good indicators of the climatic conditions of the past (see e.g. 93 Combourieu Nebout et al., 2015; Martinetto et al., 2017 and references therein). Second, we 94 use mean ordinated hypsodonty of large herbivorous mammals from Pliocene to Pleistocene 95 European fossil localities as a proxy of precipitation in the corresponding terrestrial 96 environments (Fortelius et al., 2002, 2003, 2006; 2016; Eronen et al., 2010; Žliobaitė et al., 97 2016, 2018; Oksanen et al., 2019; Saarinen et al., 2021). Mean ordinated hypsodonty of large 98 herbivorous mammal communities correlates with annual precipitation, even in extant 99 terrestrial environments (Eronen et al., 2010; Žliobaitė et al., 2016, 2018), because of factors 4 100 that favour hypsodonty in dry environments where the combined effect of air-transported 101 mineral dust and phytolith-rich vegetation causes increased wear on the teeth of herbivorous 102 mammals (Fortelius et al., 2002; Damuth & Janis, 2011; Kaiser et al., 2013; Madden, 2015; 103 Sanson et al., 2017). 104 2. Material and Methods 105 2.1 Selection of taxa 106 In this study, we selected taxa for ENM analyses that are characterized by selective 107 extirpations in Europe, testified by a well-documented fossil record. The fossil remains of the 108 selected taxa should be easily identifiable at least at genus level, so that their past distribution 109 deduced from the fossil record is not affected by misinterpretations and erroneous 110 identifications. Due to scarce or uncertain fossil record, the following taxa were excluded: the 111 anuran Discoglossus (whose fossils, if fragmentary, are easily misinterpreted as Latonia or 112 viceversa due to the uniformity of their skeletal morphology; see Biton at al., 2013, 2016); 113 the large ‘Oriental vipers’ possibly related to Macrovipera and Montivipera; the soft-shelled 114 turtles (Tryonyx); the monitor lizards (Varanus); the gecko Euleptes (that not only has a very 115 poor fossil record, but whose only extant species E. europea is considered to have originated 116 in Central Mediterranean islands and to be currently in expansion; Delaugerre et al., 2011; 117 Daza et al., 2014; Villa & Delfino, 2019a; Villa et al., 2022). 118 The fossil record of the European herpetofauna was assessed by the analysis of the 119 fosFARbase (Böhme & Ilg, 2003), the Paleobiology Database (McClennen et al., 2017), and 120 the Italian Palaeoherpetofaunas Database (Delfino, 2002 [updated to 2022]), combined with 121 literature (e.g. Delfino, 2003, 2004; Delfino et al., 2005; Venczel & Sanchiz, 2005; Chesi et 5 122 al., 2007, 2009; Blain et al., 2009; Georgalis et al., 2018a, b, 2019; Villa et al., 2018a, b; Villa 123 & Delfino, 2019a; Collareta et al., 2020; Syromyatnikova et al., 2021; Loréal et al., 2023). 124 Modern occurrences of extant taxa were obtained from the Global Biodiversity Information 125 Facility (www.GBIF.org), including in the query all "human observations" of the European 126 species of the selected genera (see below), with coordinates uncertainty equal or less than 127 5000 meters, to match with the spatial resolution of the current bioclimatic variables (2.5 128 arc/min) as suggested by Sillero & Barbosa (2021). Data point occurrences downloaded from 129 GBIF included all the available occurrences of the following taxa: Pseudopus apodus 130 (https://doi.org/10.15468/dl.m8au23), Mauremys leprosa, M. caspica, and M. rivulata 131 (https://doi.org/10.15468/dl.zjesfb), Blanus spp. (https://doi.org/10.15468/dl.nndkny), 132 Chioglossa lusitanica (https://doi.org/10.15468/dl.nn9pmz), Speleomantes spp. 133 (https://doi.org/10.15468/dl.u4w87f), Hierophis cypriensis, H. gemonensis, and H. 134 viridiflavus (https://doi.org/10.15468/dl.fzefep). Data about the species of Blanus from the 135 Anatolian region were obtained also from the recently published paper by Şahin et al. (2021), 136 whereas data about Speleomantes were added using all records of the extant species of this 137 genus available in the atlas database of the Societas Herpetologica Italica (SHI). The sand 138 boa Eryx was excluded from the analysis because public data concerning its current 139 distribution are evidently incomplete. 140 The data points were subsequently projected using QGIS 3.28 (QGIS, 2022) and compared 141 with the current distribution reported by Sindaco and Jeremčenko (2008) and Sindaco et al. 142 (2013); occurrences outside the current distribution of the taxa were excluded from the 143 dataset. Furthermore, duplicates were removed, and spatial thinning was used to space 144 occurrence records a defined distance from each other to avoid spatial autocorrelation 145 (Aiello-Lammens et al., 2015). The total of kept occurrences per each genus after the 6 146 cleaning procedure reported above is as follows: Pseudopus, 509 occurrences; Mauremys, 147 1737 occurrences; Blanus, 854 occurrences; Chioglossa, 284 occurrences; Speleomantes, 526 148 occurrences; Hierophis, 7800 occurrences. 149 2.2 Ecological Niche Modelling 150 Nineteen bioclimatic variables (see Appendix S1 for a complete list) for the current (1979– 151 2013) climate scenario were downloaded from PaleoClim (Brown et al., 2018) at a resolution 152 of 2.5 arc-minutes. Bioclimatic variables were clipped at the European scale. Pearson’s 153 pairwise correlation coefficients were calculated with R (R Core Team, 2013) to measure the 154 autocorrelation between the variables. To reduce co-correlation between variables and 155 prevent overfitting, we retained the combinations of climate variables with a Pearson’s 156 coefficient of less than 0.75 (see Appendix S2 and S3). Seven independent variables were 157 selected: temperature seasonality, mean temperature of the wettest quarter, mean temperature 158 of the warmest quarter, mean temperature of the coldest quarter, precipitation seasonality, 159 precipitation of the wettest quarter, and precipitation of the warmest quarter. The same 160 independent bioclimatic variables were downloaded for different time bins, chosen as 161 representatives of different climatic phases during the Plio-Pleistocene: 162  Last Glacial Maximum (LGM; ca. 21,000 years BP; Karger et al., 2021); 163  Last Interglacial (LIG; ca. 130,000 years BP; Otto-Bliesner et al., 2006); 164  MIS 19, Pleistocene (Brown et al., 2018); 165  mid-Piacenzian warm period (3.205 Ma; Dowsett et al., 2013; Hill, 2015); 166  Marine Isotope Stage M2, Pliocene (ca. 3.3 Ma; Dolan et al., 2015). 167 Climatic niches and projections on past climatic scenarios were estimated using the Ensemble 168 Modelling approach through the ‘biomod2’ package of R v. 4.2-2 (Thuiller et al., 2014). As 7 169 all the geographic range of the genera is covered, 1000 pseudo-absences were randomly 170 chosen within the background pixels for a total of five replications. We implemented five 171 ENM methods, including regression (MARS, multivariate adaptive regression splines; 172 Friedman, 1991), classification (GBM, generalized boosting models; Friedman et al., 2000) 173 and complex models (ANN, artificial neural networks; Thuiller et al., 2009; Random Forest; 174 Liaw & Wiener, 2002; MaxEnt v. 3.4.1; Phillips et al., 2017). Model tuning was performed 175 using the following R packages: ‘SDMtune’ v. 1.3.0 (Vignali et al., 2020) for GBM, ANN, 176 and RF; ‘ENMeval’ v. 2.0.4 (Kass et al., 2021) for MaxEnt; ‘caret’ v. 6.0.94 (Kuhn, 2008) 177 for MARS. Tuning results and settings and the selected hyperparameters (based on higher 178 testing Area Under the ROC Curve) are reported in Appendix S4. Cross-validation was done 179 based on 10 repetitions of data splitting (70% of data used for training and 30% used for 180 testing). Model performance was assessed using: the true skill statistic (TSS), the receiver 181 characteristic operator (ROC), and the continuous Boyce index. In all these cases, evaluation 182 scores span from -1 to +1: positive values indicate a model presenting predictions that are 183 consistent with the presence distribution in the evaluation dataset, whereas values lower than 184 0 are considered worse than a random model (Allouche et al. 2006). For the ensemble 185 forecasting maps, the selection criteria are TSS values, with a proportional weights method 186 (i.e. best-performing models weighting more in the construction of the ensemble forecasting 187 maps). The maps obtained as output from the software show the potential climatic niche and 188 were modified using QGIS 3.28 (QGIS, 2022). To allow the discussion of the influence of 189 certain bioclimatic variables, the Multivariate Environmental Similarity surface method, or 190 MESS (Elith et al. 2010), was also used in selected time bins (see below). 191 2.3 Hypsodonty of large mammals 8 192 The humidity patterns for Pliocene to Pleistocene in Europe and the Mediterranean realm 193 were evaluated analysing the hypsodonty of large mammals for 5.3 – 3.5 Ma (Early 194 Pliocene), 3.5 – 2.6 Ma (Late Pliocene), 2.6 – 1.8 Ma (early Early Pleistocene, 1.8 – 0.781 195 Ma (late Early Pleistocene), 0.781 – 0.425 Ma (early Middle Pleistocene), and 0.425 – 0.13 196 Ma (late Middle Pleistocene). Following the methodology introduced by Fortelius et al. 197 (2002), we use ordinated hypsodonty values for large herbivorous mammal species: 1 198 (brachydont = low-crowned), 2 (mesodont = medium-crowned), and 3 (hypsodont = high- 199 crowned). We present the locality mean ordinated hypsodonty-based humidity estimates as 200 colour-interpolated hypsodonty maps following the methodology of Fortelius et al. (2002) 201 and Eronen et al. (2010). For interpolations, we used MapInfo 12.5 for thematic mapping and 202 grid interpolation, with the following settings: 30 km grid size; 600 km search radius; 600 203 grid border. For the interpolation method, we used an inverse distance weighted algorithm 204 (IDW). The colour codes in the hypsodonty maps range from blue to red, with blue indicating 205 humid conditions (very low mean ordinated hypsodonty close to 1), green indicating sub- 206 humid conditions (mean ordinated hypsodonty around 1.4), yellow indicating intermediate 207 humidity (mean ordinated hypsodonty around 1.8), orange indicating relatively dry 208 conditions (mean ordinated hypsodonty around 2.2), and red indicating arid conditions (mean 209 ordinated hypsodonty around 2.6 and higher). For the hypsodonty maps, we used data from 210 the NOW database (www.now.org, data downloaded 17.06.2022) for the Early and Late 211 Pliocene, with and addition of Red Crag Nodule Bed from UK for the Late Pliocene based on 212 published faunal lists and taxonomic revisions (Lister, 1999; Rivals & Lister, 2016). For the 213 Early and Middle Pleistocene, we used mean ordinated hypsodonty data mainly from 214 Saarinen et al. (2021), with the following additions: 9 215  Red Crag and Norwich Crag (UK), Ahl Al Oughlam (Morocco), and Ain Boucherit, 216 Ain Jourdel, and Saint Arnaud (Algeria) for the early Early Pleistocene (2.6 – 1.8 217 Ma); 218  East Runton Early Pleistocene assemblage (UK), Tighenif (Morocco) and Ain Hanech 219 (Algeria) for the late Early Pleistocene (1.8 – 0.781 Ma); 220  Kolkotova Balka / Tiraspol (Moldova), Petralona Cave (Greece), and Trlica levels 221 TRL 5-7 (Bosnia-Herzegovina) for the early Middle Pleistocene (0.781 – 0.425 Ma); 222  Marathousa 1 (Greece) and Treugol´naya Cave (Russia) for the late Middle 223 Pleistocene (0.425 – 0.13 Ma). 224 The references for these additions can be found in Appendix S5. In addition to these, the 225 following updates to stratigraphic ages of fossil large mammal localities were made 226 compared to information in the NOW-database: 227  early Early Pleistocene: Blassac la Girondie, France (Paquette et al., 2021), Villaroya, 228 Spain (Saarinen et al., 2021; Cirilli et al., 2023), Tsalka, Georgia (Kahlke et al., 229 2011), and Gulyazi, Turkey (Kostopoulos and Sen, 1999) 230  late Early Pleistocene: Peyrolles (Valli et al., 2006; Nomade et al., 2014) 231 2.4 Palaeobotanical data 232 The qualitative analyses of the floral composition of Europe and the consequent climatic 233 deductions from pollen data discussed in the present paper refer to the results of published 234 palynological studies (see in 3.3.) sometimes including multi-method approaches too (e.g. 235 Climatic Amplitude Method, Modern Analog Technique, and Weighted Averaging Partial 236 Least-Squares Regression). The latter have been especially applied to several pollen sequences 237 of the Mediterranean area to provide quantitative estimates of climate and bioclimate 10 238 parameters (e.g. Fauquette et al., 1998, 1999, 2006; Fauquette & Bertini, 2003; Combourieu 239 Nebout et al., 2009, 2015; Bertini et al., 2010; Sinopoli et al., 2019). 240 3. Results 241 3.1 Fossil record of the herpetofauna 242 Under a literal reading of the fossil record of extant taxa, a strong disparity between the 243 amphibians and reptiles that inhabit the Mediterranean peninsulas is clearly rooted in the past 244 (Table 1). Few taxa display a reduction of the geographic range at northern latitudes, but they 245 survived in all the peninsulas where they were already present in origin (i.e. they were not 246 extirpated from any peninsula where they were living). Among these taxa, the urodeles 247 Speleomantes and Chioglossa were distributed up to central Europe during Miocene and they 248 currently survive only in the Italian and Iberian peninsulas respectively, being extirpated 249 from northern latitudes (Fig. 1A-B, 2; Venczel & Sanchiz, 2005; Blain et al., 2009; Villa et 250 al., 2018a). The snake Hierophis was present in northern areas of Europe during the Pliocene 251 (Szyndlar, 2012), but currently shows a more southern distribution throughout northwestern 252 Europe and the Apennine and Balkan Peninsulas (Appendix S6); concerning the Iberian 253 Peninsula, this genus is now present only at its northeastern edge (Sillero et al., 2014) and it 254 apparently never had a broader Iberian range (Böhme and Ilg, 2003). 255 The fossil record of other taxa, on the other hand, testifies for an uneven survival at southern 256 latitudes as well. Some taxa were present in two or three of the Mediterranean peninsulas and 257 currently only survive in one of them, whereas some others were previously present in all of 258 the three peninsulas and currently persist in two of them (Table 1). More in detail, according 259 to the fossil record, there is no evidence for range shrinkage of reptile taxa that were once 260 widespread in Europe and now exclusively survive in the Italian province (except for the 11 261 above-discussed case of Euleptes; see Villa et al., 2022, for a discussion of the fossil record 262 and biogeographic patterns of the gekkotan lineage leading to the European leaf-toad gecko). 263 Indeed, the latter geographic unit lost in the last one or two million years at least three genera 264 that still inhabit other northern Mediterranean peninsulas (Delfino, 2002, 2003, 2006; Chesi 265 et al., 2007; Delfino et al., 2008): that is, the legless lizard Pseudopus (Fig. 3A, still present in 266 a very restricted area of the North-East of the Italian geographic province, an area somehow 267 overlapping with the geographic definition of the Balkan Peninsula), the terrapin Mauremys 268 (Fig. 3B), the amphisbaenian Blanus (Fig. 3C). The large ‘Oriental vipers’ possibly related to 269 Macrovipera and Montivipera, soft-shelled turtles (Tryonyx), and monitor lizards (Varanus) 270 may be further included in this context of losses for the Italian province. However, their poor 271 (or poorly recognizable) fossil record makes them unsuitable for the analyses presented 272 herein. 273 On the contrary, two of the three Italian endemic urodeles, Salamandrina and Speleomantes, 274 had a larger spatial range in the past that, at least for the former, extended from Iberia to the 275 Balkans during the Miocene (Böhme, 2003; Venczel & Sanchiz, 2005; Delfino, 2006; 276 Delfino & Sala, 2007; see also Vieites et al., 2007; Macaluso et al., 2021b). Italy also hosted 277 one of the last European occurrences of the giant alytid frog Latonia, which went extirpated 278 before the end of the Pliocene in the rest Europe but survived until the Early Pleistocene in 279 central Italy (Rage & Rocek, 2003; Sorbelli et al., 2021) and until the Middle Pleistocene in 280 North Hungary (Szentesi, 2019). Latonia only survives in a very restricted, humid area in the 281 Middle East today (Biton et al., 2013; Perl et al., 2017). The allocaudate Albanerpeton also 282 represents a clade of amphibians that was thought to go extinct in the Early Pliocene, but was 283 found still thriving in an Early Pleistocene locality in northern Italy (Venczel & Gardner, 284 2005; Delfino & Sala, 2007; Villa et al., 2018a; Gardner et al., 2021). 12 285 3.2 Ecological Niche Modelling and hypsodonty of mammals 286 The ENM run using ‘biomod2’ resulted in the ensemble models reported in Figures 1H, 2D, 287 4A-C.5, and Appendix S6. The average values of the evaluation scores for each algorithm are 288 reported in Appendix S7. Based on average TSS (ranging from 0.86 and 0.986), ROC 289 (ranging from 0.966 and 0.992), and Boyce index (ranging from 0.61 and 0.964) values for 290 testing data (Appendix S8), the models are well-supported. Projections on past climate 291 scenarios are reported in Figure 1C-G, 2A-C, 5A-C.1-5, and Appendix S6. The three 292 Mediterranean peninsulas clearly show different suitability levels during the different time 293 bins, depending on the taxon. This difference in suitability is widely commented in the 294 Discussion section. 295 The series of hypsodonty maps indicates that dry climatic conditions were present in the 296 Iberian and Balkan peninsulas already during the Pliocene, and that they became even drier 297 during the Pleistocene, with the exception of the northeast of the Iberian peninsula (Fig. 5). In 298 contrast, Italy remained relatively more humid throughout the Early Pliocene to late Middle 299 Pleistocene period, with mean hypsodonty values being mostly closer to those of central and 300 northwestern Europe, compared to the relatively more arid Iberian and Balkan peninsulas. In 301 general, the mean ordinated hypsodonty values mostly indicate a latitudinal aridity pattern, 302 with relatively more humid conditions north of the Alps and drier conditions in the 303 Mediterranean realm from the Pliocene to the Pleistocene (see also Fortelius et al., 2002, 304 2003, 2006). Italy stands out as markedly more humid than the rest of the Mediterranean 305 realm throughout the Pliocene to late Middle Pleistocene interval. 306 3.3 Clues from palaeobotany 13 307 Palaeobotanical data from peri-Mediterranean sedimentary successions point out the key role 308 of southern European latitudes as vegetation refugia under the effects of palaeoenvironmental 309 and palaeoclimatic changes during the late Neogene and Quaternary (e.g. Tzedakis, 2007; 310 Svennings et al., 2008; Tzedakis et al., 2009; Médail & Diadema, 2009; Donders et al., 2021; 311 Bertini & Combourieu-Nebout, 2023). The Mediterranean flora and vegetation suffered their 312 major changes (e.g. prominent development of steppe) only after the Messinian Salinity 313 Crisis, in coincidence with the phase of maximal expansion of the Arctic glaciation and the 314 start of glacial/interglacial (G/I) cycles (duration 40 ky) at about 2.6 Ma. The longitudinal 315 temperature gradient was specifically attested by the preservation of thermophilous elements 316 in the eastern Mediterranean areas under the effects of the Asiatic monsoon (e.g. Suc & 317 Popescu, 2005). In southeastern Spain, aridity is documented, in the Middle Miocene and 318 from the Zanclean to the Gelasian (e.g. Carrión et al., 2010, Jimenez-Moreno et al., 2010, and 319 references therein), by the high percentage of herbs in pollen spectra (e.g. 320 Poaceae, Asteraceae with Artemisia, Amaranthaceae), which also include some subdesertic 321 taxa such as Nitraria, 322 Neurada, Prosopis, Lygeum, Ephedra, Caesalpiniaceae, Acacia, and Calligonum. A thermal 323 threshold, according to palynological evidence and climatic quantifications on Messinian and 324 Zanclean sites, was traced between Barcelona and Tarragona. A similar threshold also occurs 325 today, separating thermo-Mediterranean from meso-Mediterranean formations (Fauquette et 326 al., 1999). In North Africa and in southern European sites (Sicily), similar conditions were 327 detected (Suc & Bessais, 1990; Bertini et al., 1998; Fauquette et al., 2006), with assemblages 328 dominated by herbs (e.g. Poaceae, Asteraceae, Plantago, Erodium) including sub-desertic 329 plants (e.g. Lygeum, Nitraria, Calligonum, and Neurada), nowadays typical of the southern 330 Mediterranean region. On the contrary, in sites with latitude higher than 42°, prevalent warm 331 and humid conditions were palynologically attested by wet, subtropical to temperate (sensu 14 332 Trewartha-Köppen; see Belda et al., 2014) forest taxa as well as by the paucity of herbs 333 during the Messinian, including the MSC (e.g. Bertini, 2006; Bertini & Martinetto, 2011, 334 2014), and the Pliocene (in particular from the mid-Piacenzian warm period; Bertini & 335 Roiron, 1997; Bertini, 2010, 2013; Martinetto et al., 2017, Vieira & Zetter, 2020). At a 336 European scale, a major plant extirpation corresponding to the Piacenzian-Gelasian transition 337 (Martinetto, 1999; Martinetto et al., 2017; Suc et al., 2018) was probably caused by the 338 absence of sufficiently humid refugia in southern areas. However, unlike the rest of Europe, a 339 few thermophilous plants requiring wet summers (Cephalotaxus, Toddalia, Sequoia, 340 Symplocos subgen. Hopea) managed to survive in some areas such as central Italy, at least, 341 until the Gelasian (Martinetto, 2001; Martinetto et al., 2017). In central Europe, on the other 342 hand, the foregoing taxa are documented by macrofossils just up to the Piacenzian. During 343 the Quaternary, in central and southern Europe, alternations of Artemisia steppe and 344 thermophilous forest marked the overall glacial-interglacial vegetation changes, especially 345 along the Mediterranean littoral. The subtropical forest disappeared from the Mediterranean 346 area at ca 1.2 Ma, together with a reduction of other forests and a spreading of a steppe 347 ecosystem. However, in peculiar areas (e.g. northern Italy), the alternations are rather 348 expressed by a spread of altitudinal coniferous forest (mainly Picea), without significant 349 expansion of steppe vegetation, followed by the typical reestablishment of thermophilous 350 forest (Bertini, 2001, 2010, and references therein). The Picea-dominated forests during 351 Early Pleistocene glacial phases are documented in northern Italy at the foothills of both 352 Apennines (Emilia-Romagna region) and Alps (Lombardia and Piemonte regions), where 353 also the macroflora testifies for the persistence, even during cooler vegetational phases, of 354 forest types dominated by Picea and some deciduous angiosperms still with distinct exotic 355 affinity (Carya, Magnolia, Phellodendron, Symplocos sect. Palura; Cavallo & Martinetto, 356 2002). These plants evidently found wet and rather ‘warm’ (Mean Coldest Month T > 0°C) 15 357 refugial conditions in southern Europe until the Early-Middle Pleistocene transition 358 (Martinetto & Sami, 2001; Denk et al., 2022). In both southern Italy and the eastern 359 Mediterranean, some pollen records indicate the spread of alpine vegetation during glacial 360 phases (e.g. in southern Italy) and the settlement of wooded steppes in earlier phases of 361 interglacials (see references in Bertini, 2010). The Early-Middle Pleistocene climate 362 transition (Head & Gibbard, 2015) and the instauration of G/I cycles of 100 kyrs gave rise to 363 new major changes in the floristic and vegetational assemblages, which progressively 364 attained a modern aspect. A general decline in winter temperature and annual precipitation 365 from the Early Pleistocene to the Holocene, with main changes around 2.0 Ma, 1.4-1.3 Ma, 366 and 0.5 Ma, was detected for the Italian Peninsula (Combourieu-Nebout et al., 2015). During 367 the last 0.3 Ma, the reduction in moisture availability and/or the increase in seasonal drought 368 intensity promoted a further general reduction of those habitats that had been previously more 369 largely distributed in the Mediterranean (Combourieu-Nebout et al., 2015; Magri et al., 370 2017). 371 4. Discussion 372 The fossil record indicates that several taxa composing the extant southern European 373 herpetofauna had a much wider distribution during the Miocene and Pliocene. The 374 geographic ranges of three genera have been identified herein as having been affected by a 375 southern contraction, without an extirpation from the peninsulas where they were already 376 present: these taxa are the two urodeles Speleomantes (Fig. 1) and Chioglossa (Fig. 2), and 377 the snake Hierophis (Appendix S6). The former two taxa are currently endemic respectively 378 of the Italian geographic province (as defined by Lanza & Corti, 1996) and of the Iberian 379 Peninsula, but during the Miocene they were present in central Europe as well (Macaluso et 380 al., 2022). The third genus, Hierophis, shows a somehow similar southern contraction of the 16 381 range (Appendix S6). Climate seems to be an important driver of these geographic shifts. The 382 persistence of Hierophis in central-eastern Europe during the Late Pliocene is supported by 383 the high climatic suitability during the mid-Piacenzian Warm Period (mPWP, Appendix S6). 384 ENM analyses highlight that the climatic suitability of central Europe is low or zero from the 385 Late Pliocene onward for the two urodele genera (Fig. 1-2). Similarly to what was concluded 386 for Salamandrina by Macaluso et al. (2021a), the lack of a Pliocene fossil record out from 387 their current geographic range prevents the detection of a direct connection between the Late 388 Pliocene/Pleistocene climate and the extirpation of these urodeles, as it is possible that their 389 geographic range was already reduced at that time. The increasing aridity during the Early 390 Pliocene (as well as during the Messinian; Fig. 5) calculated through the hypsodonty analyses 391 suggests that the eradication probably started during this time interval. The same hypothesis 392 is also supported for southeastern Spain by the presence of abundant herbs (including 393 subdesertic taxa documenting aridity) from the Zanclean to the Gelasian. 394 The analysis of the European fossil record of the herpetofauna in the interval Miocene-Recent 395 clearly indicates a prolonged survival in the Italian Peninsula of amphibians that went 396 extirpated or extinct elsewhere: not only Speleomantes and Salamandrina, but also the frog 397 Latonia (Sorbelli et al., 2021) and the enigmatic amphibian Albanerpeton (Gardner et al., 398 2021). At the same time, this geographic region shows a progressive impoverishment in 399 reptiles that in many cases still inhabit the other northern Mediterranean peninsulas. This is 400 the case of Pseudopus, currently only present in the Balkans and in Anatolia but occupying in 401 the past also the Italian Peninsula (Fig. 3), and of Mauremys and Blanus, that went extirpated 402 from Italy but survived in Iberia and in the Balkans. 403 The reason of this asymmetry can be found in the climatic evolution of Europe during the 404 Plio-Pleistocene. In fact, the climate of the Italian Peninsula was already proved to be 17 405 particularly suitable for Salamandrina, especially when compared with the lower suitability 406 shown by the other two peninsulas (Macaluso et al., 2021a). The present work provides 407 similar data concerning Speleomantes (Fig. 1). Noteworthy is the fact that both cited urodeles 408 are characterized by the absence or the reduction of lungs and are therefore invariably tied to 409 cool and humid environments (Lanza, 1983; Lanza et al., 1995) to maintain the gas exchange 410 rate within a physiologically acceptable range (Spotila, 1972). It could be deduced that the 411 Italian geographic province provided at least locally and seasonally a high level of moisture 412 allowing the survival of these amphibians during the late Cenozoic. The evaluation of 413 humidity patterns in Europe and the Mediterranean area from the Early Pliocene to the 414 Middle Pleistocene based on mean ordinated hypsodonty of large herbivorous mammal 415 communities indicates that Italy remained consistently less dry than the Iberian and Balkan 416 peninsulas throughout this period (Fig. 5). The Picea-dominated forests during Early 417 Pleistocene glacial phases in northern Italy further support this hypothesis: as a matter of fact, 418 they indicate the presence of wet conditions in contrast to what is seen in the rest of Europe, 419 where habitats were dominated by steppe. Interestingly, plant remains testifying these 420 climatic conditions were found at the foothills of the Apennines (Emilia-Romagna region; 421 Fauquette & Bertini, 2003), not far from the area currently showing the highest endemism 422 rate for the amphibians (the contact zone between Alps and Apennines; fig. 5 of Sillero et al., 423 2014). A refugium with similar condition was present in the northwestern coast of Iberia for 424 the salamander Chioglossa, also characterized by lung reduction (Lanza, 1983; García-París 425 et al., 2004; Fig. 2). Wet climate is documented by the presence of Ericaceae along the 426 Atlantic coast of the Iberian Peninsula and by the persistence of plants requiring wet and 427 humid summers in the Italian Peninsula until the Gelasian. The rather humid conditions of 428 Italy are also confirmed by high-resolution palynological data from 6 to 21 ka (Bartlein et al., 429 2011). Higher humidity during the Pliocene and Quaternary in the Italian Peninsula compared 18 430 to Iberia and Balkans are confirmed by all the proxies used in this work. Despite being more 431 humid than the other peninsulas, it is essential to underline that during the Pleistocene Italy is 432 still to be considered dry on average (Fig. 5), explaining the fact that some of the amphibian 433 genera still went extinct (in the case of Albanerpeton) or extirpated (in the case of Latonia) 434 during this time bin, probably due to more strict physiological requirements related to 435 humidity or other environmental factors. 436 Conversely, the Italian Peninsula was climatically unsuitable for the reptile taxa studied 437 herein, especially during the Last Interglacial and the Last Glacial Maximum (Fig. 4). More 438 in detail, the suitability of the legless lizard Pseudopus through time is highly compatible 439 with its fossil record, as the distribution of this genus is strongly reduced in the Italian 440 Peninsula during the LGM and the last fossil occurrences are Late Pleistocene in age (Fig. 441 4A), supporting the hypothesis of a climatic explanation for its extirpation. The suitability in 442 certain areas of the Balkans and in Anatolia is high in all the examined time bins. 443 Interestingly, Iberia seems suitable as well (Fig. 4A), but fossil record suggests that 444 Pseudopus did not reach the southern part of the peninsula during Plio-Pleistocene (Fig. 4), 445 for some unrelated reason. 446 The turtle Mauremys was present in all the three Mediterranean peninsulas from Miocene to 447 Late Pleistocene and now only survives in the Balkans and Iberian ones (Fig. 4B). The 448 persistence of suitable climatic conditions for this taxon in Italy until the Late Pleistocene is 449 demonstrated by the ENM analyses. During the LIG a good suitability is present in restricted 450 areas, including the one where the last Italian fossil occurrence was found (in Latium; Chesi 451 et al. 2007). Subsequently, during the LGM, the Italian suitability drops (with this peninsula 452 being almost entirely white in the map; Fig. 4B.5), whereas suitable areas persist in the 453 southern part of the other two peninsulas. Interestingly, there is no time bin where the 19 454 suitability of Sardinia is low to the point of explaining the extirpation of Mauremys from this 455 island, suggesting that other factors may have acted on this (like interspecific competitions or 456 predations). Climate had also a limited impact on the persistence of Speleomantes in Sardinia, 457 as it seems that its climate was unsuitable for this genus during the LGM (when this genus 458 was already present on the island). In fact, these salamanders are highly resistant to external 459 unfavourable climatic conditions because of their lifestyle, as they inhabit subterranean 460 systems (Vignoli et al., 2008). 461 As far as the amphisbaenian Blanus (Fig. 4C) is concerned, the suitability in southern Italy 462 drops earlier than for the other two reptile genera Pseudopus and Mauremys, being already 463 low during the MIS19 (Fig. 4C.3) and becoming especially low during the Late Pleistocene 464 (Fig. 4C.4-C.5). Once again, this supports a climatic explanation for the extirpation of this 465 genus from the area, considering that the last Italian fossil occurrence is from the Early 466 Pleistocene of southeastern Italy. Concerning this genus, it is important to underline that due 467 to the problematic identification at genus level of fossils of amphisbaenians represented 468 exclusively by vertebrae (Villa & Delfino, 2019b; Villa et al., 2019), the rarity of fossils in 469 western Europe (where fossil amphisbaenians are known and somehow abundant but not 470 identified at the genus level) should not be considered as evidence of absence of Blanus. 471 The data presented herein therefore suggest that the absence of suitable climate during the 472 Last Glacial Maximum (or other rather recent glacial phases) caused the extirpation of 473 reptilian taxa from Italy, whereas the relatively warmer climate of the southern parts of Iberia 474 and Balkans and western part of Anatolia allowed the persistence of reptiles in these latter 475 peninsulas. These results are consistent with the high genetic diversity of Mauremys leprosa 476 in southern Iberia (Fritz et al., 2006) and of Pseudopus apodus in western Anatolia (Jandzik 477 et al., 2021). In particular, an important role might have been played by the colder summer 20 478 temperature in the Italian province during the LGM. In fact, based on the MESS, the 479 temperature of the warmest quarter of the year of the Italian province during the LGM is the 480 most different variable compared to the extant variables in the areas associated with the 481 occurrence points (Appendix S9). This parameter might have been important for reptiles as 482 the temperature during development is specifically important for the physiologic equilibrium 483 of the embryos (Singh et al., 2020), especially for genera with Temperature-Dependent Sex 484 Determination such as Mauremys (Okada et al., 2010) and for generally thermophilus taxa 485 such as Blanus and Pseudopus. 486 Interestingly, a distribution similar to the one of Blanus and Mauremys is shown by the 487 cyprinid freshwater genus Messinobarbus (Bianco, 1998), suggesting that the patterns 488 highlighted herein might be valid for other groups of ectothermic vertebrates as well. 489 5. Conclusions 490 Despite them being generally grouped together as southern refugia during the Plio- 491 Pleistocene climatic oscillations, this work, focused on longitudinal differences, demonstrates 492 that the Iberian, Italian, and Balkan peninsulas acted as selective refugia for different taxa of 493 the herpetofauna. Climate was herein proved as an important driver that led to such 494 biogeographic pattern (with the exception of the island of Sardinia). The Italian geographic 495 province was unsuitable during the LGM for some reptile genera (Blanus, Mauremys, and 496 Pseudopus), that persisted in the Italian Peninsula until the Early or Late Pleistocene, without 497 surviving up to the current day. The Balkan (and the adjacent Anatolian) and the Iberian 498 peninsulas, especially when it comes respectively to their eastern and southern portions, had a 499 better role as refugia for the considered reptiles, which is consistent with both fossil record 500 and genetic analyses. Conversely, the Italian geographic province acted as selective refugium 501 for amphibians (Albanerpeton, Latonia, Salamandrina, Speleomantes) thanks to its higher 21 502 humidity during the Pliocene and Pleistocene, as highlighted by ENM and hypsodonty 503 analyses, as well as by the prolonged survival of plants requiring wet summers. Half of these 504 genera are now either extirpated (Latonia) or extinct (Albanerpeton) from the peninsula, 505 possibly due to their stricter physiological requirements in terms of humidity that did not 506 allow their survival during the comparatively drier Pleistocene. However, the persistence in 507 the Italian geographic province of two endemic genera, Salamandrina and Speleomantes, is 508 of exceptional relevance. The future persistence of Salamandrina was already reported to be 509 in danger (Macaluso et al., 2021a) due to the human-induced climate change, and the 510 European cave salamander (Speleomantes) might have to face a similar fate. In an ideal 511 world, no prioritisation would be needed for conservation practices, but, as we are running 512 out of time and the conservation efforts are still too limited, setting conservation priorities is 513 becoming more and more relevant. We suggest that combining ENM and palaeoclimatic 514 approaches with fossil record should help determining priority areas for conservation. In this 515 case study, we demonstrate that, given the unique role Italy played as an amphibian European 516 refugium during Pliocene and Pleistocene times, the conservation efforts concerning this 517 group of ectothermic tetrapods should be necessarily intensified in this area. 518 Acknowledgements 519 This work was supported by Grants (ex-60% 2018-2020) from the Università degli Studi di 520 Torino. AV is funded by the Agencia Estatal de Investigación through a Juan de la Cierva- 521 Formación grant (FCI2019-039443-I/AEI/10.13039/501100011033). MD acknowledges the 522 Generalitat de Catalunya (CERCA Program), and the Spanish Agencia Estatal de 523 Investigación (CGL2016-76431-P, AEI/FEDER, EU). JS wishes to acknowledge Academy of 524 Finland (project nr. 340775). LM thanks P. Mannion, L. Jones, and A. A. Chiarenza 525 (University College London) for providing practical and theoretical help concerning the 22 526 methodology and S. Salvidio for critical comments on the literature. The authors also thank 527 three anonymous reviewers for the constructive comments on a previous draft. The 528 "Commissione Atlante" (Atlas Commission) of the Societas Herpetologica Italica provided 529 the distribution data of extant Speleomantes spp. on the UTM grid. 530 References 531 Aiello‐Lammens, M.E., Boria, R.A., Radosavljevic, A., Vilela, B. & Anderson, R.P. (2015) 532 spThin: an R package for spatial thinning of species occurrence records for use in 533 ecological niche models. Ecography, 38(5), 541–545. 534 Allouche, O., Tsoar, A. & Kadmon, R. (2006) Assessing the accuracy of species distribution 535 models: Prevalence, kappa and the true skill statistic (TSS). Journal of Applied 536 Ecology, 43, 1223–1232. 537 Atkinson, R.J., Rokas, A. & Stone, G.N. (2007) Longitudinal patterns in species richness and 538 genetic diversity in European oaks and oak gallwasp. In: Weiss, S. & Ferrand, N. (eds) 539 Phylogeography of southern European refugia. Evolutionary perspectives on the 540 origins and conservation of European biodiversity. Springer, Berlin Heidelberg New 541 York, pp 127–151. 542 Bartlein, P. J., Harrison, S. P., Brewer, S., Connor, S., Davis, B. A. S., Gajewski, K. et al. 543 (2011) Pollen-based continental climate reconstructions at 6 and 21 ka: a global 544 synthesis. Climate Dynamics, 37(3), 775–802. 545 Belda, M., Holtanová, E., Halenka, T. & Kalvová, J. (2014) Climate classification revisited: 546 from Köppen to Trewartha. Climate Research, 59 (1), 1–13. 23 547 Bertini, A. (2001) Pliocene climatic cycles and altitudinal forest development from 2.7 Ma in 548 the Northern Apennines (Italy): evidence from the pollen record of the Stirone section 549 (∼ 5.1 to∼ 2.2 Ma). Geobios, 34(3), 253–265. 550 Bertini, A. (2006) The Northern Apennines palynological record as a contribute for the 551 reconstruction of the Messinian palaeoenvironments. Sedimentary Geology, 188, 235– 552 258. 553 Bertini, A. (2010) Pliocene to Pleistocene palynoflora and vegetation in Italy: state of the 554 art. Quaternary International, 225(1), 5–24. 555 Bertini, A. (2013) Climate and vegetation in the Upper Valdarno Basin (central Italy) as a 556 response to Northern Hemisphere insolation forcing and regional tectonics in the late 557 Pliocene-early Pleistocene. Italian Journal of Geosciences, 132(1), 137–148. 558 Bertini, A. & Martinetto, E. (2011) Reconstruction of vegetation transects for the Messinian- 559 Piacenzian of Italy by means of comparative analysis of pollen, leaf and carpological 560 records. Palaeogeography, Palaeoclimatology, Palaeoecology, 304 (3-4), 230–246. 561 Bertini, A. & Martinetto, E. (2014) The Neogene flora of the Italian peninsula and Sicily. In: 562 Kustatscher, E., Roghi, G., Bertini, A., Miola, A (eds.): Palaeobotany of Italy. Trento, 563 Naturmuseum Sudtirol, pp. 248–279. 564 Bertini, A. & Roiron, P. (1997) Evolution de la végétation et du climat pendant le Pliocène 565 moyen, en Italie centrale : apport de la palynologie et de la macroflore à l’étude du 566 bassin du Valdarno supérieur (coupe de Santa Barbara). Comptes rendus de l’Académie 567 des sciences. Série 2. Sciences de la terre et des planètes, 324(9), 763–771. 24 568 Bertini, A. & Combourieu-Nebout N. (2023) Piacenzian to Late Pleistocene flora and 569 vegetation in Italy: a moving sketch. Alpine and Mediterranean Quaternary, 36 (1), 570 2023, 1–29 (http://amq.aiqua.it). https://doi.org/10.26382/AMQ.2023.05 571 Bertini, A., Londeix, L., Maniscalco, R., Di Stefano, A., Suc, J. P., Clauzon, G. et al. (1998) 572 Paleobiological evidence of depositional conditions in the Salt Member, Gessoso- 573 Solfifera Formation (Messinian, Upper Miocene) of Sicily. Micropaleontology, 413– 574 433. 575 Bertini, A., Magi, M., Mazza, P. P. & Fauquette, S. (2010) Impact of short-term climatic 576 events on latest Pliocene land settings and communities in Central Italy (Upper 577 Valdarno basin). Quaternary International, 225(1), 92–105. 578 Biton, R., Geffen, E., Vences, M., Cohen, O., Bailon, S., Rabinovich, R. et al. (2013) The 579 rediscovered Hula painted frog is a living fossil. Nature Communications, 4, 1959. 580 Biton, R., Boistel, R., Rabinovich, R., Gafny, S., Brumfeld, V. & Bailon S. (2016) 581 Osteological observations on the alytid Anura Latonia nigriventer with comments on 582 functional morphology, biogeography, and evolutionary history. Journal of 583 Morphology, 277, 1131–1145. 584 Bianco, P.G. (1998) Diversity of Barbinae fishes in southern Europe with description of a 585 new genus and a new species (Cyprinidae). Italian Journal of Zoology, 65(1), 125–136. 586 Blain, H.A., López-García, J.M., Cuenca-Bescós, G., Alonso, C., Vaquero, M. & Alonso, S. 587 (2009). Première mise en évidence fossile du chioglosse portugais Chioglossa 588 lusitanica (Amphibia, Caudata) et son implication pour l’histoire biogéographique de 589 l’espèce. Comptes Rendus Palevol, 8(8), 693–703. 25 590 Böhme, M. (2003) Miocene Climatic Optimum: evidence from Lower Vertebrates of Central 591 Europe. Palaeogeography, Palaeoclimatology, Palaeoecology, 195(3–4), 389–401. 592 Böhme, M. & Ilg, A. (2003) FosFARbase, www.wahre-staerke.com/ (accessed December 2, 593 2021). 594 Bologna, M. A., & Mazzotti, S. (2006). Analisi biogeografica. Atlante degli anfibi e dei rettili 595 d’Italia. SHI, Polistampa, Firenze, 655–678. 596 Brown, J.L., Hill, D.J., Dolan, A.M., Carnaval, A.C. & Haywood, A.M. (2018) PaleoClim, 597 high spatial resolution paleoclimate surfaces for global land areas. Scientific data, 5(1), 598 1–9. 599 Carrión, J. S., Fernández, S., Jiménez-Moreno, G., Fauquette, S., Gil-Romera, G., González- 600 Sampériz, P. & Finlayson, C. (2010) The historical origins of aridity and vegetation 601 degradation in southeastern Spain. Journal of arid environments, 74(7), 731–736. 602 Cavallo, P. & Martinetto, E. (2001) Flore carpologiche del Pliocene di Castelletto Cervo 603 (Biella). Bollettino del Museo Regionale di Scienze Naturali Torino, 18(2), 277–343. 604 Chesi, F., Delfino, M., Abbazzi, L., Carboni, S., Lecca, L. & Rook, L. (2007) New fossil 605 vertebrates from San Giovanni in Sinis (Late Pleistocene, Sardinia): the last Mauremys 606 (Reptilia, Testudines) from the central Mediterranean area. Rivista Italiana di 607 Paleontologia e Stratigrafia, 113, 287–297 608 Chesi, F., Delfino, M. & Rook, L. (2009) Late Miocene Mauremys (Testudines, 609 Geoemydidae) from Tuscany (Italy): evidence of terrapin persistence after a mammal 610 turnover. Journal of Paleontology, 83(3), 379–388. 26 611 Cirilli, O., Pandolfi, L., Alba, D.M., Madurell-Malapeira, J., Bukhsianidze, M., Kordos, L., 612 Lordkidpanidze, D., Rook, L. & Bernor, R. (2023) The last Plio-Pleistocene hipparions 613 of Western Eurasia. A review with remarks on their taxonomy, paleobiogeography and 614 evolution. Quaternary Science Reviews, 306, 107976. 615 Collareta, A., Casati, S., Zuffi, M.A. & Di Cencio, A. (2020) First authentic record of the 616 freshwater turtle Mauremys from the upper Pliocene of Italy, with a new occurrence of 617 the rarely reported ichnotaxon Thatchtelithichnus holmani. Carnets Geol, 20(16), 301. 618 Combourieu Nebout, N., Peyron, O., Dormoy, I., Desprat, S., Beaudouin, C., Kotthoff, U. & 619 Marret, F. (2009) Rapid climatic variability in the west Mediterranean during the last 25 620 000 years from high resolution pollen data. Climate of the Past, 5(3), 503–521. 621 Combourieu-Nebout, N., Bertini, A., Russo-Ermolli, E., Peyron, O., Klotz, S., Montade, V. et 622 al. (2015) Climate changes in the central Mediterranean and Italian vegetation 623 dynamics since the Pliocene. Review of Palaeobotany and Palynology, 218, 127–147. 624 Damuth, J. & Janis, C.M. (2011) On the relationship between hypsodonty and feeding 625 ecology in ungulate mammals, and its utility in palaeoecology. Biological Reviews, 86, 626 733–758. 627 Daza, J.D., Bauer, A.M. & Snively, E.D. (2014) On the fossil record of the Gekkota. The 628 Anatomical Record, 297, 433–462. 629 Delaugerre, M., Ouni, R. & Nouira, S. (2011) Is the European leaf-toed gecko Euleptes 630 europaea also an African? Its occurrence on the Western Mediterranean landbridge 631 islets and its extinction rate. Herpetology Notes, 4, 127–137. 27 632 Delfino, M. (2002) Erpetofaune italiane del Neogene e del Quaternario. PhD Dissertation, 633 Università degli Studi di Modena e Reggio Emilia. 634 Delfino, M. (2003) A Pleistocene amphisbaenian from Sicily. Amphibia-Reptilia, 24, 407– 635 414. 636 Delfino, M. (2004) The Middle Pleistocene herpetofauna of Valdemino Cave (Liguria, North- 637 Western Italy). Herpetological Journal, 14, 113–128. 638 Delfino, M., Razzetti, E. & Salvidio, S. (2005) European plethodontids: palaeontological data 639 and biogeographical considerations. Annali del Museo Civico di Storia Naturale 640 Giacomo Doria, 97, 45–58. 641 Delfino, M. (2006) Fossil records of modern reptile and amphibian species in Italy. Pp 96– 642 119. In: Sindaco, R., Doria, G., Razzetti, E. & Bernini, F. (eds), Atlas of Italian 643 Amphibians and Reptiles. Firenze, Societas Herpetologica Italica, Edizioni Polistampa. 644 Delfino, M. & Sala, B. (2007) Late Pliocene Albanerpetontidae (Lissamphibia) from Italy. 645 Journal of Vertebrate Palaeontology, 27, 716–719. 646 Delfino, M., Kotsakis, T., Arca, M., Tuveri, C., Pitruzzella, G. & Rook, L. (2008) Agamid 647 lizards from the Plio-Pleistocene of Sardinia (Italy) and an overview of the European 648 fossil record of the family. Geodiversitas, 30(3), 641–656. 649 Denk, T., Sami, M., Teodoridis, V. & Martinetto, E. (2022) The late Early Pleistocene flora 650 of Oriolo, Faenza (Italy): assembly of the modern forest biome. Fossil Imprint, 78(1), 651 217–262. 28 652 Dolan, A.M., Haywood, A.M., Hunter, S.J., Tindall, J.C., Dowsett, H.J., Hill, D.J. & 653 Pickering, S.J. (2015) Modelling the enigmatic late Pliocene glacial event—Marine 654 Isotope Stage M2. Global and Planetary Change, 128, 47–60. 655 Donders, T., Panagiotopoulos, K., Koutsodendris, A., Bertini, A., Mercuri, A. M., Masi, A. et 656 al. (2021) 1.36 million years of Mediterranean forest refugium dynamics in response to 657 glacial–interglacial cycle strength. Proceedings of the National Academy of 658 Sciences, 118(34), e2026111118.Eronen, J.T. (2006) Eurasian Neogene large 659 herbivorous mammals and climate. Acta Zoologica Fennica, 216, 1–72. 660 Dowsett, H.J., Foley, K.M., Stoll, D.K., Chandler, M.A., Sohl, L.E., Bentsen, M. et al. (2013) 661 Sea surface temperature of the mid-Piacenzian ocean: a data-model comparison. 662 Scientific Reports, 3(1), 1–8. 663 Elith, J., Kearney, M. & Phillips, S. (2010) The art of modelling range‐shifting species. 664 Methods in ecology and evolution, 1(4), 330–342. 665 Eronen, J.T., Puolamaki, K., Liu, L., Lintulaakso, K., Damuth, J. et al. (2010) Precipitation 666 and large herbivorous mammals II: application to fossil data. Evolutionary Ecology 667 Research, 12, 235–48. 668 Fauquette, S. & Bertini, A. (2003) Quantification of the northern Italy Pliocene climate from 669 pollen data: evidence for a very peculiar climate pattern. Boreas, 32(2), 361–369. 670 Fauquette, S., Quezel, P., Guiot, J., & Suc, J. (1998) Bioclimatic signification of marker taxa 671 of the Mediterranean Pliocene. Geobios, 31(2), 151–169. 29 672 Fauquette, S., Suc, J. P., Guiot, J., Diniz, F., Feddi, N., Zheng, Z. et al. (1999) Climate and 673 biomes in the West Mediterranean area during the Pliocene. Palaeogeography, 674 Palaeoclimatology, Palaeoecology, 152(1-2), 15–36. 675 Fauquette, S., Suc, J.P., Bertini, A., Popescu, S.M., Warny, S., Taoufiq, N.B. et al. (2006) How 676 much did climate force the Messinian salinity crisis? Quantified climatic conditions from 677 pollen records in the Mediterranean region. Palaeogeography, Palaeoclimatology, 678 Palaeoecology, 238(1–4), 281–301. 679 Fortelius, M., Eronen, J.T., Jernvall, J., Liu, L., Pushkina, D., Rinne, J. et al. (2002) Fossil 680 mammals resolve regional patterns of Eurasian climate change during 20 million years. 681 Evolutionary Ecology Research, 4, 1005–1016. 682 Fortelius, M., Eronen, J.T., Liu, L., Pushkina, D., Tesakov, A., Vislobokova, I.A. & Zhang, 683 Z. (2003) Continental-scale hypsodonty patterns, climatic palaeobiogeography, and 684 dispersal of Eurasian Neogene large mammal herbivores. In: Reumer, J.W.F. & 685 Wessels, W. (eds) Distribution and migration of Tertiary mammals in Eurasia. A 686 volume in honour of Hans de Bruijn. Deinsea, 10, 1–11. 687 Fortelius, M., Eronen, J.T., Liu, L., Pushkina, D., Tesakov, A., Vislobokova, I.A. & Zhang, 688 Z. (2006) Late Miocene and Pliocene large land mammals and climatic changes in 689 Eurasia. Palaeogeography, Palaeoclimatology, Palaeoecology, 238, 219–227 690 Fortelius, M., Zliobaité, I., Kaya, F., Bibi, F., Bobe, R., Leakey, L., Leakey, M., Patterson, 691 D., Rannikko, J., Werdelin, L. (2016) An ecometric analysis of the fossil mammal 692 record of the Turkana Basin. Philosophical Transactions of the Royal Society B, 371, 1– 693 13. 30 694 Friedman, J. H. (1991) Estimating functions of mixed ordinal and categorical variables using 695 adaptive splines. Stanford Univ CA Lab for Computational Statistics. 696 Friedman, J., Hastie, T. & Tibshirani, R. (2000) Additive logistic regression: a statistical view 697 of boosting (with discussion and a rejoinder by the authors). The annals of statistics, 698 28(2), 337–407. 699 Fritz, U., Barata, M., Busack, S.D., Fritzsch, G. & Castilho, R. (2006) Impact of mountain 700 chains, sea straits and peripheral populations on genetic and taxonomic structure of a 701 freshwater turtle, Mauremys leprosa (Reptilia, Testudines, Geoemydidae). Zoologica 702 Scripta, 35, 97–108. 703 García-París, M., Montori, A. & Herrero, P. (2004) Fauna Iberica. Museo Nacional de 704 Ciencia Naturales, Consejo Superior de Investigaciones Cientificas, Madrid. 705 Gardner, J.D., Villa, A., Colombero, S., Venczel, M. & Delfino, M. (2021) A Messinian 706 (latest Miocene) occurrence for Albanerpeton Estes & Hoffstetter, 1976 (Lissamphibia: 707 Albanerpetontidae) at Moncucco Torinese, Piedmont Basin, northwestern Italy, and a 708 review of the European Cenozoic record for albanerpetontids. Geodiversitas, 43(14), 709 391–404. 710 GBIF.org (21 September 2022) GBIF Occurrence Download [for DOIs see Material and 711 methods] 712 Georgalis, G.L., Villa, A. & Delfino, M. (2018a). The last amphisbaenian (Squamata) from 713 continental Eastern Europe. Annales de Paléontologie, 104(2), 155–159. 31 714 Georgalis, G.L., Halaçlar, K., Mayda, S., Kaya, T. & Ayaz, D. (2018b). First fossil find of the 715 Blanus strauchi complex (Amphisbaenia, Blanidae) from the Miocene of Anatolia. 716 Journal of Vertebrate Paleontology, 38(2), e1437044. 717 Georgalis, G.L., Villa, A., Ivanov, M., Vasilyan, D. & Delfino, M. (2019). Fossil amphibians 718 and reptiles from the Neogene locality of Maramena (Greece), the most diverse 719 European herpetofauna at the Miocene/Pliocene transition boundary. Palaeontologia 720 Electronica, 22(3), 1–99. 721 Head, M.J. & Gibbard, P.L. (2015) Early–Middle Pleistocene transitions: linking terrestrial 722 and marine realms. Quaternary International, 389, 7–46. 723 Hill, D.J. (2015) The non-analogue nature of Pliocene temperature gradients. EPSL, 425, 724 232–241. 725 Kahlke, R.-D., García, N., Kostopoulos, D.S., Lacombat, F., Lister, A.M., Mazza, P.P.A, 726 Spassov, N. & Titov, V. (2011) Western Palaearctic palaeoenvironmental conditions 727 during the Early and early Middle Pleistocene inferred from large mammal 728 communities, and implications for hominin dispersal in Europe. Quaternary Science 729 Reviews, 30, 1368–1395. 730 Kaiser, T.M., Müller, D.W.H., Fortelius, M., Schulz, E., Codron, D., Clauss, M. (2013) 731 Hypsodonty and tooth facet development in relation to diet and habitat in herbivorous 732 ungulates: implications for understanding tooth wear. Mammal Review, 43, 34–46. 733 Karger, D.N., Nobis, M.P., Normand, S., Graham, C.H. & Zimmermann, N.E. (2021) 734 CHELSA-TraCE21k v1. 0. Downscaled transient temperature and precipitation data 735 since the last glacial maximum. Climate of the Past Discussion, 2021, 1–27. 32 736 Kass, J.M., Muscarella, R., Galante, P.J., Bohl, C.L., Pinilla‐Buitrago, G.E., Boria, R.A. et al. 737 (2021) ENMeval 2.0: Redesigned for customizable and reproducible modeling of 738 species’ niches and distributions. Methods in Ecology and Evolution, 12(9), 1602–1608. 739 Kostopoulos, D.S. & Sen, S. (1999) Late Pliocene (Villafranchian) mammals from Sarikol 740 Tepe, Ankara, Turkey. Mitteilungen der Bayerischen Staatssammlung für Paläontologie 741 und historische Geologie, 39, 165–202. 742 Kuhn, M. (2008) Building predictive models in R using the caret package. Journal of 743 statistical software, 28, 1–26. 744 Jandzik, D., Jablonski, D., Zinenko, O., Kukushkin, O. V., Moravec, J. & Gvoždík, V. (2018) 745 Pleistocene extinctions and recent expansions in an anguid lizard of the genus 746 Pseudopus. Zoologica Scripta, 47(1), 21–32. 747 Jiménez-Moreno, G., Fauquette, S., & Suc, J. P. (2010) Miocene to Pliocene vegetation 748 reconstruction and climate estimates in the Iberian Peninsula from pollen data. Review 749 of Palaeobotany and Palynology, 162(3), 403–415. 750 Lanza, B. (1983) Guide per il riconoscimento delle specie animali delle acque interne 751 italiane. 27. Anfibi, Rettili (Amphibia, Reptilia). Roma, CNR. 752 Lanza, B. & Corti, C. (1996) Evolution of knowledge on the Italian herpetofauna during the 753 20th century. Boll Mus civ St nat Verona, 20, 373–436. 754 Lanza, B., Caputo, V., Nascetti, G. & Bullini, L. (1995) Morphologic and genetic studies of 755 the European plethodontid salamanders: taxonomic inferences (genus: Hydromantes). 756 Monografie XVI. Museo Regionale di Scienze Naturali, Torino. 33 757 Liaw, A. & Wiener, M. (2002) Classification and regression by randomForest. R News, 2, 758 18–22. 759 Lister, A. (1999) The Pliocene deer of the red crag Nodule bed (UK). Deinsea, 7(1), 215– 760 222. 761 Loréal, E., Syromyatnikova, E.V., Danilov, I.G. & Čerňanský, A. (2023) The easternmost 762 record of the largest anguine lizard that has ever lived–Pseudopus pannonicus 763 (Squamata, Anguidae): new fossils from the late Neogene of Eastern Europe. Fossil 764 Record, 26(1), 51–84. 765 Macaluso, L., Villa, A., Carnevale, G. & Delfino, M. (2021a) Past, present, and future 766 climate space of the only endemic vertebrate genus of the Italian peninsula. Scientific 767 reports, 11(1), 1–8. 768 Macaluso, L., Villa, A., Pitruzzella, G., Rook, L., Carnevale, G. & Delfino, M. (2021b) A 769 progressive extirpation: an overview of the fossil record of Salamandrina 770 (Salamandridae, Urodela). Historical Biology, 33(12), 3723–3740. 771 Macaluso, L., Mannion, P.D., Evans, S.E., Carnevale, G., Monti, S., Marchitelli, D. & 772 Delfino, M. (2022) Biogeographic history of Palearctic caudates revealed by a critical 773 appraisal of their fossil record quality and spatio-temporal distribution. Royal Society 774 Open Science, 9(11), 220935. 775 Madden, R.H. (2015) Hypsodonty in mammals: Evolution, geomorphology and the role of 776 Earth surface processes. Cambridge University Press, UK, 423 p. 34 777 Magri, D., Di Rita, F., Aranbarri, J., Fletcher, W. & González-Sampériz, P. (2017) 778 Quaternary disappearance of tree taxa from Southern Europe: Timing and 779 trends. Quaternary Science Reviews, 163, 23–55. 780 Martinetto, E. (1999) Chronological framing of Pliocene to Early Pleistocene plant macrofossil 781 assemblages from northern Italy. Acta Palaeobotanica, 2, 503–511. 782 Martinetto, E. (2001) The role of Central Italy as a centre of refuge for thermophilous plants 783 in the late Cenozoic. Acta Palaeobotanica, 41, 299–319. 784 Martinetto, E., Momohara, A., Bizzarri, R., Baldanza, A., Delfino, M., Esu, D. & Sardella, R. 785 (2017). Late persistence and deterministic extinction of humid thermophilous plant taxa of 786 East Asian affinity (HUTEA) in southern Europe. Palaeogeography, Palaeoclimatology, 787 Palaeoecology, 467, 211–231. 788 Martinetto, E. & Sami, M. (2001) Paleoflora delle "sabbie gialle" medio-pleistoceniche di 789 Oriolo, presso Faenza (Ravenna). Quaderni di Storia Naturale della Romagna, 14, 1– 790 28. 791 McClennen, M., Jenkins, J. & Uhen, M. (2017) Paleobiology Database. [Accessed on 12 792 December 2021] 793 Médail, F. & Diadema, K. (2009) Glacial refugia influence plant diversity patterns in the 794 Mediterranean Basin. Journal of biogeography, 36(7), 1333–1345. 795 Nomade, S., Pastre, J.F., Guillou, H., Faure, M., Guérin, C., Delson, E., Debard, E., 796 Voinchet, P., & Messager, E. (2014) 40Ar/39Ar constraints on some French landmark 797 Late Pliocene to Early Pleistocene large mammalian paleofaunas: Paleoenvironmental 798 and paleoecological implications. Quaternary Geochronology, 21, 2–15. 35 799 Okada, Y., Yabe, T. & Oda, S. I. (2010) Temperature-dependent sex determination in the 800 Japanese pond turtle, Mauremys japonica (Reptilia: Geoemydidae). Current 801 Herpetology, 29(1), 1–10. 802 Oksanen, O., Žliobaitė, I., Saarinen, J., Lawing, A.M., Fortelius, M. (2019) A Humboldtian 803 approach to life and climate of the geological past: Estimating palaeotemperature from 804 dental traits of mammalian communities. Journal of Biogeography, 2019, 1–17. 805 Otto-Bliesner, B.L., Marshall, S.J., Overpeck, J.T., Miller, G.H. & Hu, A. (2006) Simulating 806 Arctic climate warmth and icefeld retreat in the last interglaciation. Science, 311(5768), 807 1751–1753. 808 Paquette, J.-L., Médard, E., Poidevin, J.-L. & Barbet, P. (2021). Precise dating of middle to 809 late Villafranchian mammalian paleofaunae from the Upper Allier River valley (French 810 Massif Central) using U–Pb geochronology on volcanic zircons. Quaternary 811 Geochronology, 65, 101198. 812 Perl, R.G.B., Gafny, S., Malka, Y., Renan,Woodhams, D.C., Rollins-Smith, L., Pask, J.D., 813 Bletz, M.C., Geffen, E. & Vences, M. (2017) Natural history and conservation of the 814 rediscovered Hula painted frog, Latonia nigriventer. Contributions to Zoology, 86, 11– 815 37. 816 Phillips, S.J., Anderson, R.P., Dudík, M., Schapire, R.E. & Blair, M.E. (2017) Opening the 817 black box: An opensource release of Maxent. Ecography, 40, 887–893. 818 QGIS (2022) QGIS: A Free and Open Source Geographic Information System. (Version 819 3.28). [Computer software] https://www.qgis.org/en/site/index.html 36 820 R Core Team (2013) R: A language and environment for statistical computing. R Foundation 821 for Statistical Computing, Vienna, Austria. http://www.R-project.org/. 822 Rage, J.C. & Rocek, Z. (2003) Evolution of anuran assemblages in the Tertiary and 823 Quaternary of Europe, in the context of palaeoclimate and palaeogeography. Amphibia- 824 Reptilia, 24, 133–250 825 Rivals, F. & Lister, A. M. (2016) Dietary flexibility and niche partitioning of large herbivores 826 through the Pleistocene of Britain. Quaternary Science Reviews, 146, 116–133. 827 Saarinen, J., Oksanen, O., Zliobaite, I., Fortelius, M., DeMiguel, D., Azanza, B. et al. (2021) 828 Pliocene to Middle Pleistocene climate history in the Guadix-Baza Basin, and the 829 environmental conditions of early Homo dispersal in Europe. Quaternary Science 830 Reviews, 268, 107–132. 831 Şahin, M.K., Candan, K., Caynak, E.Y., Kumlutaş, Y. & Ilgaz, Ç. (2021) Ecological niche 832 divergence contributes species differentiation in worm lizards (Blanus sp.) (Squamata: 833 Amphisbaenia: Blanidae) in Mediterranean part of Anatolian peninsula and the 834 Levantine region. Biologia, 76(2), 525–532. 835 Sanson, G.D., Kerr, S. & Read, J. (2017) Dietary exogenous and endogenous abrasives and 836 tooth wear in African buffalo. Biosurface and Biotribology, 3, 211–223. 837 Schmitt, T., Fritz, U., Delfino, M., Ulrich, W. & Habel, J.C. (2021) Biogeography of Italy 838 revisited: genetic lineages confirm major phylogeographic patterns and a pre- 839 Pleistocene origin of its biota. Frontiers in Zoology, 18(1), 34. 840 Sillero, N. & Barbosa, A.M. (2021) Common mistakes in ecological niche models. 841 International Journal of Geographical Information Science, 35(2), 213–226. 37 842 Sillero, N., Campos, J., Bonardi, A., Corti, C., Creemers, R., Crochet, P. A. et al. (2014). 843 Updated distribution and biogeography of amphibians and reptiles of 844 Europe. Amphibia-Reptilia, 35(1), 1–31. 845 Sindaco R. & Jeremčenko, V.K. (2008) The reptiles of the western Palearctic. 1. Annotated 846 checklist and distributional atlas of the turtles, crocodiles, amphisbaenians and lizards 847 of Europe, North Africa, Middle East and Central Asia. Monografie della Societas 848 Herpetologica Italica I. Latina, Belvedere. 580 pp. 849 Sindaco, R., Jeremčenko, V.K. & Venchi, A. (2006) Mediterranean reptiles: state of 850 knowledge, hot spots, areas of endemism, conservation. Pp 101–102. In: Bologna, 851 M.A., Capula, M., Carpaneto, G.M., Luiselli, L., Marangoni, C. & Venchi, A. (eds) 852 Abstracts of the VIth Congress of the Societas Herpetologica Italica (Roma, September 853 27 - October 1 2006). Roma, Stilgrafica. 854 Sindaco, R., Venchi, A. & Grieco C. (2013) The reptiles of the western Palearctic. 2. 855 Annotated checklist and distributional atlas of the snakes of Europe, North Africa, 856 Middle East and Central Asia, with an update to the Vol. 1. Monografie della Societas 857 Herpetologica Italica II. Latina, Belvedere. 543 pp. 858 Singh, S. K., Das, D. & Rhen, T. (2020) Embryonic temperature programs phenotype in 859 reptiles. Frontiers in Physiology, 35. 860 Sinopoli, G., Peyron, O., Masi, A., Holtvoeth, J., Francke, A., Wagner, B., & Sadori, L. 861 (2019). Pollen-based temperature and precipitation changes in the Ohrid Basin 862 (western Balkans) between 160 and 70 ka. Climate of the Past, 15(1), 53–71. 38 863 Sorbelli, L., Villa, A., Gentili, S., Cherin, M., Carnevale, G., Tschopp, E. & Delfino, M. 864 (2021) The Early Pleistocene ectothermic vertebrates of Pietrafitta (Italy) and the last 865 Western European occurrence of Latonia Meyer, 1843. Comptes Rendus Palevol, 866 20(26), 555–583. 867 Speybroeck, J., Beukema, W., Dufresnes, C., Fritz, U., Jablonski, D., Lymberakis, P. et al. 868 (2020). Species list of the European herpetofauna–2020 update by the Taxonomic 869 Committee of the Societas Europaea Herpetologica. Amphibia-Reptilia, 41(2), 139– 870 189. 871 Spotila, J.R. (1972) Role of temperature and water in the ecology of lungless salamanders. 872 Ecological Monographs, 42, 95–125. 873 Suc, J.P. & Bessais, E. (1990) Pérennité d'un climat thermo-xérique en Sicile avant, pendant, 874 après la crise de salinité messinienne. Comptes rendus de l'Académie des sciences. Série 875 2, Mécanique, Physique, Chimie, Sciences de l'univers, Sciences de la Terre, 310(12), 876 1701–1707. 877 Suc, J.P. & Popescu, S. M. (2005) Pollen records and climatic cycles in the North 878 Mediterranean region since 2.7 Ma. Geological Society, London, Special 879 Publications, 247(1), 147–158. 880 Suc, J.P., Popescu, S. M., Fauquette, S., Bessedik, M., Jiménez Moreno, G., Bachiri Taoufiq, 881 N. et al. (2018) Reconstruction of Mediterranean flora, vegetation and climate for the last 882 23 million years based on an extensive pollen dataset. Ecologia mediterranea, 44(2), 53– 883 85. 39 884 Svenning, J.C., Normand, S. & Kageyama, M. (2008) Glacial refugia of temperate trees in 885 Europe: insights from species distribution modelling. Journal of Ecology, 96(6), 1117– 886 1127. 887 Syromyatnikova, E.V., Kovalenko, E.S. & Kaloyan, A. (2021) A fossil record of the Eastern 888 clade of Blanus (Amphisbaenia: Blanidae) from the late Miocene of Ukraine. Geobios, 889 69, 69–75. 890 Szentesi, Z. (2019) Latonia gigantea (Lartet, 1851) remains and other herpetological 891 observations from the middle Pleistocene Esztramos 6 locality (Aggtelek–Rudabánya 892 Mountains, North Hungary). Fragmenta Palaeontologica Hungarica, 36, 141–158. 893 Szyndlar, Z. (2012) Early Oligocene to Pliocene Colubridae of Europe: a review. Bulletin 894 Société géologique de France, 183(6), 661–681. 895 The NOW Community (2022) New and Old Worlds Database of Fossil Mammals (NOW). 896 Licensed under CC BY 4.0. Retrieved 17.6.2022 897 from https://nowdatabase.org/now/database/. 898 Thuiller, W., Lafourcade, B., Engler, R. & Araújo, M. B. (2009) BIOMOD–a platform for 899 ensemble forecasting of species distributions. Ecography, 32(3), 369–373. 900 Thuiller, W., Georges, D. & Engler, R. (2014) biomod2: Ensemble platform for species 901 distribution modelling. R package version 3.1–64. http://CRAN.R- 902 project.org/package=biomod2. 903 Tzedakis, P.C. (2007) Seven ambiguities in the Mediterranean palaeoenvironmental 904 narrative. Quaternary Science Reviews, 26(17-18), 2042–2066. 40 905 Tzedakis, P.C., Raynaud, D., McManus, J.F., Berger, A., Brovkin, V. & Kiefer, T. (2009) 906 Interglacial diversity. Nature Geoscience, 2(11), 751–755. 907 Valli, A.M.F., Caron, J.-B., Debard, E., Guérin, C., Pastre, J.-F. & Argant, J. (2006). Le 908 gisement paléontologique villafranchien terminal de Peyrolles (Issoire, Puy-de-Dôme, 909 France): résultats de nouvelles prospections. Geodiversitas, 28 (2), 297–317. 910 Venczel, M. & Gardner, J.D. (2005) The geologically youngest albanerpetontid amphibian, 911 from the Lower Pliocene of Hungary. Palaeontology, 48, 1273–1300. 912 Venczel, M. & Sanchíz, B. (2005) A fossil plethodontids salamander from the Middle Miocene 913 of Slovakia (Caudata, Plethodontidae). Amphibia-Reptilia, 26, 408–411. 914 Vieira, M. & Zetter, R. (2020) Pollen record of megathermal and mesothermal elements in the 915 late Pliocene from west Portugal revealed by combined light and scanning electron 916 microscopy studies. Grana, 59(1), 114–126. 917 Vieites, D. R., Min, M. S. & Wake, D. B. (2007) Rapid diversification and dispersal during 918 periods of global warming by plethodontid salamanders. Proceedings of the National 919 Academy of Sciences, 104(50), 19903–19907. 920 Vignali, S., Barras, A.G., Arlettaz, R. & Braunisch, V. (2020) SDMtune: An R package to 921 tune and evaluate species distribution models. Ecology and Evolution, 10(20), 11488– 922 11506. 923 Vignoli, L., Caldera, F. & Bologna, M.A. (2008) Spatial niche of the Italian cave salamander, 924 Speleomantes italicus (Dunn, 1923) (Plethodontidae, Amphibia), in a subterranean 925 system of Central Italy. Italian Journal of Zoology, 75(1), 59–65. 41 926 Villa, A., Blain, H.A. & Delfino, M. (2018a) The Early Pleistocene herpetofauna of Rivoli 927 Veronese (Northern Italy) as evidence for humid and forested glacial phases in the 928 Gelasian of Southern Alps. Palaeogeography, Palaeoclimatology, Palaeoecology, 490, 929 393–403. 930 Villa, A., Blain, H.A., van den Hoek Ostende, L.W. & Delfino, M. (2018b) Fossil amphibians 931 and reptiles from Tegelen (Province of Limburg) and the early Pleistocene palaeoclimate 932 of The Netherlands. Quaternary science reviews, 187, 203–219. 933 Villa, A. & Delfino, M. (2019a) Fossil lizards and worm lizards (Reptilia, Squamata) from 934 the Neogene and Quaternary of Europe: an overview. Swiss Journal of 935 Palaeontology, 138(2), 177–211. 936 Villa, A. & Delfino, M. (2019b) A comparative atlas of the skull osteology of European 937 lizards (Reptilia: Squamata). Zoological Journal of the Linnean Society, 187(3), 829– 938 928. 939 Villa, A., Kirchner, M., Alba, D.M., Bernardini, F., Bolet, A., Luján, À.H. et al. (2019) 940 Comparative cranial osteology of Blanus (Squamata: Amphisbaenia). Zoological 941 Journal of the Linnean Society, 185(3), 693–716. 942 Villa, A., Wings O. & Rabi M. (2022). A new gecko (Squamata, Gekkota) from the Eocene 943 of Geiseltal (Germany) implies long-term persistence of European Sphaerodactylidae. 944 Papers in Palaeontology, 8, e1434. 945 Weiss, S. & Ferrand, N. (eds) (2007) Phylogeography of southern European refugia. 946 Evolutionary perspectives on the origins and conservation of European biodiversity. 42 947 Springer, Berlin Heidelberg New York.Willis, K.J. & Whittaker, R.J. (2000) The 948 refugial debate. Science, 287(5457), 1406–1407. 949 Žliobaitė, I., Rinne, J., Toth, A., Mechenich, M., Liu, L. P., Behrensmeyer, A. K., Fortelius, 950 M. (2016) Herbivore teeth predict climatic limits in Kenyan ecosystems. Proceedings of 951 the National Academy of Sciences of the United States of America, 113, 12751–12756. 952 Žliobaitė, I., Tang, H., Saarinen, J., Fortelius, M., Rinne, J., Rannikko, J. (2018) Dental 953 ecometrics of tropical Africa: Linking vegetation types and communities of large plant‐ 954 eating mammals. Evolutionary Ecology Research, 19, 127–147. 955 Captions 956 Figure 1. Distribution of the fossil record of Speleomantes sp. (spanning from Miocene and 957 Pliocene, in A, to Pleistocene, in B) and climatic suitability for the genus during different 958 time bins (from Late Pliocene to present), based on modern occurrences of Speleomantes spp. 959 (reported in H). Abbreviations: LGM = Last Glacial Maximum; LIG = Last Interglacial; M2 960 = Marine Isotope Stage M2; PWP = Pliocene Warm Period. Darker colours show areas with 961 better-predicted conditions. To notice that the Italian Peninsula is consistently suitable for 962 this genus in all the time bins. 43 963 964 44 965 Figure 2. Ensemble model projections on different time bins (from Late Pliocene to present) for 966 Chioglossa spp. and distribution of Miocene fossil record. Current distribution in D. Abbreviations: 967 LIG = Last Interglacial; M2 = Marine Isotope Stage M2. 968 969 970 45 971 Figure 3. Fossil record of reptile taxa that were extirpated from the Italian geographic 972 province (Pseudopus in A, Mauremys in B, and Blanus in C) compared with their current 973 distribution (A-C.2; based on the occurrences used to train the models in this study). Fossil 974 record from Late Pliocene onward is reported in Figure 4. 975 976 46 977 Figure 4. Climatic suitability of Pseudopus (A), Mauremys (B), and Blanus (C) during different time 978 bins (from Late Pliocene to present). Fossil record from of each genus from Late Pliocene on is 979 reported in the small squares within the corresponding time bins. Abbreviations: LGM = Last Glacial 980 Maximum; LIG = Last Interglacial; M2 = Marine Isotope Stage M2; PWP = Pliocene Warm 981 Period. Darker colours show areas with better-predicted conditions. To notice that suitability in the 982 Italian Peninsula drops during the LGM for both Pseudopus and Mauremys (A.5, B.5) and during 983 Early Pleistocene for Blanus (C.3), matching the fossil record and therefore supporting a climatic 984 explanation for their extirpation. 985 47 986 Figure 5. Colour-interpolated maps showing mean ordinated hypsodonty in fossil large 987 herbivorous mammal communities from the Early Pliocene to the late Middle Pleistocene. 988 Blue indicates humid conditions (low mean ordinated hypsodonty, close to 1), red indicates 989 arid conditions (high mean ordinated hypsodonty, close to 2.6 and above). The spectrum in 990 between from green to yellow and orange indicates progressively dryer intermediate climatic 991 conditions. 992 993 48 994 Table 1. Presence data of selected components of the European herpetofauna, showing 995 extirpation patterns testified by the fossil record. Genus Central Iberian Peninsula Italian Balkans Climatic analyses Europe Peninsula Peninsula Past Present Past Present Past Present Past Present Chioglossa X X X This work Speleomantes X X X This work Euleptes X X X No Hierophis X Limited to X X X X This work part close to Pyrenees Salamandrina X X X X X Macaluso et al. 2021a Pseudopus X Limited X X X This work to part close to Pyrenees Mauremys X X X X X X This work Blanus X X X X X This work 996 49