This is the accepted version of the journal article: Grau-Camats, Montserrat; Bertrand, Ornella C.; Prieto, Jérôme; [et al.]. «A Miopetaurista (Rodentia, Sciuridae) cranium from the Middle Miocene of Bavaria (Germany) and brain evolution in flying squirrels». Papers in Palaeontology, Vol. 8, Issue 4 (July-August 2022), art. e1454. DOI 10.1002/spp2.1454 This version is available at https://ddd.uab.cat/record/264443 under the terms of the license 1 A MIOPETAURISTA (SCIURIDAE, RODENTIA) CRANIUM FROM THE 2 MIDDLE MIOCENE OF BAVARIA (GERMANY) AND BRAIN EVOLUTION IN 3 FLYING SQUIRRELS 4 by MONTSERRAT GRAU-CAMATS1, ORNELLA C. BERTRAND2*, JÉRÔME 5 PRIETO3, SERGI LÓPEZ-TORRES4,5,6, MARY T. SILCOX7 and ISAAC 6 CASANOVAS-VILAR1* 1 7 Institut Català de Paleontologia Miquel Crusafont, Universtitat Autònoma de Barcelona, 8 Edifici Z, Carrer de les Columnes, s/n, E-08193 Cerdanyola del Vallès, Barcelona, Spain; 9 e-mails: montse.grau@icp.cat, isaac.casanovas@icp.cat 2 10 School of GeoSciences, University of Edinburgh, Grant Institute, King's 11 BuildingsJames Hutton Road, Edinburgh, EH9 3FE, Edinburgh, Scotland, UK; e-mail: 12 ornella.bertrand@ed.ac.uk 3 13 Department für Geo- und Umweltwissenschaften, Paläontologie, Ludwig-Maximillians- 14 Universität München, Richard-Wagner-Str. 10, 80333 Munich, Germany; e-mail: 15 j.prieto@lrz.uni-muenchen.de 4 16 Institute of Evolutionary Biology, Faculty of Biology, Biological and Chemical 17 Research Centre, University of Warsaw, Żwirki i Wigury 101, 02-089 Warsaw, Poland; 18 e-mail: s.lopez-torres@uw.edu.pl 5 19 Division of Paleontology, American Museum of Natural History, 200 Central Park 20 West, 10024 New York, NY, United States of America 6 21 New York Consortium in Evolutionary Primatology, The Graduate Center, CUNY, 365 22 Fifth Avenue, 10016 New York, NY, United States of America 1 7 23 Department of Anthropology, University of Toronto Scarborough, Toronto, ON, M1C 24 1A4, Canada; e-mail: mary.silcox@utoronto.ca 25 *Corresponding authors 26 Abstract: Flying squirrels (Sciurinae, Pteromyini) are the most successful group of 27 gliding mammals. However, their fossil record mostly consists of isolated dental remains 28 which provide very limited insights into their paleobiology and evolution. Only recently, 29 the first skeleton of a fossil flying squirrel, belonging to the species Miopetaurista 30 neogrivensis, has been described. It presents all the diagnostic gliding-related postcranial 31 features of its extant relatives and shows that this group has undergone very little 32 morphological change for almost 12 million years. However, the associated cranium is 33 badly crushed, so particular details of the cranial morphology cannot be described. Here 34 we describe a well-preserved cranium of the closely-related Miopetaurista crusafonti 35 from 12.5–12.0 Ma from Bavaria (Germany). Its cranial morphology is found to be almost 36 identical to extant large flying squirrels, even in details such as the position of the 37 foramina. The virtual endocast also shows close affinities to living large flying squirrels 38 in morphology and in the relative volume of different brain regions, showing diagnostic 39 features such as the size reduction of petrosal lobules and olfactory bulbs. However, the 40 encephalization quotient (EQ) and neocortical ratio are lower than observed in extant 41 flying squirrels. EQ is known to increase through time in squirrels, but might also be 42 related to locomotion, as arboreal and gliding squirrels display higher EQs than terrestrial 43 ones. Because Miopetaurista was certainly a glider, its comparatively lower EQ and 44 neocortical size support the existence of an independent trend of increasing EQ and 45 neocortical complexity in this flying squirrel subclade. 2 46 Key words: Rodentia, Sciuridae, Pteromyini, cranial morphology, endocast, 47 encephalization quotient 48 INTRODUCTION 49 Gliding has independently evolved multiple times in mammals. There are at least two 50 different groups of Mesozoic gliding mammaliaforms (volaticotheres and haramyidians; 51 see Meng et al. 2006; Han et al. 2017; Luo et al. 2017), several gliding marsupials 52 (acrobatids, pseudocheirids, petaurids), the colugos (dermopterans), and at least four 53 different rodent families (Jackson 2012; Jackson & Thorington 2012). Gliding rodents 54 include the extant flying squirrels (pteromyins); the scaly-tailed flying squirrels 55 (anomalurids); at least one species of the completely extinct eomyid family from the late 56 Oligocene of Germany (Storch et al. 1996); and one extinct dormouse (glirid) species 57 from the Late Miocene of France (Mein & Romaggi 1991). However, of all gliding 58 mammals, only flying squirrels have achieved a significant diversity and wide 59 geographical distribution, being present in all northern continents. For a long time 60 considered to belong to a family or subfamily of their own (Simpson 1945; McKenna & 61 Bell 1997) molecular analyses have shown that they are indeed a tribe (Pteromyini) within 62 the Holarctic tree squirrels (Sciurinae) (Mercer & Roth 2003; Herron et al. 2004; Steppan 63 et al. 2004; Casanovas-Vilar et al. 2018; see Fig. 1). Even though transitional forms have 64 yet to be found, flying squirrels certainly evolved from tree squirrels and all share 65 specialized wrist anatomy to support and extend the patagium (Thorington 1984; 66 Thorington & Darrow 2000; Thorington et al. 2002). 67 Until very recently the fossil record of flying squirrels (Sciurinae, Pteromyini) 68 consisted solely of isolated cheek teeth and a few mandibular and maxillary fragments 69 (see discussion in Casanovas-Vilar et al. 2018). Their identification was (and for most of 70 them still is) controversial since many of the dental diagnostic characters used to 3 71 recognize extinct flying squirrels are also present in non-gliding species (Thorington et 72 al. 2005). This has led to major disagreements regarding the time of divergence and 73 diversification of the group. While for most mammal clades molecular-based estimates 74 yield older dates than those based on the fossil record (for example for the origin of 75 placental mammal orders; e.g., Springer et al. 2003; Bininda-Emonds et al. 2007; 76 O’Leary et al. 2013; Murphy et al. 2021), for flying squirrels fossil evidence (specifically 77 isolated cheek teeth belonging to the species Hesperopetes thoringtoni) unusually 78 suggested a much older divergence time (ca. 36 Ma; Emry and Korth, 2007) than most 79 molecular phylogenies (ca. 23 Ma; Mercer and Roth, 2003; Steppan et al. 2004). Contrary 80 to dental material, postcranial bones are unassailably diagnostic for the group since they 81 are shaped by the anatomical adaptations related to gliding (Thorington 1984; Thorington 82 & Darrow 2000; Thorington et al. 2002, 2005). The recent description of a partial skeleton 83 of Miopetaurista neogrivensis from the Middle/Late Miocene (11.6 Ma) of Catalonia, 84 Spain, showed that this taxon, previously tentatively assigned to flying squirrels based on 85 cheek tooth morphology, displayed the gliding-related postcranial features observed in 86 extant forms (Casanovas-Vilar et al. 2018). Its diagnostic wrist anatomy further revealed 87 that M. neogrivensis belonged to the subtribe Pteromyina, which today includes large 88 flying squirrels, implying that the two flying squirrel subtribes (Pteromyina and 89 Glaucomyina) had already diverged at that time (Fig. 1). Moreover, it allowed for a 90 recalibration of estimates of flying squirrel divergence time, resulting in an older time 91 frame (late Oligocene, 31–25 Ma instead of 25–20 Ma). This new estimate is further 92 consistent with the age of some of the oldest purported pteromyinan fossils. Perhaps even 93 more importantly, this fossil showed that flying squirrels are a morphologically 94 conservative group. Total evidence phylogenetic analyses recognized Miopetaurista as 95 the sister taxon of the extant giant flying squirrel (Petaurista; Casanovas-Vilar et al. 2018; 4 96 see Fig. 1), with the two genera showing numerous similarities in cranial morphology and 97 a virtually identical postcranial anatomy. The only marked differences were found in the 98 cheek tooth morphology, but the striking morphological similarities in the postcranial 99 skeleton show that giant flying squirrels have undergone little evolutionary change for at 100 least 12 million years so they could be well regarded as ‘living fossils’. Phylogenetic 101 analyses combining morphological and molecular data showed that flying squirrels 102 radiated during the Miocene (mostly between 18 and 15 Ma; Casanovas-Vilar et al. 2018), 103 with several genera reported from Eurasia and North America (Jackson & Thorington 104 2012; Casanovas-Vilar et al. 2018). However, the attribution of all these taxa to the flying 105 squirrel clade (excepting Miopetaurista) is questionable until more diagnostic material 106 (i.e., certain postcranial elements, particularly wrist bones related to the extension of the 107 patagium) is found. 108 The remarkably complete Miopetaurista skeleton recently found in Catalonia 109 included an associated skull, and an additional second cranium was also recovered from 110 a nearby site. Both crania are mostly complete, but one is crushed laterally and the second 111 one dorso-ventrally, so the skull of M. neogrivensis was virtually reconstructed using 112 elements from both specimens that were scaled and repositioned. Even though those two 113 crania were CT-scanned, the internal morphology proved difficult to reconstruct and was 114 therefore not considered by Casanovas-Vilar et al. (2018). Here we describe a partial 115 cranium from the late Middle Miocene (12.5–12.0 Ma, see below) of Gumpersdorf near 116 Marktl (Bavaria, Germany) attributed to Miopetaurista crusafonti, a closely related 117 species. The cranial cavity was infilled with sediment, which created a natural endocast 118 that is partly exposed in the fossil. By means of CT-scanning we were able to reconstruct 119 a virtual endocast of this specimen and observe in detail endocranial morphological 120 features. Even though there are a few studies that comprehensively describe endocranial 5 121 morphology of some Oligocene sciurids (Bertrand et al. 2017, 2018), this is the first time 122 a fossil flying squirrel endocast is studied. The aims of this work are twofold. On the one 123 hand it provides an accurate description of the external and internal cranial morphology 124 for the genus Miopetaurista based on a well-preserved, uncrushed specimen, and thus 125 completing and significantly improving that of Casanovas-Vilar et al. (2018). On the 126 other hand, the endocast is accurately described and compared to that of other extant 127 flying squirrels to reveal if diagnostic brain features are also present. Locomotor mode 128 has been previously related to overall relative brain size and the size of the different brain 129 regions in sciurids (Bertrand et al. 2017, 2018, 2019a, 2021), although this is debated 130 because of the existence of temporal trends in relative brain size in rodents (i.e., 131 encephalization quotient; Bertrand et al. 2019a). Since Miopetaurista was certainly a 132 glider as its extant relatives, we finally discuss whether locomotion or these temporal 133 trends had a greater impact on brain size and in the relative size of the different brain 134 regions. 135 MATERIAL AND METHODS 136 Material, provenance and chronology 137 The described cranium (Figs 2–5, Appendix S1) was found isolated in 1978 in a gravel 138 quarry at the Gumpersdorf site (near Marktl, Bavaria, Germany), located in the North 139 Alpine Foreland Basin (NAFB) or Molasse Basin, situated on the northern side of the 140 Alps. The NAFB is a large foreland basin, 1000 km long by a maximum width of 130 141 km, that expands from western France to Austria. It formed during the uplift of the Alps 142 and served as a sink for the sediments eroded from that mountain range (Abdul Aziz et 143 al. 2010). In the German part of the basin, the sedimentary infill is divided into five main 144 depositional units: the Lower Marine Molasse (early and middle Oligocene); the Lower 6 145 Freshwater Molasse (late Oligocene to Early Miocene); the Upper Marine Molasse (Early 146 Miocene); the Upper Brackish-water Molasse (Early Miocene); and the Upper Freshwater 147 Molasse (OSM: Obere Süẞwassermolasse; Middle to Late Miocene; Prieto and Rummel, 148 2016). The Gumpersdorf site is situated in eastern Bavaria, in the OSM unit, although its 149 correlation to the OSM local biozones (Prieto & Rummel 2016) is uncertain. The cranium 150 was found 2–3 meters below a marl layer, within reworked sediments, that has yielded a 151 diverse paleoflora (Gregor 1982). 152 The age of these deposits is not well constrained because of the lack of 153 biostratigraphically informative fossils. However, Gumpersdorf had been tentatively 154 correlated to Mammal Neogene (MN) zone MN9 (early Vallesian, 11.2–9.9 Ma; MN zone 155 ranges after Hilgen et al. 2012) assuming a chronological proximity to Marktl, a richer 156 site located only a few kilometres to the southwest (Mayr, 1979 and references therein). 157 However, recent biostratigraphic studies (Prieto & Rummel 2016) propose a late MN7+8 158 (late Astaracian, 13.1/12.6–11.2 Ma) age for Marktl. Another key mammal locality of the 159 uppermost OSM is Hammerschmide, which is correlated to the Middle/Late Miocene 160 boundary (11.6 Ma) on the basis of detailed bio- and magnetostratigraphic data (Kirscher 161 et al. 2016). Lithostratigraphic data indicate that Gumpersdorf is a little older than both 162 Marktl and Hammerschmiede, so an age of 12.5 to 12.0 Ma seems a realistic 163 approximation. Therefore, the Gumpersdorf specimen is slightly older than the skeleton 164 of M. neogrivensis recovered at Abocador de Can Mata site ACM/C5-D1 (Casanovas- 165 Vilar et al. 2018). 166 This specimen was briefly described and figured by Fahlbusch (1979) who 167 tentatively attributed it to Miopetaurista crusafonti. It is curated in the SNSB-Bayerische 168 Staatssammlung für Paläontologie und Geologie, (Munich, Germany) with collection 169 number 1978 V 1. 7 170 Three-dimensional data acquisition and reconstruction 171 The specimen was scanned using a Nanotom M (Phoenix X-ray) micro-CT scanner at the 172 Staatliche Naturwissenschaftliche Sammlungen Bayerns – Zoologishe Staatssammlung 173 Museum (Munich, Germany). Scanning parameters used were 130 kV and 120 mA, 174 including 1505 slices with a voxel size of 32 μm. Raw data were imported to Avizo 8.0.1 175 software for segmentation and visualization. Cranial bones and endocast were segmented 176 separately using semiautomatic thresholding tools to remove sediment and bone, 177 respectively (Figs 4–5). 178 Comparative sample 179 Cheek teeth measurements of the studied specimen were compared to those of other 180 Miopetaurista species taken from the literature (Table 1; Grau-Camats et al. 2021, table 181 S1). Cranial anatomy and measurements were compared to original specimens 182 (IPS56468h, IPS88677) and to the virtual reconstruction of M. neogrivensis (Casanovas- 183 Vilar et al. 2018: Fig. 4B) as well as to a representative sample of extant flying squirrels 184 (Fig. 6, Table 2). These comprise the extant Pteromyina Petaurista petaurista (ZMA 185 131418), Eupetaurus cinereus (RMNH 19524), Aeromys tephromelas (24670), Belomys 186 pearsonii (RMNH 56.046) and Pteromys volans (RMNH 40035), and the extant 187 Glaucomyina represented by Hylopetes sagitta (RMNH 15512), Glaucomys volans 188 (RMNH 19786), Glaucomys sabrinus (IPS60584) and Iomys horsfieldii (RMNH 15937). 189 All specimens of the extant species are housed in the Naturalis Biodiversity Center 190 (Leiden, the Netherlands), except for G. sabrinus which comes from the collections of 191 the Institut Català de Paleontologia Miquel Crusafont (ICP). The M. neogrivensis original 192 specimens are also curated at the ICP. 8 193 The virtual endocast of M. crusafonti was compared to already published rodent 194 endocasts (Bertrand et al. 2016a, 2017, 2018, 2019b; Bertrand & Silcox 2016; Grau- 195 Camats et al. 2021, tables S2–S4; Fig. 8). These include two extant flying squirrels 196 belonging to different subtribes: Pe. petaurista (USNM 589079; Fig. 8A) and G. volans 197 (AMNH 240290; Fig. 8B). Extinct squirrels are represented by Cedromus wilsoni 198 (Cedromurinae; USNM 256584; Fig. 8C) from the Orellan (late Oligocene) of the White 199 River Formation (Wyoming, USA); and the early tree squirrel Protosciurus cf. rachelae 200 (Sciurini, Sciurinae; YPM 14737; Fig. 8D) from late early Arikareean (late Oligocene– 201 early Miocene) of the John Day Formation (Oregon, USA). The ischyromyid Paramys 202 delicatus (Paramyinae; AMNH 12506; Fig. 8E) from the Bridgerian (middle Eocene) of 203 the Wind River Formation (Wyoming, USA) as well as other previous published virtual 204 endocasts of Ischyromyidae are also included (Bertrand et al. 2019b). The inclusion of 205 ischyromyids in our study is crucial. Phylogenetic relationships among the Ischyromyidae 206 and with other rodent groups are still debated, but Paramys is generally regarded as one 207 of the most basal rodents (e.g., Korth 1994; Meng et al. 2003; Asher et al. 2019) . Indeed, 208 this group of rodents has a conservative endocranial morphology (Bertrand et al. 2019a) 209 and is therefore a good representation of the plesiomorphic state for rodents and in this 210 case for squirrels. Aplodontiidae, is the sister group to Sciuridae (e.g., see Fabre et al. 211 2012), for which fossil endocasts are known. However, these belong to Prosciurus 212 (Bertrand et al. 2018) and Mesogaulus (Bertrand et al. 2021), dating back to the 213 Oligocene and Early Miocene, respectively, and are already too derived to provide 214 information on the ancestral condition from which sciurids evolved, so they are not 215 considered here. Phylogenetic relationships among the considered taxa, as well as their 216 age is illustrated in Fig. 1. 217 Anatomical terminology and measurement methods 9 218 Dental terminology and measurement methods for squirrel cheek teeth follow Casanovas- 219 Vilar et al. (2015) and references therein. For cranial anatomy, especially cranial 220 foramina, Wahlert (1985, 2000), Wible (2008), Sinitsa et al. (2019) and Wible and 221 Shelley (2020) were used as primary references. The description of muscular insertion 222 areas follows Ball and Roth (1995). Linear cranial measurements are after Nicolas et al. 223 (2008) and Bertrand et al. (2016b) and were taken on physical specimens. Endocast 224 morphology as well as linear and surface measurements follow Bertrand et al. (2016a, 225 2017, 2018, 2019b) and Bertrand and Silcox (2016). Endocranial volumes and surface 226 areas for extant and fossil squirrels and other rodents are taken from Bertrand et al. 227 (2016a, 2017, 2018, 2019b, 2021). Endocast and brain region volumes were calculated 228 using Avizo, 8.0.1. Because the right side of the endocast is missing a significant portion, 229 the total endocast volume was calculated by doubling the volume of the left half of the 230 endocast. The endocast exhibits some deformation, so it was divided by digitally cutting 231 it between the olfactory bulbs, along the superior sagittal sinus, and the midline of the 232 vermis in dorsal view using the ‘volume edit’ module. The neocortical surface area was 233 estimated by selecting the area above the orbitotemporal canal on the left side of the 234 endocast only, and excluding the circular fissure and the confluence of sinuses (=NS1; 235 following Jerison 2012; Long et al. 2015). The rhinal fissure represents the separation 236 existing between the paleo- and the neocortex (Martin 1990). The relationship between 237 the rhinal fissure and the orbitotemporal canal in rodents exists as in many other 238 mammalian orders (e.g., lemurs; Martin 1990). In the illustration of the brain of Sciurus 239 vulgaris (Brauer & Schober 1970), the rhinal fissure is in the same location as the 240 orbitotemporal canal in the virtual endocast of Sciurus carolinensis (Bertrand & Silcox 241 2016: Fig 7c). The selected area in this hemisphere was then doubled. Because the right 242 side is damaged, the total endocranial surface (TS), which includes the olfactory bulb 10 243 length (see Bertrand & Silcox 2016) was estimated by using the left side only as well and 244 doubling it. Volume and surface area ratios are expressed as percentages. 245 Body mass and encephalization quotient estimations 246 Body mass (BM) was estimated using a linear regression of body mass vs. cranial length 247 computed with the software R version 4.0.2 (R Core Team 2020) following the methods 248 of Bertrand et al. (2016a). The encephalization quotient (EQ) was used to compare 249 endocranial volume among species with different body masses. The EQ equation used 250 [EQ = 0.0097 (BM)0.06419] was the one proposed by Pilleri et al. (1984), which is 251 specifically adapted to rodents. Statistical tests and plots were made using R version 4.0.2 252 (R Core Team 2020). For comparison purposes, we included the EQ produced by 253 Eisenberg and Wilson (1978). 254 Phylogenetic signal analysis 255 We performed regressions on the endocranial data using Phylogenetic Generalized Least 256 Squares (PGLS) regressions (Grafen & Hamilton 1989). The regressions were calculated 257 and plotted using R version 4.0.2 (R Core Team 2020). To perform the PGLS analyses, 258 we used the tree topology proposed by Bertrand et al. (2021), based on Korth and Emry 259 (1991) and Mercer andRoth (2003), with the addition of M. crusafonti as sister taxon of 260 Petaurista following Casanovas-Vilar et al. (2018). The phylogenetic tree used is given 261 in Fig. 1. We used Bayesian method to time calibrate our tree with the function 262 “createMrBayesTipDatingNexus” in the paleotree package v3.3.25 (Bapst 2012) and 263 followed the protocol of Bapst (2013, 2014) to create a script that was subsequently ran 264 in the software MrBayes (v3.2.7; Huelsenbeck & Ronquist 2001; Ronquist & 265 Huelsenbeck 2003). We employed the “gls” function of the nlme package (version 3.1- 266 142) to run the PGLS analysis (Pinheiro et al. 2020) and used the Lambda model 11 267 (Brownian motion with internal branches multiplied by Pagel’s λ [lambda]; Pagel 1999). 268 We obtained a total of six final PGLS regressions: (1) endocranial volume in relation to 269 body mass, (2) petrosal lobule volume in relation to body mass, (3) olfactory bulb volume 270 in relation to body mass, (4) neocortex surface area in relation to endocranial surface area, 271 (5) petrosal lobule volume in relation to endocranial volume, (6) olfactory bulb volume 272 in relation to endocranial volume. We computed Pagel’s λ (Pagel 1999) for each PGLS 273 regression, a scaling coefficient that detects whether the shared evolutionary histories as 274 specified by the phylogeny are responsible for the patterns of similarity observed in the 275 data. Values below 1 correspond to traits being less similar amongst species than expected 276 from their phylogenetic relationships, while values above 1 suggest the reverse. We also 277 provide the 95 % confidence intervals for each PGLS regression following the 278 recommendations from Symonds & Blomberg (2014). 279 We generated the predicted and residual values for each PGLS regression using 280 the function “predict” and “residuals” from the stats package (version 3.6-2). The 281 functions “R2.pred” and “R2.lik” from the rr2 package (version 1.0-2) were used to 282 generate the coefficient of correlation for each regression. We decided to report two 283 different estimates of the coefficient of correlation using the R2 obtained from the 284 predicted and from the residual values. The first one is better than other R2 estimators in 285 determining how much variation is explained by the model, while the second one is more 286 useful when assessing the significance of each variable used in the model (Ives 2019). 287 Institutional abbreviations. AMNH, American Museum of Natural History, New York, 288 USA; IPS, Institut Català de Paleontologia Miquel Crusafont, Sabadell, Spain; RMNH, 289 Naturalis Biodiversity Center, Leiden, the Netherlands; SNSB, Bayerische 290 Staatssammlung für Paläontologie und Geologie, Munich, Germany; USNM, National 291 Museum of Natural History, Smithsonian Institution, Washington, D.C., USA; YPM, 12 292 Peabody Museum of Natural History, Yale University, New Haven, USA; ZMA, 293 Zoological Museum of Amsterdam collection, Naturalis Biodiversity Center, Leiden, the 294 Netherlands. 295 296 DESCRIPTIONS AND COMPARISONS 297 Cheek tooth morphology 298 The upper cheek teeth show smooth enamel without lophules; the ridges are relatively 299 simple and thick (Fig. 3E). The metaloph and the protoloph are parallel and do not 300 converge towards the protocone. The posteroloph and anteroloph are lower than the 301 protoloph and metaloph. The protocone is longitudinally elongated. The P4–M3 have a 302 large, antero-posteriorly elongated root under the protocone and two smaller cylindrical 303 roots below the metacone and paracone. 304 P3. It is small and conical, with a single cylindrical root. It has a prominent anterior cusp 305 and a posterior ridge which encloses a small basin. A small posterior cusp is present on 306 the posterior ridge. 307 P4. The right P4 is missing. The left one is subtriangular, longer than wide, and with a 308 well-developed anterior region. There are three main cusps: metacone, paracone and 309 protocone. The hypocone is reduced and integrated into the endoloph, although this cusp 310 is more marked than on the molars. The mesostyle is evident and closes the central valley. 311 The parastyle is labio-lingually elongated and is as large as the main cusps and only 312 slightly less prominent. There are two longitudinal spurs that weakly connect the 313 metaloph with the posteroloph. The anterior valley is relatively wide and labially open, 314 while the posterior one is very narrow. 13 315 M1/M2. Both molars are subrectangular, their width is notably greater than their length. 316 The M2 is relatively longer and squarer than the M1. Each of these molars possesses three 317 main cusps: metacone, paracone and protocone. The M1 has a diminutive mesostyle, 318 whereas it is missing in the M2. The protoloph is constricted near the point where it 319 merges with the protocone. The central valley is wide and labially open. In both molars 320 the metaloph shows a tiny posterior spur that just reaches the posteroloph but does not 321 merge with it. The anterior and posterior valleys are open, the latter being much narrower. 322 M3. It is subtriangular, with a conspicuously narrower posterior half. This tooth has two 323 main cusps: paracone and protocone. The metacone is reduced to a small cusp integrated 324 into a ridge that defines the posterolabial border of the molar. There is a highly reduced 325 postero-lingual valley defined by a thin and arched ridge, presumably corresponding to 326 the metaloph. This minute valley is partially closed by a small cingulum descending from 327 the labial side. As described for the other molars, the protoloph is constricted near the 328 point where it merges with the protocone. The anterior valley is very narrow and shallow 329 while the central valley is wider and deeper. Both valleys are labially open. Cheek teeth 330 measurements are given in Table 1 and comparisons are given in Grau-Camats et al. 331 (2021, table S1). 332 Comparisons and species attribution. The Gumpersdorf specimen is assigned to the genus 333 Miopetaurista because of its large size; the absence of enamel crenulations on the upper 334 cheek teeth; and the protoloph and metaloph of P4–M2 lacking conules and being parallel 335 to one another rather than lingually convergent (Mein 1970; Daxner-Höck & Mein 1975; 336 de Bruijn 1999). The specimen belongs to a medium-sized Miopetaurista species, clearly 337 larger than the Early Miocene species (early to middle Orleanian, MN3–MN4) 338 Miopetaurista diescalidus, Miopetaurista dehmi and Miopetaurista lappi. It further 339 differs from these species by the absence of a mesoloph, and from M. diescalidus in 14 340 particular by the different shape of the P4, which presents a characteristically reduced and 341 more rounded P4 (Daams 1977). In the case of Miopetaurista gibberosa from the Middle 342 Miocene (late Orleanian, MN5) of Göriach (Austria), only lower cheek teeth are known 343 (Daxner-Höck & Höck 2015), but these clearly belong to a smaller species. Younger 344 species include Miopetaurista neogrivensis (late Astaracian to early Vallesian, MN7+8– 345 MN9) and Miopetaurista thaleri (late Turolian to Villanyan, MN13–MN16), which are 346 the largest species of the genus, rivalling in size the extant giant flying squirrels of the 347 genus Petaurista (see Casanovas-Vilar et al. 2018). The material from Gumpersdorf is 348 slightly smaller than that ascribed to these two species and the shape of the P4 is more 349 triangular, being markedly narrower in its lingual part. In addition, the morphology of the 350 upper cheek teeth is more complex in M. thaleri, the M1/M2 always having a short 351 anterior spur in the protoloph (Mein 1970). 352 The Gumpersdorf material closely matches the size of the species Miopetaurista 353 gaillardi (Astaracian, MN6–MN7+8) and Miopetaurista crusafonti (Vallesian to early 354 Turolian, MN9–MN11), although it is closer to the latter. The former is known from 355 several sites ranging from Portugal to Turkey while the latter has previously only been 356 reported from Catalonia and southern France (Casanovas-Vilar et al. 2015). The 357 described material is slightly larger than M. gaillardi, further differing in the shape of the 358 P4, which is more triangular and presents a well-developed parastyle as in M. crusafonti. 359 Other aspects of the P4 morphology resemble these two species, such as the presence of 360 a pronounced mesostyle and two short posterior spurs in the metaloph directed towards 361 the posteroloph (see Mein 1970; Casanovas-Vilar et al. 2015). In M. gaillardi the P4 362 possesses an additional small cusp between the paracone and the parastyle, which is 363 absent in the Gumpersdorf material and in M. crusafonti (Casanovas-Vilar et al. 2015). 364 The size and morphology of the described material supports its ascription to M. crusafonti 15 365 as already suggested by Fahlbusch (1979), even though the M1/M2 may show a more 366 complex morphology in this species, with additional short spurs (Casanovas-Vilar et al. 367 2015). The P3 morphology of M. crusafonti is here described for the first time. As 368 compared to M. neogrivensis the P3 is more rounded and presents two well-defined cusps. 369 The material of Gumpersdorf confirms the occurrence of M. crusafonti in Germany 370 during the latest Middle Miocene (late Astaracian, MN7+8), being the only record of this 371 species in Central Europe. Younger occurrences of the genus Miopetaurista in Germany, 372 such as that from Dorn-Dürkheim 1 (early Turolian, MN11), have not been identified to 373 the species level but apparently correspond to a larger-sized species (Franzen & Storch 374 1975; Franzen et al. 2013). 375 Cranial anatomy and comparisons 376 Description. The cranium is short and wide (Figs 2–5 and Appendix S1). The total 377 estimated length of the cranium is 67.7 mm from the tip of the snout to the posterior part 378 of the braincase (for cranial measurements and comparisons see Table 2). The rostrum is 379 laterally crushed and deviated towards the right side. It is short and appears to have also 380 been wide, although extensive damage to this region obscures its morphology. The nasals 381 and premaxillary bones are crushed, and the different fragments displaced. The left 382 incisor is missing, but the right one is complete, with only minor damage on its tip. The 383 incisor is clearly orthodont and relatively slender. Only the anteriormost border of the 384 right incisive foramen is preserved, so it is impossible to evaluate its shape and size. The 385 infraorbital foramen is small and rounded. The infraorbital canal is long and opens lateral 386 to the rostrum. The orbital region, particularly the left side, is well preserved. The 387 braincase is not notably deformed or crushed but important parts are missing, particularly 388 in the dorsal part of the cranium (i.e., most part of the frontal and the right parietal; Figs 389 3–4, Appendix S1). The zygomatic arches are broken, and only the zygomatic process of 16 390 the squamosal and the ventral half of the zygomatic plate are preserved. Most of the 391 occipital region is crushed, and the right half of the occipital bone is entirely absent. The 392 auditory bullae are missing. The brain cavity is infilled with terrigenous sediment, thus 393 producing an almost complete natural endocast (Fig. 3, Appendix S1). 394 In ventral view, the masseteric tubercles are bulbous and prominent (Fig. 4A). The 395 zygomatic plate is relatively vertical and wide as in other sciurids. A semicircular ridge 396 in front of the cheek teeth marks the insertion area for the anterior portion of the 397 buccinator muscle. The palate is not particularly wide and is pierced by an anterior pair 398 of posterior palatine foramina at the level of the M2. The posterior pair of posterior 399 palatine foramina are just tiny punctures in the maxillary bone close to the suture with the 400 palatine bones. The palatines are pierced by a well-developed, rounded and closed 401 posterior maxillary foramen just posterior to the M3. The posterior margin of the palatines 402 is straight but there is a short and rounded posterior nasal spine. Only the anterior part of 403 the pterygoid ridges is preserved but these appear to have been slightly divergent 404 posteriorly. The left foramen ovale is preserved. It is large, rounded and located 405 anteromedially to the inferred position of the auditory bulla. The lateral flange of the 406 pterygoid encloses the alisphenoid canal, which has moved from its original position. The 407 transverse canal is smaller and medial to the foramen ovale. A fragment of the 408 basisphenoid, a rectangular flat plate, is preserved on the left side. 409 In the orbital region, a rounded and relatively large sphenopalatine foramen is 410 clearly visible in the left side, whereas only its anterior border is preserved in the right 411 side (Fig. 5A). This foramen is located at the level of the anterior edge of the third molar 412 and is mostly included within the maxillary bone. An oval upper ethmoid foramen is well 413 preserved in the left frontal. The lower ethmoid foramen can only be observed in the left 414 frontal and it is somewhat damaged. It is much smaller than the upper foramen and it is 17 415 located more posteriorly, close to the suture with the squamosal. The posterior part of the 416 orbital region comprising the orbitosphenoid, as well as part of the palatine and frontal 417 bones is damaged, so that the optic and dorsal palatine foramina are not preserved. A 418 large and elliptical masticatory foramen (including the buccinator foramen) is visible on 419 both sides, although it is better preserved on the right side. Medial to the masticatory 420 foramen, the edge of a large sphenorbital fissure can also be recognized on the right 421 alisphenoid. Finally, a small and elliptical postglenoid foramen is visible just below the 422 posterior edge of the posterior zygomatic root. This foramen is only preserved on the left 423 side of the cranium. 424 The postorbital processes are not preserved (Fig. 4B). The interorbital distance is 425 relatively narrow compared to other squirrels (Table 2; see also Fig. 6). Even though the 426 zygomatic arch is broken, its posterior root is preserved on both sides of the cranium. It 427 is robust and almost horizontal (Fig. 4B). The cranial vault looks relatively convex 428 because most of the region of the cranium anterior to it is crushed and poorly preserved 429 (Fig. 5A). When considered in isolation, it appears to have been quite flat, thus resembling 430 M. neogrivensis. Only the left half of the parietal is preserved, but it shows a marked 431 temporal ridge for the insertion of the temporalis muscle (Fig. 4B). The posterior half of 432 the ridge curves medially before merging with a marked nuchal crest. 433 As already stated, the auditory bullae are not preserved. Furthermore, the right 434 half of the cranial vault is entirely missing. However, this makes it possible to observe 435 the petrosal, which is well preserved on the left side of the cranium (Fig. 5B, Appendix 436 S1). In medial view, the petrosal crest defines the posterior margin of a large middle 437 cranial fossa that houses the posterior part of the cerebrum (Fig. 2, Appendix S1). A short 438 sulcus on the inner surface of the anterior lamina of the petrosal likely only carried the 439 internal carotid nerve, as the proximal stapedial artery (=transpromontorial portion of the 18 440 artery in Wible, 1984) is absent in modern Sciuroidea. Within the alisphenoid bone, the 441 large internal opening of the foramen ovale can also be observed on the anteroventral side 442 of this middle cranial fossa. A smaller posterior depression, the subarcuate fossa, occupies 443 the posterointernal side of the petrosal and would have housed the petrosal lobule 444 (=paraflocculus), part of the cerebellum (Fig. 2, Appendix S1). Ventral and anteromedial 445 to the subarcuate fossa there is a large internal auditory meatus, subdivided by a tiny 446 transverse crest, slightly more recessed than the meatus. The upper fossa of the internal 447 acoustic meatus, namely the superior acoustic foramen, is subdivided by an even weaker 448 crest perpendicular to the transverse crest. This perpendicular crest delimits an anterior 449 round opening into the facial canal and posteriorly a smaller superior vestibular area, as 450 well as a lower opening for the inferior vestibular area. The lower fossa of the internal 451 acoustic meatus, termed the inferior acoustic foramen, is also subdivided by a tiny ridge. 452 Its anterior half has a large and elliptical opening, the spiral cribiform tract. Posterolateral 453 to this foramen, is the much smaller foramen singulare for the posterior ampullary nerve 454 (branch of the vestibulocochlear nerve VIII) to the ampulla of the posterior semicircular 455 canal. The ridge separating the internal acoustic meatus from the subarquate fossa 456 presents a tiny circular foramen, likely the vestibular aqueduct. A small foramen, 457 presumably corresponding to the mastoid foramen (see Wible & Shelley 2020: 6, 29), is 458 located posterodorsally to the subarcuate fossa. Another small foramen pointing towards 459 the back of the cranium is located posteroventrally to the internal auditory meatus. This 460 foramen likely corresponds to the hypoglossal foramen for nerve XII. Finally, two small 461 foramina can be observed at the back of the skull, close together and near the occipital 462 margin. These foramina likely correspond to two condyloid canals. 463 Comparisons. The cranium is short and wide, resembling that of the large flying squirrels 464 Aeromys and Petaurista (Fig. 6), further being comparable in size to M. neogriviensis and 19 465 Pe. petaurista (Table 2). The rostrum is short and was probably broad, as in 466 Miopetaurista, Petaurista and Aeromys. It exhibits orthodont upper incisors as seen in 467 our sample of extant flying squirrels. In palatal view, the masseteric tubercles are bulbous 468 and prominent, being similar to those of Petaurista. The zygomatic plate is similar in 469 morphology and size to that of other studied flying squirrels. The infraorbital foramen is 470 small and rounded as in Petaurista. The ridge for the insertion of the anterior portion of 471 the buccinator muscle is clearly marked as a semicircular scar similar in shape to that of 472 Aeromys. The palate is not particularly wide, being narrower than in Glaucomyina, such 473 as Glaucomys, Hylopetes and Iomys. The shape and width of the palate is again close to 474 Petaurista. The posterior palatine foramina are located at the M2 level, in the maxillary 475 bone. This resembles the condition seen in Aeromys whereas in other compared extant 476 flying squirrels the palatine foramina are located in the palatine, just in the suture with 477 the maxillary or immediately behind it. The palatines enclose two posterior maxillary 478 foramina, which differ from those of other Pteromyina. These foramina are not 479 completely closed in the studied Pe. petaurista, Eupetaurus cinereus and Pteromys volans 480 as well as in Glaucomyina species. Aeromys tephromelas and Biwamoyopterus laoensis 481 (see Li et al. 2019) are the only studied Pteromyina that show closed posterior maxillary 482 foramina, although they are more elongated in A. tephromelas than in M. crusafonti. 483 However, the intrageneric and intraspecific variability of this character has not been 484 evaluated, and the foramen is known to be closed in one species of the genus 485 Biswamoyopterus (B. laoensis) but not in the other two (B. biswasi, B. gaoligongensis; 486 see Li et al. 2019). Miopetaurista crusafonti presents a short and rounded posterior nasal 487 spine, similar in shape to that of Pe. petaurista and E. cinereus. The pterygoid ridges are 488 parallel in our sample of flying squirrels, but in M. neogrivensis and apparently in M. 489 crusafonti as well they diverge slightly posteriorly. The foramen ovale is large and 20 490 relatively close to the anterior margin of the auditory bulla, similar in size and shape to 491 that of large-sized flying squirrels. 492 The interorbital distance is relatively narrow as compared to our sample of extant 493 Pteromyina (Table 2; Fig. 6). Only Belomys and Pteromys show a similar interorbital 494 constriction, and the interorbital distance is wider in other Pteromyina including M. 495 neogrivensis. In the orbital region the frontal bone is pierced by an oval upper ethmoid 496 foramen at the level of the M3 whereas in other Pteromyina such as Pe. petaurista and A. 497 tephromelas it is more anteriorly placed, at the level of M2. Glaucomyina also have a 498 more anterior upper ethmoid foramen compared to M. crusafonti, at the M2 level. The 499 sphenopalatine foramen is almost entirely included within the maxillary bone, as 500 observed in A. tephromelas. However, in extant large flying squirrels, as well as in some 501 small taxa such as Hylopetes sagitta, the position of this foramen is more anterior (at the 502 contact between M1 and M2) than in M. crusafonti (at the contact between M2 and M3). 503 The squamosal is relatively low and presents a small postglenoid foramen, which within 504 the compared specimens only occur in Petaurista and Eupetaurus. 505 The root of the jugal is deep and robust approaching the morphology of Petaurista. 506 On initial overview, the cranial vault of M. crusafonti seems relatively convex, more 507 closely resembling smaller-sized flying squirrels. However, the dorsal part of the 508 cranium, particularly the frontal, is poorly preserved thus giving this false impression of 509 convexity. Therefore, the morphology of the cranial vault was certainly somewhat 510 originally flatter, although probably not as much as in Petaurista or Aeromys species. The 511 preserved part of the parietal bone shows a prominent temporal ridge that curves medially 512 towards the nuchal crest similarly to M. neogrivensis and E. cinereus. These ridges do not 513 come as close towards the posterior side of the cranium in other flying squirrels, such as 514 in Petaurista, Aeromys or Pteromys. 21 515 Except for its greater interorbital constriction, the cranium of M. crusafonti does 516 not differ significantly from that of M. neogrivensis (Fig. 6B), hence being remarkably 517 similar to the large flying squirrels Petaurista and Aeromys. Yet, as remarked by 518 Casanovas-Vilar et al. (2018), the cheek tooth morphology is clearly different and there 519 are subtle cranial differences. Miopetaurista has the temporal ridges that converge 520 towards the posterior edge of the cranium, resembling the condition seen in Eupetaurus. 521 On the other hand, the morphology of the posterior maxillary foramina is different from 522 that of Petaurista and Aeromys, but the intrageneric and intraspecific variability of this 523 character has not been assessed, so it may not be diagnostic. 524 Endocranial anatomy and comparisons 525 Olfactory bulbs. The olfactory bulbs of M. crusafonti are located above the M1 (Fig. 2, 526 Appendix S1), which is different to the position observed in all fossil and extant squirrels 527 but similar to some ischyromyids (see Bertrand et al. 2018). It is possible that the position 528 of the olfactory bulbs could have resulted from deformation of the rostral region being 529 telescoped anteriorly compared to the braincase; however better-preserved specimens 530 will be necessary test this hypothesis. In ischyromyid rodents past studies have found that 531 the olfactory bulbs are positioned more posteriorly relative to the toothrow, above the 532 M1–M2 (Bertrand et al. 2016b, 2019b). This might be related to the fact the rostrum 533 appears to have shortened and the braincase expanded due to brain size increase in 534 Sciuridae compared to Ischyromyidae (Bertrand et al. 2019a). The olfactory bulbs 535 represent 1.2 % of the total endocranial volume in M. crusafonti (Grau-Camats et al. 536 2021, table S4). Miopetaurista crusafonti has a smaller olfactory bulb volume ratio 537 compared to Eocene and Oligocene Ischyromyidae (3.2 % to 6.1 %), the Oligocene 538 squirrel Cedromus wilsoni (3.0 %), the late Oligocene Sciurini Protosciurus cf. rachelae 539 (3.7 % and 4.9 %), and extant squirrels, which range from 1.6% to 4.7 %. Miopetaurista 22 540 crusafonti also has a lower ratio compared to extant flying squirrels, which range from 541 1.6 % to 3.5 % in olfactory bulb ratio (Table 3; Grau-Camats et al. 2021, table S4). Among 542 the considered extant squirrel sample, Ratufa affinis and Pe. petaurista, the closest living 543 relative of M. crusafonti, show the lowest olfactory bulb volume ratios (1.6 %). Log- 544 transformed olfactory bulb volume shows a positive and highly significant correlation 545 with both log-transformed endocranial volume and body size (Fig. 9A, B; Table 4). Both 546 endocranial volume and body mass explain a high portion of the variance according to 547 PGLS regression (Table 4; r2=0.86 against body mass and r2=0.91 against endocranial 548 volume). The phylogenetic signal is moderate when assessed against body mass (λ~0.6) 549 and high (λ~0.9) if endocranial volume is considered (Table 4). Miopetaurista crusafonti 550 has smaller olfactory bulbs than would be expected for its endocranial volume compared 551 to ischyromyids, other fossil sciurids and the majority of extant squirrels (Fig. 9A). It also 552 has smaller olfactory bulbs than would be expected when assessed against body mass 553 compared to extant sciurids (Fig. 9B). The flying squirrels Pe. petaurista and Pteromys 554 buechneri have high negative residuals and occupy a similar position relative to the 555 regression line for squirrels to M. crusafonti (Grau-Camats et al. 2021, tables S6–S7) 556 although they fall within the cluster of Ischyromyidae when the comparison is made with 557 respect to body mass (Fig. 9B). 558 Cerebrum and midbrain. The circular fissure (Fig. 7A, Appendix S1) of M. crusafonti is 559 shorter antero-posteriorly than in ischyromyid rodents but similar in length compared to 560 that of sciurids, including the extinct C. wilsoni and Pr. cf. rachelae (Fig. 8; Bertrand et 561 al. 2017, 2018, 2019b). The ratio of cerebellum maximum width to cerebrum maximum 562 width (CLW/CRMW) is variable in the studied taxa (Grau-Camats et al. 2021, table S3). 563 Miopetaurista crusafonti has a ratio of 65.8 %, which is very close to the value for the 564 early squirrel C. wilsoni (66.2 %) and lower than the values for the two specimens of Pr. 23 565 cf. rachelae (69.4 % and 79.27 %). The ratio of M. crusafonti is lower than the range 566 observed for extant squirrels (68.8 % to 82.4 %). This result contrasts with the condition 567 in ischyromyids, for which the ratio is higher, between 82.5 % and 103.6 %, suggesting 568 that the cerebellum and cerebrum have more similar widths in this family of early rodents 569 (Bertrand et al. 2019b). Previous research concluded that the lower ratio exhibited by 570 squirrels is explained by the lateral expansion of the cerebrum relative to the cerebellum 571 compared to the configuration in ischyromyid rodents (Bertrand et al. 2018, 2019b). The 572 lowest ratios in the extant dataset occur in flying squirrels, with two species displaying 573 the lowest values for extant taxa: Hylopetes spadiceus (68.8 %) and Pt. buechneri 574 (69.3%). The midbrain is not visible on the endocranial surface of M. crusafonti, which 575 implies complete coverage of this area by the cerebrum (Figs 7A, C). The observed 576 condition resembles all extant squirrels (e.g., Fig. 8A–B) but contrasts with the situation 577 in other extinct squirrels which exhibit only a partially covered midbrain (Fig. 8C–D; 578 Bertrand et al. 2017, 2019b). 579 A temporal fossa is visible in the fossil taxa C. wilsoni (Fig. 8C), Pr. cf. rachelae 580 (Fig. 8D), M. crusafonti (Fig. 7C) as well as in all studied extant squirrels. In contrast, the 581 fossa is absent in most ischyromyid rodents (e.g, Fig. 8E) with the exception of 582 Pseudotomus horribilis, Pseudotomus hians, one specimen of Ischyromys typus and in 583 Reithroparamys sciuroides (Bertrand et al. 2019b). The Sylvian fossa is absent in the 584 fossil taxa M. crusafonti, Pr. cf. rachelae, and also in the ischyromyids (Bertrand et al. 585 2018, 2019b). The presence of this structure is variable among other extant and extinct 586 squirrels. For instance, a Sylvian fossa is visible in C. wilsoni (Fig. 8C), but is absent in 587 our sample of pteromyins except for A. tephromelas, Petinomys setosus and Pt. 588 buechneri, while an actual Sylvian sulcus can be observed in the invertivorous 589 callosciurine Rhinosciurus laticaudatus (Bertrand et al. 2017). Lateral sulci are absent in 24 590 M. crusafonti, C. wilsoni, Pr. cf. rachelae and in the majority of our extant squirrel 591 sample, while the presence of lateral sulci is variable in the ischyromyids (see Bertrand 592 et al. 2019a). However, they are visible in large flying squirrels, such as the pteromyins 593 A. tephromelas and Pe. petaurista (Fig. 8A; Bertrand et al. 2017), which also have the 594 highest endocranial volumes within the Sciuridae (11.5 cm3 in A. tephromelas and 12.3 595 cm3 in Pe. Petaurista; Table 3; Grau-Camats et al. 2021, table S4). Calculated endocranial 596 volume for M. crusafonti is 10.82 cm3, thus being lower than both extant taxa. In general, 597 presence or absence of neocortical sulci is related to the endocranial volume, with brains 598 of less than 5 cm3 generally being lissencephalic (Macrini et al. 2007). For rodents, both 599 gyrencephalic and lissencephalic brains can be found in an interval of brain masses 600 between 3 and 30 cm3 (Pilleri et al. 1984). The absence of sulci in all but the largest 601 squirrels suggests that the value below which gyrification occurs may be higher than in 602 some other mammalian groups. As such, M. crusafonti may differ from the large extant 603 flying squirrels in lacking the lateral sulcus because of its somewhat smaller brain size. 604 The position of the orbitotemporal canal of M. crusafonti (Fig. 7C) is similar to 605 that of the late Oligocene sciurid Pr. cf. rachelae (Fig. 8D) in being near the ventral extent 606 of the temporal lobe. This structure is positioned more ventrally in M. crusafonti 607 compared to ischyromyid rodents and the early Oligocene sciurid C. wilsoni (Figs 8C, E; 608 Bertrand et al. 2017, 2019b). A ventrally positioned canal reflects a ventral expansion of 609 the neocortex, based on the inferred relationship between the canal and the rhinal fissure 610 in sciurids (Bertrand et al. 2017). In extant squirrels the orbitotemporal canal is even more 611 ventrally positioned than in Pr. cf. rachelae (e.g., Fig. 8A–B), thus resulting in a greater 612 neocortical surface area. This canal is straight in M. crusafonti and morphologically 613 similar to that of the flying squirrels A. tephromelas and Pe. petaurista (Bertrand et al. 614 2017). The neocortex of M. crusafonti represents 30.2 % of the total endocast surface 25 615 area, which is in the range of the values obtained for other extinct squirrels (Fig. 9C, Table 616 3; Grau-Camats et al. 2021, table S4). In extant squirrels the ratio is higher, ranging from 617 33.8 % to 39.3 % and for flying squirrels specifically, the range of values is between 33.8 618 % and 36.7 % (Fig. 9C, Table 3; Bertrand et al. 2017; Grau-Camats et al. 2021, table S4). 619 This result contrasts with ischyromyids, which range from 16.3 % to 23 % (Bertrand et 620 al. 2016b, 2018, 2019b). Log-transformed neocortical surface area shows a positive and 621 highly significant correlation with log-transformed endocranial surface area, the latter 622 variable explaining a significant part of existing variance (r2>0.90; Table 4). Our analyses 623 show that phylogenetic signal is strong (λ>1; see Table 4), which would suggest that more 624 closely related species display more similar neocortical sizes in our sample (Fig. 9D). 625 However, this result appears to be mainly driven by ischyromyid rodents as they have a 626 much lower neocortical surface area percentage compared to extinct and extant squirrels. 627 Indeed, after rerunning the PGLS analysis without Ischyromyidae, λ dropped close to 0 628 (and correlation increased to r2>0.99; Table 4), suggesting that the impact of phylogeny 629 on the relative size of the neocortex is not strong among extinct and extant squirrels. 630 Nevertheless, the neocortical surface area in M. crusafonti is slightly smaller than would 631 be expected for its endocranial surface area (Fig. 9D). Miopetaurista crusafonti has 632 residual values from the regression line that are close to those calculated for other fossil 633 squirrels, Pr. cf. rachelae and C. wilsoni (Fig. 9D, Bertrand et al. 2017, 2018; Grau- 634 Camats et al. 2021, tables S6–S7). 635 The hypophyseal fossa is poorly defined, although there is a slight bulge in the 636 endocast just posterior to the optic chiasm that might mark its position. A poorly 637 demarcated fossa is also characteristic of C. wilsoni (Fig. 8C) and most extant sciurids 638 studied here, although it is more salient in some ischyromyids (e.g., Fig. 8E; Bertrand et 639 al. 2016b, 2017, 2018, 2019b; Bertrand & Silcox 2016). 26 640 Cerebellum. The caudal region of the endocast of M. crusafonti is damaged. The vermis 641 of the cerebellum is visible, separated from the left lateral lobe of the cerebellum by the 642 paramedian fissure (Fig. 7A, C; Appendix S1). Only the left petrosal lobe 643 (=paraflocculus, see Bertrand et al. 2020) is preserved, therefore, the volume calculated 644 for the left petrosal lobe was compared to the volume for the left side of the endocast 645 (Table 3; Grau-Camats et al. 2021, table S4). Miopetaurista crusafonti has relatively 646 smaller petrosal lobules (1.9 % of total endocast volume) compared to the other fossil 647 squirrels, C. wilsoni (3.2 %) and Pr. cf. rachelae (3.3 %), thus being in the range of extant 648 squirrels (0.9 %–2.3 %) and in the upper part of the range of ischyromyid rodents (0.4 649 %–2.1 %; Bertrand et al. 2017, 2018, 2019b). Compared to extant pteromyins, M. 650 crusafonti is in the upper range of variation for the petrosal lobule volume ratio (0.9 %– 651 1.7 %). Log-transformed petrosal lobule volume shows a positive and highly significant 652 correlation with both log-transformed endocranial volume and body mass. Both 653 endocranial volume and body mass explain a low proportion of the existing variance (r2 654 <0.60; Table 4). Furthermore, the phylogenetic signal is low (λ~0.3; see Table 4) 655 suggesting that closely related species are less likely to exhibit similar relative and 656 absolute petrosal lobule sizes and that other factors may affect the size of this brain 657 structure. The size of the petrosal lobules of M. crusafonti is equivalent to the value 658 expected for its endocranial volume compared to modern sciurids (Fig. 9E). Four extant 659 flying squirrels are below the regression line while the remaining three (i.e., Pe. 660 petaurista, A. tephromelas and M. crusafonti) are above it. This relationship is nearly 661 identical when looking at the relationship between petrosal lobule size and body mass; 662 however, in this instance, M. crusafonti lies slightly closer to Pe. petaurista and well 663 above the regression line (Fig. 9F). It is worth noting that one extant flying squirrel 664 (Petinomys setosus) in our dataset lacks the subarcuate fossa entirely (Table 3; Grau- 27 665 Camats et al. 2021, table S4) and therefore has no petrosal lobules on the endocast 666 (Bertrand et al. 2017). 667 Cranial nerves and blood vessels. The ventral region of the endocast is damaged, 668 particularly on its posterior half, but several cranial nerves can be observed (Fig. 7B, 669 Appendix S1). Casts of both optic nerves and the optic chiasm are partly preserved in M. 670 crusafonti, with the chiasm being located just anterior to the point at which the 671 oculomotor nerve would have left the brain. Its location is difficult to identify as this area 672 of the cranium is not well-preserved; however, the chiasm appears to be located dorsal to 673 the posterior nasal spine (Fig. 7B). The sphenorbital fissure is not preserved. Casts of the 674 buccinator and masseteric nerves (branches of nerve V3) are partly preserved on both 675 sides and exit through the masticatory foramen, which demonstrates that they would have 676 had a course separate from the other parts of V3 exiting through the foramen ovale (Fig. 677 7). The casts of mandibular (V3), facial (VII) and vestibulocochlear (VIII) nerves are 678 observable on the left side of the endocast only. The position of the different casts of the 679 nerves resembles the closely-related Pe. petaurista (Fig. 8A). The superior sagittal sinus 680 is demarcated (Fig. 7A), being similar in morphology to C. wilsoni, Pr. cf. rachelae (Fig. 681 8C–D) and most modern sciurids in our comparative sample (see Bertrand et al. 2017, 682 2018). It differs from the condition observed in ischyromyid rodents, in which the 683 superior sagittal sinus is less well marked (e.g., Bertrand et al. 2019b). The transverse and 684 sigmoid sinuses are visible on the left side of the endocast of M. crusafonti (Fig. 7C), but 685 the region where the inferior petrosal sinus would have passed is not preserved. The 686 stapedial canal for the distal section of the stapedial artery (Wible & Shelley 2020) can 687 be observed on the ventral region (Fig. 7B–C) and is similarly positioned compared to 688 sciurids and ischyromyids (Bertrand et al. 2017, 2018, 2019b). 689 Brain size and encephalization quotient 28 690 The endocranial volume of M. crusafonti (estimated doubling the volume of the 691 most complete side, see Materials and Methods) is 10.82 cm3. The encephalization 692 quotient (EQ) calculated using the equation by Pilleri et al. (1984) is 1.13 (Table 3; Grau- 693 Camats et al. 2021, table S5). This value is higher than those calculated for ischyromyids 694 (0.50 to 0.88) and very similar to those for C. wilsoni (1.0) and Pr. cf. rachelae (1.06; 695 Fig. 10A; Bertrand et al. 2017, 2018, 2019b). Miopetaurista crusafonti has an EQ below 696 those of extant Sciurinae, being in the lower range of variation for Callosciurinae (Fig. 697 10A). Log-transformed endocranial volume shows a positive and highly significant 698 correlation with log-transformed body mass, the latter variable explaining great part of 699 the existing variance (r2=0.89; see Table 4). Phylogenetic signal is relatively low (λ~0.4; 700 see Table 4) implying that closely related species may not necessarily have a similar 701 brain-body mass relationship, suggesting that other factors may impact relative brain size. 702 When assessed against body mass, the endocranial volume of M. crusafonti is above the 703 regression line for extant squirrels, below its close relative Pe. petaurista (Fig. 10B). 704 DISCUSSION 705 Several studies have shown that the encephalization quotient (EQ) increases with 706 time in different mammalian orders (Jerison 1973; Radinsky 1976; Gurche 1982; Silcox 707 et al. 2010; Orliac & Gilissen 2012; Yao et al. 2012; Bertrand et al. 2017, 2018, 2019b, 708 2021). On the other hand, a series of papers have shown that EQ also varies as a function 709 of ecology in chiropterans and rodents (Eisenberg & Wilson 1978; Harvey et al. 1980; 710 Mace et al. 1981; Roth & Thorington 1982; Meier 1983; Pilleri et al. 1984; Bertrand et 711 al. 2017, 2018, 2019b). These studies found that arboreality was associated with a higher 712 EQ in rodents, likely because of the requirements of foraging in a complex three- 713 dimensional environment such as found in the treetops. Recently, Bertrand et al. (2016b, 714 2017, 2018, 2019b, 2021) studied the endocasts of ischyromyids and early sciurids 29 715 (Cedromus wilsoni and Protosciurus cf. rachelae) as compared to extant aplodontids and 716 sciurids. The Oligocene C. wilsoni (subfamily Cedromurinae) and Pr. cf. rachelae 717 (subfamily Sciurinae, tribe Sciurini), both considered to be scansorial or arboreal squirrels 718 (Korth & Samuels 2015; Bhagat et al. 2021), had not yet reached the EQ of their extant 719 relatives. Indeed, Cedromus is in the upper part of the range of extant ground squirrels, 720 while Protosciurus is in the lower part of the range of tree squirrels. Similarly, the EQ of 721 M. crusafonti is 1.07, which is slightly below the range of variation of extant flying 722 squirrels (1.11-1.39; Fig. 10A, Table 3; Grau-Camats et al. 2021, table S5), yet there is 723 unquestionable evidence of gliding locomotion in the genus Miopetaurista (Casanovas- 724 Vilar et al. 2018). Accordingly, the lower EQ suggests that there were temporal trends in 725 this parameter within the sciurid family, as observed in other mammalian families and 726 orders (e.g., Jerison 1973; Radinsky 1976; Silcox et al. 2010; Bertrand et al. 2019a) . In 727 addition, the increase in relative brain size would have occurred independently in the 728 Sciurini and Pteromyini, because early members of both of these Sciurinae tribes 729 (Protosciurus, Miopetaurista) show lower EQs than extant representatives. Considering 730 that two events of EQ increase occurred in closely related clades, and that the 731 phylogenetic signal is low for the brain-body mass regression, it can be inferred that 732 changes in relative brain size may relate to factors other than phylogeny, such as 733 ecological transitions (Bertrand et al. 2017, 2018, 2021). 734 The comparison of the endocast of M. crusafonti to those of extant and extinct 735 sciurids, as well as to the more primitive ischyromyid rodents suggests some significant 736 changes in brain morphology in the flying squirrel lineage. Previous studies have shown 737 that an increase in neocortical surface area ratio occurs in the transition from 738 ischyromyids to sciurids (Bertrand et al. 2017, 2018, 2019a). The late Oligocene fossil 739 Sciurini Pr. cf. rachelae and the middle Miocene fossil Pteromyini M. crusafonti both 30 740 exhibit lower neocortical surface area ratios compared to their respective extant relatives 741 (Fig. 9C; Grau-Camats et al. 2021, table S4). Therefore, neocortical surface ratio would 742 have increased independently in tree and flying squirrels, which is the likely cause for the 743 apparent independent increases in EQ discussed above. However, the independent 744 increase in neocortical surface area cannot be solely explained by a possible shift in 745 ecology because all extinct and extant squirrels have larger neocortices than ischyromyid 746 rodents. The gradual increase in neocortical surface area appears to more closely follow 747 a temporal trend than relative brain size and suggests that phylogeny may have more 748 impact on the neocortical size than on overall brain size in our sample. We note that if 749 Ischyromyidae are not included in the estimation of λ for the neocortex, the phylogenetic 750 signal for the relative size of this brain region is markedly lower (Table 4) as all extant 751 squirrels have a higher neocortical surface area compared to extinct squirrels. 752 Previous studies have shown that an increase in the absolute size of the petrosal 753 lobules occurred in the evolution of the sciurids from ischyromyids as a response to a 754 more arboreal lifestyle in the former group (Bertrand et al. 2017, 2018, 2019b). Our 755 results show a low phylogenetic signal, thus agreeing with these earlier studies, in that 756 factors other than shared evolutionary story have an important effect on the relative and 757 absolute size of the petrosal lobules. Extant pteromyins of our sample have relatively 758 smaller petrosal lobules compared to sciurins (Fig. 9E; Bertrand et al. 2017; Grau-Camats 759 et al. 2021, table S4). This difference is already apparent in Miocene taxa. Indeed, while 760 the tree squirrel Pr. cf. rachelae has relatively large petrosal lobules, the flying squirrel 761 M. crusafonti shows a lower relative size, thus resembling its extant relatives (Fig. 9E; 762 Grau-Camats et al. 2021, table S4). This suggests a decrease in the relative size of the 763 petrosal lobules in the evolution of flying squirrels which might be related to a packing 764 problem due to the constraints inherent to a skull built to be aerodynamic for gliding (see 31 765 also Bertrand et al. 2021). The functional elements of the petrosal lobules might still be 766 present but in the non-petrosal lobule part of the cerebellum. However, an alternative 767 explanation could be related to a specific adaptation. Tree squirrels spend a considerable 768 amount of time on thin and unstable branches, which implies that balance and control of 769 the head and eye movements are crucial for these animals. In contrast, flying squirrels 770 have a different lifestyle in the trees and to propel themselves in the air, they use more 771 stable and sturdy branches and trunks (Bishop 2006). This suggests that flying squirrels 772 might require less overall eye movement control compared to tree squirrels while 773 navigating the canopy. Pteromyini also have larger auditory bullae compared to tree 774 squirrels (Lu et al. 2014). These structures are in contact with the braincase and more 775 specifically with the subarcuate fossa. Using allometric shape deformations, Lu et al. 776 (2014) showed that the shape of different cranial regions (e.g., auditory bulla and vault- 777 occiput roundness) changed simultaneously from small to large squirrel specimens. 778 Therefore, enlarged bullae may influence and possibly reduce the space for the petrosal 779 lobules. Alternatively, the auditory bulla may represent a separate module, meaning that 780 the development of both units could be independent. Ultimately, more work will be 781 required to test these hypotheses. The same authors suggested that because the bullae are 782 so large in flying squirrels, hearing might be crucial while gliding. The parietal region of 783 the brain of flying squirrels appears to be more developed, which could potentially 784 suggest enhanced hearing (Bertrand et al. 2019a). Other studies have indicated that flying 785 squirrels, which are nocturnal gliders, might be using ultrasounds like echolocating bats 786 to ease navigation while gliding (Muul & Alley 1963; but see Chattin 1969). More 787 recently, Murrant et al. (2013) identified that flying squirrels were reacting to their own 788 high-frequency vocalizations and therefore were capable to hear ultrasonic sounds. The 789 use of echolocation in flying squirrels is debated and has just begun to be studied in detail, 32 790 but there is no direct evidence of echo-based navigation so far (Newar & Bowman 2020). 791 Flying squirrels have large eyes and undoubtedly use visual cues during glides, but they 792 may be also assisted by the use of ultrasound (Murrant et al. 2013). Therefore, the 793 observed decrease in the size of the petrosal lobules could result from an accommodation 794 for other biological needs such as hearing. More work is required to better understand the 795 relationship between the neurosensory system of flying squirrels and how they navigate 796 through the canopy. 797 Finally, previous studies have shown that a decrease in the size of the olfactory 798 bulbs occurred through time from ischyromyids to sciurids (Bertrand et al. 2017, 2018, 799 2019b). The relative size of the olfactory bulbs in M. crusafonti is slightly below the range 800 of to those of other members of the subtribe Pteromyina (i.e., large flying squirrel clade), 801 including the closely related Pe. petaurista, while that of the sciurin Pr. cf. rachelae is 802 conspicuously in the upper range of the extant tree squirrels variation (Grau-Camats et al. 803 2021, table S4). Interestingly, Glaucomys volans, Hylopetes spadiceus and Petinomys 804 setosus, the studied members of the subtribe Glaucomyina (i.e., small flying squirrel 805 clade), all show relatively larger olfactory bulbs, in the range of other extant squirrel 806 groups (see Grau-Camats et al. 2021, table S4). This could suggest that a decrease in the 807 relative size of the olfactory bulbs occurred after the Pteromyina/Glaucomyina split. 808 Despite exhibiting a relatively small neocortex, closer to the condition in other Oligocene 809 squirrels rather than to extant species, M. crusafonti already exhibits features 810 characteristic of Pteromyini (and more specifically the Pteromyina) such as low petrosal 811 lobule and olfactory bulb volume ratios. 812 CONCLUSIONS 813 Squirrels are generally regarded as an anatomically conservative group and flying 814 squirrels are no exception, the oldest skeleton of a flying squirrel, belonging to 33 815 Miopetaurista neogrivensis and dating back to 11.6 Ma, being very similar in many 816 aspects of morphology to the extant giant flying squirrel Petaurista (Casanovas-Vilar et 817 al. 2018). The crania of both M. neogrivensis and the closely-related Miopetaurista 818 crusafonti described here are also remarkably similar to those of Petaurista and other 819 large flying squirrels (subtribe Pteromyina) such as Aeromys and Biswamoyopterus, 820 differing only in cheek tooth morphology and a few minor details in the cranium, such as 821 the morphology of the temporal ridges. Endocranial morphology, as well as the relative 822 volume of the different brain regions are very similar to other Pteromyina, showing for 823 example the relatively reduced petrosal lobules characteristic of flying squirrels in 824 comparison to other extant squirrels. However, the encephalization quotient (EQ) of M. 825 crusafonti is slightly below that of extant flying squirrels. A temporal trend in EQ increase 826 has been previously reported for sciurids and other mammal families (Jerison 1973; 827 Radinsky 1976; Gurche 1982; Silcox et al. 2010; Orliac & Gilissen 2012; Yao et al. 2012; 828 Bertrand et al. 2017, 2018, 2019b), but this parameter has also been shown to vary with 829 locomotor mode, arboreal and gliding rodent species showing higher values than 830 terrestrial ones (Bertrand et al. 2021). Since Miopetaurista was certainly a glider, like its 831 extant relatives, the conspicuously lower EQ must be attributed to its older age rather than 832 to different locomotion. Even though this should be tested in additional squirrel groups, 833 these results support an independent temporal increase in EQ in different squirrel clades, 834 further showing that, in terms of relative brain size, flying squirrels are not so 835 conservative. 836 Author contributions. MGC, OCB, MTS and ICV contributed to the conception and 837 design of the study. JP acquired the CT data, MGC, OCB, SLT segmented the specimens 838 and acquired quantitative measurements. MGC and OCB carried out the statistical 34 839 analyses. MGC and OCB drafted the article with important contributions by all other 840 authors. All authors gave final approval before submission. 841 Acknowledgements. This publication is part of project I+D+i PID2020-117289GBI00 842 funded by MCIN/AEI/10.13039/501100011033/. It has also been supported by the 843 Spanish Agencia Estatal de Investigación and European Regional Development Fund of 844 the European Union (CGL2016-76431-P and CGL2017-82654-P); the Generalitat de 845 Catalunya (CERCA Programme); a Marie Skłodowska-Curie Actions: Individual 846 Fellowship (H2020-MSCA-IF-2018-2020; No. 792611) to O.C.B; an NSERC Discovery 847 Grant to M.T.S.; and the Ministerio de Economía y Competitividad (RYC-2013-12470 to 848 I.C.-V.). I.C.-V. is member of the consolidated research group 2017 SGR 116 of the 849 Generalitat de Catalunya. The authors thank Sergio Llácer (ICP) for his help with 850 specimen segmentation, and Sergio Llácer and Javier Serrano (ICP) for preparing the 851 video for Appendix S1. We also thank L. W. Van den Hoek Ostende, S. Van der Mije 852 and P. Kaminga (Naturalis Biodiversity Center, Leiden, the Netherlands) for their 853 assistance during the study of the squirrel collections kept at their institution. We thank 854 S. Thomas for her technical comments and assistance in the preparation of tables and 855 figures. The constructive comments by the two reviewers of this manuscript (J. Maugoust 856 and M.V. Sinitsa) as well as by handling editor (L. Hautier) certainly helped to improve 857 the final result. 858 DATA ARCHIVING STATEMENT 859 Data for this study are available in the Dryad Digital Repository: 860 https://doi.org/10.5061/dryad.5qfttdz4pç; 3D surface models of the Miopetaurista 861 crusafonti cranium and endocast SNSB 1978 V 1 from Gumpersdorf are available at 862 MorphoSource (www.morphosource.org) with identifier xxxxx. 35 863 SUPPORTING INFORMATION 864 Additional supporting information can be found in the online version of this article. 865 Appendix S1. Animated rendering of the virtual model of the cranium of Miopetaurista 866 crusafonti (SNSB 1978 V 1) from Gumpersdorf (Bavaria) based on μCT 867 data. Main anatomical elements of the cranium and the endocast are 868 indicated. 869 REFERENCES 870 ABDUL AZIZ, H., BÖHME, M., ROCHOLL, A., PRIETO, J., WIJBRANS, J. R., 40 871 BACHTADSE, V. and ULBIG, A. 2010. Integrated stratigraphy and Ar/39Ar 872 chronology of the early to middle Miocene Upper Freshwater Molasse in western 873 Bavaria (Germany). International Journal of Earth Sciences, 99, 1859–1886. 874 ASHER, R. J., SMITH, M. R., RANKIN, A. and EMRY, R. J. 2019. Congruence, fossils 875 and the evolutionary tree of rodents and lagomorphs. Royal Society Open Science, 876 6, 190387. 877 BALL, S. S. and ROTH, V. L. 1995. Jaw muscles of New World squirrels. Journal of 878 Morphology, 224, 265–291. 879 BAPST, D. W. 2012. paleotree: an R package for paleontological and phylogenetic 880 analyses of evolution. Methods in Ecology and Evolution, 3, 803–807. 881 ———. 2013. A stochastic rate-calibrated method for time-scaling phylogenies of fossil 882 taxa. Methods in Ecology and Evolution, 4, 724–733. 883 ———. 2014. Preparing paleontological datasets for phylogenetic comparative methods. 884 In GARAMSZEGI, L. Z. (ed.) Modern Phylogenetic Comparative Methods and 885 Their Application in Evolutionary Biology: Concepts and Practice, Springer, 886 Berlin, Heidelberg, 515–544 pp. 36 887 BERTRAND, O. C. and SILCOX, M. T. 2016. First virtual endocasts of a fossil rodent: 888 Ischyromys typus (Ischyromyidae, Oligocene) and brain evolution in rodents. 889 Journal of Vertebrate Paleontology, 36, e1095762. 890 ———, SCHILLACI, M. A. and SILCOX, M. T. 2016a. Cranial dimensions as 891 estimators of body mass and locomotor habits in extant and fossil rodents. Journal 892 of Vertebrate Paleontology, 36, e1014905. 893 ———, AMADOR-MUGHAL, F. and SILCOX, M. T. 2016b. Virtual endocasts of 894 Eocene Paramys (Paramyinae): oldest endocranial record for Rodentia and early 895 brain evolution in Euarchontoglires. Proceedings of the Royal Society B: 896 Biological Sciences, 283, 20152316. 897 ———, ——— and ———. 2017. Virtual endocast of the early Oligocene Cedromus 898 wilsoni (Cedromurinae) and brain evolution in squirrels. Journal of Anatomy, 230, 899 128–151. 900 ———, SAN MARTIN-FLORES, G. and SILCOX, M. T. 2019a. Endocranial shape 901 variation in the squirrel-related clade and their fossil relatives using 3D geometric 902 morphometrics: contributions of locomotion and phylogeny to brain shape. 903 Journal of Zoology, 308, 197–211. 904 ———, AMADOR‐MUGHAL, F., LANG, M. M. and SILCOX, M. T. 2018. Virtual 905 endocasts of fossil Sciuroidea: brain size reduction in the evolution of fossoriality. 906 Palaeontology, 61, 919–948. 907 ———, ———, ——— and ———. 2019b. New virtual endocasts of Eocene 908 Ischyromyidae and their relevance in evaluating neurological changes occurring 909 through time in Rodentia. Journal of Mammalian Evolution, 26, 345–371. 37 910 ———, PÜSCHEL, H. P., SCHWAB, J. A., SILCOX, M. T. and BRUSATTE, S. L. 911 2021. The impact of locomotion on the brain evolution of squirrels and close 912 relatives. Communications Biology, 4, 460. 913 ———, SHELLEY, S. L., WIBLE, J. R., WILLIAMSON, T. E., HOLBROOK, L. T., 914 CHESTER, S. G. B., BUTLER, I. B. and BRUSATTE, S. L. 2020. Virtual 915 endocranial and inner ear endocasts of the Paleocene ‘condylarth’ Chriacus: new 916 insight into the neurosensory system and evolution of early placental mammals. 917 Journal of Anatomy, 236, 21–49. 918 BHAGAT, R., BERTRAND, O. C. and SILCOX, M. T. 2021. Evolution of arboreality 919 and fossoriality in squirrels and aplodontid rodents: Insights from the semicircular 920 canals of fossil rodents. Journal of Anatomy, 238, 96–112. 921 BININDA-EMONDS, O. R. P., CARDILLO, M., JONES, K. E., MACPHEE, R. D. E., 922 BECK, R. M. D., GRENYER, R., PRICE, S. A., VOS, R. A., GITTLEMAN, J. 923 L. and PURVIS, A. 2007. The delayed rise of present-day mammals. Nature, 446, 924 507–512. 925 BISHOP, K. L. 2006. The relationship between 3-D kinematics and gliding performance 926 in the southern flying squirrel, Glaucomys volans. Journal of Experimental 927 Biology, 209, 689–701. 928 BRAUER, K. and SCHOBER, W. 1970. Katalog der Säugetiergehirne. Catalogue of 929 mammalian brains. VEB Gustav Fischer, Jena, Germany. 930 DE BRUIJN, H. 1999. Superfamily Sciuroidea. In RÖSSNER, G. E. and HEISSIG, K. 931 (eds.) The Miocene Land Mammals of Europe, Verlag Dr. Friedrich Pfiel, Munich, 932 271–280 pp. 933 CASANOVAS-VILAR, I., ALMÉCIJA, S. and ALBA, D. M. 2015. Late Miocene flying 934 squirrels from Can Llobateres 1 (Vallès-Penedès Basin, Catalonia): systematics 38 935 and palaeobiogeography. Palaeobiodiversity and Palaeoenvironments, 95, 353– 936 372. 937 ———, GARCIA-PORTA, J., FORTUNY, J., SANISIDRO, Ó., PRIETO, J., 938 QUEREJETA, M., LLÁCER, S., ROBLES, J. M., BERNARDINI, F. and ALBA, 939 D. M. 2018. Oldest skeleton of a fossil flying squirrel casts new light on the 940 phylogeny of the group. ELife, 7, e39270. 941 CHATTIN, D. 1969. Acoustical echolocation in flying squirrels.BA Thesis, Reed 942 College, Portland, 76pp. 943 DAAMS, R. 1977. Aragonian sciuroptera (Sciuridae, Rodentia, Mammalia) from Spain. 944 Proceedings of the Koninklijke Nederlandse Akademie van Wetenschappen, 945 80(5), 356–359. 946 DAXNER-HÖCK, G. and MEIN, P. 1975. Taxonomische Probleme um das Genus 947 Miopetaurista Kretzoi, 1962 (Fam. Sciuridae). Paläontologische Zeitschrift, 49, 948 75–77. 949 DAXNER-HÖCK, G. and HÖCK, E. 2015. Catalogus Fossilium Austriae. Ein 950 systematisches Verzeichnis aller auf österreichsichem Gebiet festgestellten 951 Fossilien. Band 4. Rodentia neogenica. Catalogus Fossilium Austriae. Ein 952 Systematisches Verzeichnis Aller Auf Österreichsichem Gebiet Festgestellten 953 Fossilien. Verlag der Österreichischen Akademie der Wissenschaften. 954 EISENBERG, J. F. and WILSON, D. E. 1978. Relative brain size and feeding strategies 955 in the Chiroptera. Evolution, 32, 740–751. 956 EMRY, R. J. and KORTH, W. W. 2007. A new genus of squirrel (Rodentia, Sciuridae) 957 from the mid-Cenozoic of North America. Journal of Vertebrate Paleontology, 958 27, 693–698. 39 959 FABRE, P.-H., HAUTIER, L., DIMITROV, D. and DOUZERY, E. J. P. 2012. A glimpse 960 on the pattern of rodent diversification: a phylogenetic approach. BMC 961 Evolutionary Biology, 12, 88. 962 FAHLBUSCH, V. 1979. Flughörnchen-Schädel (Miopetaurista crusafonti; Inv.-Nr. 1978 963 V1). 7, 18–19. 964 FRANZEN, J. L. and STORCH, G. 1975. Die unterpliozäne (turolische) Wirbeltierfauna 965 von Dorn-Dürkheim, Rheinhessen (SW-Deutschland). 1. Entdeckung, Geologie, 966 Mammalia: Carnivora, Proboscidea, Rodentia. Grabungsergebnisse 1972-1973. 967 Senckenbergiana Lethaea, 56, 233–303. 968 FRANZEN, J. L., PICKFORD, M. and COSTEUR, L. 2013. Palaeobiodiversity, 969 palaeoecology, palaeobiogeography and biochronology of Dorn-Dürkheim 1—a 970 summary. Palaeobiodiversity and Palaeoenvironments, 93, 277–284. 971 GRAFEN, A. and HAMILTON, W. D. 1989. The phylogenetic regression. Philosophical 972 Transactions of the Royal Society of London. B, Biological Sciences, 326, 119– 973 157. 974 GRAU-CAMATS, M., BERTRAND, O. C., PRIETO, J., LÓPEZ-TORRES, S., 975 SILCOX, M. T. and CASANOVAS-VILAR, I. 2021. Data from: A Miopetaurista 976 (Sciuridae, Rodentia) cranium from the middle Miocene of Bavaria (Germany) 977 and brain evolution in flying squirrels. Dryad Digital Repository. 978 GREGOR, H.-J. 1982. Die jungtertiären Floren Süddeutschlands. Paläokarpologie, 979 Phytostratigraphie, Paläoökologie, Paläoklimatologie. Ferdinand Enke, 980 Stuttgart. 981 GURCHE, J. A. 1982. Early primate brain evolution. In ARMSTRONG, E. and FALK, 982 D. (eds.) Primate Brain Evolution: Methods and Concepts, Springer US, Boston, 983 MA, 227–246 pp. 40 984 HAN, G., MAO, F., BI, S., WANG, Y. and MENG, J. 2017. A Jurassic gliding 985 euharamiyidan mammal with an ear of five auditory bones. Nature, nature24483. 986 HARVEY, P. H., CLUTTON-BROCK, T. H. and MACE, G. M. 1980. Brain size and 987 ecology in small mammals and primates. Proceedings of the National Academy 988 of Sciences, 77, 4387–4389. 989 HERRON, M. D., CASTOE, T. A. and PARKINSON, C. L. 2004. Sciurid phylogeny and 990 the paraphyly of Holarctic ground squirrels (Spermophilus). Molecular 991 Phylogenetics and Evolution, 31, 1015–1030. 992 HILGEN, F. J., LOURENS, L. J. and VAN DAM, J. A. 2012. The Neogene Period. In 993 GRADSTEIN, F. M., OGG, J. G., SCHMITZ, M. and OGG, G. (eds.) The 994 Geologic Time Scale 2012, Elsevier, Amsterdam, 923–978 pp. 995 HUELSENBECK, J. P. and RONQUIST, F. 2001. MRBAYES: Bayesian inference of 996 phylogenetic trees. Bioinformatics, 17, 754–755. 997 IVES, A. R. 2019. R2s for correlated data: phylogenetic models, LMMs, and GLMMs. 998 Systematic Biology, 68, 234–251. 999 JACKSON, S. 2012. Gliding Mammals of the World. CSIRO Publishing, Collingwood, 1000 Vic. 1001 JACKSON, S. M. and THORINGTON, R. W. Jr. 2012. Gliding Mammals: Taxonomy of 1002 Living and Extinct Species. Smithsonian Contributions to Zoology. Smithsonian 1003 Institution Scholarly Press, Washington DC. 1004 JERISON, H. J. 1973. Evolution of the Brain and Intelligence. Academic Press, New 1005 York, NY. 1006 ———. 2012. Digitized fossil brains: neocorticalization. Biolinguistics, 6, 383–392. 1007 KIRSCHER, U., PRIETO, J., BACHTADSE, V., AZIZ, H. A., DOPPLER, G., 1008 HAGMAIER, M. and BÖHME, M. 2016. A biochronologic tie-point for the base 41 1009 of the Tortonian stage in European terrestrial settings: Magnetostratigraphy of the 1010 topmost Upper Freshwater Molasse sediments of the North Alpine Foreland Basin 1011 in Bavaria (Germany). Newsletters on Stratigraphy, 49, 445–467. 1012 KORTH, W. W. 1994. The Tertiary Record of Rodents in North America. Plenum Press. 1013 ——— and EMRY, R. J. 1991. The skull of Cedromus and a review of the Cedromurinae 1014 (Rodentia, Sciuridae). Journal of Paleontology, 65, 984–994. 1015 ——— and SAMUELS, J. X. 2015. New rodent material from the John Day Formation 1016 (Arikareean, Middle Oligocene to Early Miocene) of Oregon. Annals of Carnegie 1017 Museum, 83, 19–84. 1018 LI, Q., LI, X.-Y., JACKSON, S. M., LI, F., JIANG, M., ZHAO, W., SONG, W.-Y. and 1019 JIANG, X.-L. 2019. Discovery and description of a mysterious Asian flying 1020 squirrel (Rodentia, Sciuridae, Biswamoyopterus) from Mount Gaoligong, 1021 southwest China. ZooKeys, 864, 147–160. 1022 LONG, A., BLOCH, J. I. and SILCOX, M. T. 2015. Quantification of neocortical ratios 1023 in stem primates. American Journal of Physical Anthropology, 157, 363–373. 1024 LU, X., GE, D., XIA, L., HUANG, C. and YANG, Q. 2014. Geometric morphometric 1025 study of the skull shape diversification in Sciuridae (Mammalia, Rodentia). 1026 Integrative Zoology, 9, 231–245. 1027 LUO, Z.-X., MENG, Q.-J., GROSSNICKLE, D. M., LIU, D., NEANDER, A. I., 1028 ZHANG, Y.-G. and JI, Q. 2017. New evidence for mammaliaform ear evolution 1029 and feeding adaptation in a Jurassic ecosystem. Nature, 548, 326–329. 1030 MACE, G. M., HARVEY, P. H. and CLUTTON‐BROCK, T. H. 1981. Brain size and 1031 ecology in small mammals. Journal of Zoology, 193, 333–354. 1032 MACRINI, T. E., ROUGIER, G. W. and ROWE, T. 2007. Description of a cranial 1033 endocast from the fossil mammal Vincelestes neuquenianus (Theriiformes) and 42 1034 its relevance to the evolution of endocranial characters in therians. The 1035 Anatomical Record, 290, 875–892. 1036 MARTIN, R. D. 1990. Evolution of the primate central nervous system. In Primate 1037 Origins and Evolution: A Phylogenetic Reconstruction, Princeton University 1038 Press, Princeton, New Jersey, 357–426 pp. 1039 MAYR, H. 1979. Gebissmorphologische Untersuchungen an miozänen Gliriden 1040 Mammalia, Rodentia Süddeutschlands.PhD Dissertation, Ludwig-Maximillians- 1041 Universität München, Munich, 380pp. 1042 MCKENNA, M. C. and BELL, S. K. 1997. Classification of Mammals. Columbia 1043 University Press, New York. 1044 MEIER, P. T. 1983. Relative brain size within the North American Sciuridae. Journal of 1045 Mammalogy, 64, 642–647. 1046 MEIN, P. 1970. Les sciuroptères (Mammalia, Rodentia) néogènes d’Europe Occidentale. 1047 Geobios, 3, 7–77. 1048 MEIN, P. and ROMAGGI, J.-P. 1991. Un gliridé (mammalia, rodentia) planeur dans le 1049 miocène supérieur de l’Ardèche: Une adaptation non retrouvée dans la nature 1050 actuelle. Geobios, 24, 45–50. 1051 MENG, J., HU, Y. and LI, C. 2003. The osteology of Rhombomylus (Mammalia, Glires): 1052 implications for phylogeny and evolution of Glires. Bulletin of the American 1053 Museum of Natural History, 275, 1–247. 1054 ———, ———, WANG, Y., WANG, X. and LI, C. 2006. A Mesozoic gliding mammal 1055 from northeastern China. Nature, 444, 889–893. 1056 MERCER, J. M. and ROTH, V. L. 2003. The effects of Cenozoic global change on 1057 squirrel phylogeny. Science, 299, 1568–1572. 43 1058 MURPHY, W. J., FOLEY, N. M., BREDEMEYER, K. R., GATESY, J. and SPRINGER, 1059 M. S. 2021. Phylogenomics and the genetic architecture of the placental mammal 1060 radiation. Annual Review of Animal Biosciences, 9, 29–53. 1061 MURRANT, M. N., BOWMAN, J., GARROWAY, C. J., PRINZEN, B., MAYBERRY, 1062 H. and FAURE, P. A. 2013. Ultrasonic vocalizations emitted by flying squirrels. 1063 PLoS ONE, 8, e73045. 1064 MUUL, I. and ALLEY, J. W. 1963. Night gliders of the woodlands: vociferous 1065 Glaucomys rarely appears. Natural History Magazine, 18–25. 1066 NEWAR, S. L. and BOWMAN, J. 2020. Think before they squeak: Vocalizations of the 1067 squirrel family. Frontiers in Ecology and Evolution, 8, 193. 1068 NICOLAS, V., WENDELEN, W., BARRIERE, P., DUDU, A. and COLYN, M. 2008. 1069 Morphometric variation in Hylomyscus alleni and H. stella (Rodentia: Muridae), 1070 and description of a new species. Journal of Mammalogy, 89, 222–231. 1071 O’LEARY, M. A., BLOCH, J. I., FLYNN, J. J., GAUDIN, T. J., GIALLOMBARDO, 1072 A., GIANNINI, N. P., GOLDBERG, S. L., KRAATZ, B. P., LUO, Z.-X., MENG, 1073 J., NI, X., NOVACEK, M. J., PERINI, F. A., RANDALL, Z. S., ROUGIER, G. 1074 W., SARGIS, E. J., SILCOX, M. T., SIMMONS, N. B., SPAULDING, M., 1075 VELAZCO, P. M., WEKSLER, M., WIBLE, J. R. and CIRRANELLO, A. L. 1076 2013. The Placental Mammal Ancestor and the Post–K-Pg Radiation of 1077 Placentals. Science, 339, 662–667. 1078 ORLIAC, M. J. and GILISSEN, E. 2012. Virtual endocranial cast of earliest Eocene 1079 Diacodexis (Artiodactyla, Mammalia) and morphological diversity of early 1080 artiodactyl brains. Proceedings of the Royal Society B: Biological Sciences, 279, 1081 3670–3677. 44 1082 PAGEL, M. 1999. Inferring the historical patterns of biological evolution. Nature, 401, 1083 877–884. 1084 PILLERI, G., GIHR, M. and KRAUS, C. 1984. Cephalization in rodents with particular 1085 reference to the Canadian beaver (Castor canadensis). In PILLERI, G. (ed.) 1086 Investigations on Beavers, Brain Anatomy Institute, Berne, 11–102 pp. 1087 PINHEIRO, J., BATES, D., DEBROY, S., SARKAR, D. and R CORE TEAM. 2020. 1088 Nlme: Linear and Nonlinear Mixed Effects Models. R Package. . 1089 PRIETO, J. and RUMMEL, M. 2016. Some considerations on small mammal evolution 1090 in Southern Germany, with emphasis on Late Burdigalian-Earliest Tortonian 1091 (Miocene) cricetid rodents. Comptes Rendus Palevol, 15, 837–854. 1092 R CORE TEAM. 2020. R: A language and environment for statistical computing. R 1093 Foundation for Statistical Computing, Vienna. 1094 RADINSKY, L. 1976. Oldest horse brains: more advanced than previously realized. 1095 Science, 194, 626–627. 1096 RONQUIST, F. and HUELSENBECK, J. P. 2003. MrBayes 3: Bayesian phylogenetic 1097 inference under mixed models. Bioinformatics, 19, 1572–1574. 1098 ROTH, V. L. and THORINGTON, R. W. Jr. 1982. Relative brain size among African 1099 squirrels. Journal of Mammalogy, 63, 168–173. 1100 SILCOX, M. T., BENHAM, A. E. and BLOCH, J. I. 2010. Endocasts of Microsyops 1101 (Microsyopidae, Primates) and the evolution of the brain in primitive primates. 1102 Journal of Human Evolution, 58, 505–521. 1103 SIMPSON, G. G. 1945. The principles of classification and a classification of mammals. 1104 Bulletin of the American Museum of Natural History, 85, 1–350. 1105 SINITSA, M. V., POGODINA, N. V. and КRYUCHKOVA, L. Y. 2019. The skull of 1106 Spermophilus nogaici (Rodentia: Sciuridae: Xerinae) and the affinities of the 45 1107 earliest Old World ground squirrels. Zoological Journal of the Linnean Society, 1108 186, 826–864. 1109 SPRINGER, M. S., MURPHY, W. J., EIZIRIK, E. and O’BRIEN, S. J. 2003. Placental 1110 mammal diversification and the Cretaceous–Tertiary boundary. Proceedings of 1111 the National Academy of Sciences, 100, 1056–1061. 1112 STEPPAN, S. J., STORZ, B. L. and HOFFMANN, R. S. 2004. Nuclear DNA phylogeny 1113 of the squirrels (Mammalia: Rodentia) and the evolution of arboreality from c- 1114 myc and RAG1. Molecular Phylogenetics and Evolution, 30, 703–719. 1115 STORCH, G., ENGESSER, B. and WUTTKE, M. 1996. Oldest fossil record of gliding 1116 in rodents. Nature, 379, 439–441. 1117 SYMONDS, M. R. E. and BLOMBERG, S. P. 2014. A Primer on Phylogenetic 1118 Generalised Least Squares. In GARAMSZEGI, L. Z. (ed.) Modern Phylogenetic 1119 Comparative Methods and Their Application in Evolutionary Biology: Concepts 1120 and Practice, Springer, Berlin, Heidelberg, 105–130 pp. 1121 THORINGTON, R. W., Jr. 1984. Flying squirrels are monophyletic. Science, 225, 1048– 1122 1050. 1123 ——— and DARROW, K. 2000. Anatomy of the squirrel wrist: bones, ligaments and 1124 muscles. Journal of Morphology, 246, 85–102. 1125 ———, PITASSY, D. and JANSA, S. A. 2002. Phylogenies of flying squirrels 1126 (Pteromyinae). Journal of Mammalian Evolution, 9, 99–135. 1127 ———, SCHENNUM, C. E., PAPPAS, L. A. and PITASSY, D. 2005. The difficulties 1128 of identifying flying squirrels (Sciuridae: Pteromyini) in the fossil record. Journal 1129 of Vertebrate Paleontology, 25, 950–961. 46 1130 WAHLERT, J. H. 1985. Cranial foramina of rodents. In LUCKETT, W. P. and 1131 HARTENBERGER, J.-L. (eds.) Evolutionary Relationships among Rodents, 1132 Springer US, Boston, MA, 311–332 pp. 1133 ———. 2000. Morphology of the auditory region in Paramys copei and other Eocene 1134 rodents from North America. American Museum Novitates, 3307, 1–16. 1135 WIBLE, J. R. 1984. The ontogeny and phylogeny of the mammalian cranial arterial 1136 pattern.PhD Dissertation, Duke University, Durham, North Carolina, USA, 1137 705pp. 1138 ———. 2008. On the cranial osteology of the Hispaniolan solenodon, Solenodon 1139 paradoxus Brandt, 1833 (Mammalia, Lipotyphla, Solenodontidae). Annals of 1140 Carnegie Museum, 77, 321–402. 1141 ——— and SHELLEY, S. L. 2020. Anatomy of the petrosal and middle ear of the brown 1142 rat, Rattus norvegicus (Berkenhout, 1769) (Rodentia, Muridae). Annals of 1143 Carnegie Museum, 86, 1–35. 1144 YAO, L., BROWN, J.-P., STAMPANONI, M., MARONE, F., ISLER, K. and MARTIN, 1145 R. D. 2012. Evolutionary change in the brain size of bats. Brain, Behavior and 1146 Evolution, 80, 15–25. 1147 1148 47 1149 FIGURE CAPTIONS 1150 1151 FIG. 1. Phylogenetic relationships and age for Sciuridae and Ischyromyidae discussed in 1152 the text. Within the Sciurinae, the flying (Pteromyini) and tree squirrel (Sciurini) clades 1153 are highlighted and Miopetaurista crusafonti is indicated in bold. Tree topology is mostly 1154 based in Korth & Emry (1991), Meng et al. (2003) and Casanovas-Vilar et al. (2018) (see 1155 text for details). This tree is used to perform the Phylogenetic Generalised Least Squares 1156 (PGLS) analyses (see text for details). 1157 48 1158 1159 FIG. 2. Virtual endocast of Miopetaurista crusafonti (SNSB 1978 V 1) inside the 1160 translucent cranium in: A, dorsal; B, ventral; C, left lateral; D, right lateral view. Scale 1161 bar is 10 mm. See Appendix S1 for an animated rendering of the specimen. 1162 49 1163 1164 FIG. 3. Cranium of Miopetaurista crusafonti (SNSB 1978 V 1) in: A, dorsal; B, ventral; 1165 C, left lateral; D, right lateral view. E, detail of the upper left cheek teeth in occlusal view. 1166 Scale bar is 10 mm in figures A–D and 5 mm in figure E. Cheek teeth measurements and 1167 comparisons are given in Table 1 and Grau-Camats et al. (2021, table S1), respectively, 1168 while cranial measurements and comparisons are given in Table 2. See Appendix S1 for 1169 an animated rendering of the specimen. 1170 50 1171 1172 FIG. 4. Three-dimensional model of the cranium of Miopetaurista crusafonti (SNSB 1978 1173 V 1) based on μCT data and schematic drawing indicating the different anatomical 1174 elements in: A, ventral view; B dorsal view. Gray shading indicates broken areas. Scale 1175 bar is 10 mm. See Appendix S1 for an animated rendering of the specimen. 1176 51 1177 1178 FIG. 5. Three-dimensional model of the cranium of Miopetaurista crusafonti (SNSB 1978 1179 V 1) based on μCT data and schematic drawing indicating the different anatomical 1180 elements in: A, left lateral view; B, left petrosal in medial view. Gray shading indicates 1181 broken areas. Dashed lines indicate the approximate situation of anatomical features that 1182 cannot be appreciated in this view. Scale bar is 10 mm. See Appendix S1 for an animated 1183 rendering of the specimen. 1184 52 1185 1186 FIG. 6. Crania of extinct and extant flying squirrels in dorsal, ventral and lateral views. 1187 A, Miopetaurista crusafonti (SNSB 1978 V 1); B, Miopetaurista neogriviensis (virtual 1188 reconstruction based on IPS56468h and IPS88677); C, Petaurista petaurista (ZMA 1189 13418); D, Aeromys tephromelas (RMNH 24076); E, Petinomys sagitta (RMNH 15512). 1190 Scale bars are 10 mm. Cranial measurements and comparisons are given in Table 2. 1191 53 1192 1193 FIG. 7. Endocranial morphology of the Miocene fossil flying squirrel Miopetaurista 1194 crusafonti (SNSB 1978 V 1). Virtual endocast of M. crusafonti in: A, dorsal; B, ventral; 1195 C, left lateral view. Scale bar is 10 mm. See Appendix S1 for an animated rendering of 1196 the endocast. 1197 54 1198 1199 FIG. 8. Endocasts of various extinct and extant rodents in dorsal, ventral and right lateral 1200 views. A, Petaurista petaurista (USNM 589079); B, Glaucomys volans (AMNH 1201 240290); C, Cedromus wilsoni (USNM 256584); D, Protosciurus cf. rachelae (YPM 1202 14737); E, Paramys delicatus (AMNH 12506). Scale bars are 10 mm. 1203 55 1204 1205 FIG. 9. Bivariate plots and boxplots for endocast measurements in Ischyromyidae and 1206 extant and extinct Sciuridae. A, bivariate plot of log 10 (olfactory bulb volume) versus 1207 log 10 (endocranial volume); B, bivariate plot of log 10 (olfactory bulb volume) versus 1208 log 10 (body mass); C, boxplot of neocortical surface area ratio; D, bivariate plot of log 56 1209 10 (neocortical surface area) versus log 10 (endocranial surface area); E, bivariate plot of 1210 log 10 (petrosal lobule volume) versus log 10 (endocranial volume); F, bivariate plot of 1211 log 10 (petrosal lobule volume) versus log 10 (body mass). Phylogenetic generalized 1212 least-square (PGLS) regression lines are included for extant Sciuridae. The position of 1213 Miopetaurista crusafonti SNSB 1978 V 1 from Gumpersdorf as well as that of the extant 1214 Petaurista petaurista (thick arrow) is indicated in figures (A–B) and (D–F). PGLS 1215 regression results are given in Table 4. Residuals associated with the different PGLS 1216 regressions are given in Grau-Camats et al. (2021, table S6). Body mass, olfactory bulb, 1217 petrosal lobule and endocast volume (in mm3) and neocortical surface area for 1218 Miopetaurista crusafonti and other extinct and extant sciurid and ischyromyid rodents are 1219 given in Grau-Camats et al. (2021, table S4). 1220 57 1221 1222 FIG. 10. Relationship between endocranial volume and body mass and encephalization 1223 quotients for extinct and extant Sciuridae and Ischyromyidae (see Table 1). A, Boxplot 1224 of EQs based on the equation by Pilleri et al. (1984). B, Bivariate plot of log 10 1225 (endocranial volume) versus log 10 (body mass) for Sciuridae and Ischyromyidae. 1226 Phylogenetic generalised least-square (PGLS) regression line is included for extant 1227 Sciuridae. The position of Miopetaurista crusafonti SNSB 1978 V 1 from Gumpersdorf, 1228 as well as that of the extant Petaurista petaurista (thick arrow) is indicated in figure B. 1229 PGLS regression results are given in Table 4. Residuals associated with the different 1230 PGLS regressions are given in Grau-Camats et al. (2021, table S7). Body mass and 1231 endocast volume (in mm3) for Miopetaurista crusafonti and other extinct and extant 1232 sciurid and ischyromyid rodents are given in Grau-Camats et al. (2021, table S4). 58 1233 TABLE 1. Cheek tooth measurements for Miopetaurista crusafonti 1978 V 1 from 1234 Gumpersdorf. ‘l’ stands for left side and ‘r’ stands for right side. All measurements are in 1235 mm. Element Length Width P3 l 2.49 2.27 P4 l 3.97 4.05 M1 l 3.65 4.51 M2 l 3.72 4.44 M3 l 3.62 3.9 P3 r 2.37 2.16 M1 r 3.54 4.46 M2 r 3.79 4.42 M3 r 3.73 3.99 1236 1237 59 1238 TABLE 2. Cranial measurements for extinct and extant flying squirrels as compared to 1239 Miopetaurista crusafonti 1978 V 1 from Gumpersdorf and other extant and extinct flying 1240 squirrels. Cranial measurements for the virtually reconstructed skull of M. neogrivensis 1241 are after Casanovas-Vilar et al. (2018) . Acronyms and definitions of the measurements 1242 follow Nicolas et al. (2008) and Bertrand et al. (2016a): BNAS, maximum breadth of the 1243 nasals; BRCA, maximum breadth of the braincase; BULL, maximum length of the 1244 auditory bulla; CTL, length of the upper cheek tooth row; DIA, length of the diastema 1245 (from the alveolus of the incisor to the alveolus of the P3); HEBA, henselion–basion 1246 length; HEPA, henselion–palation length; IF, length of the incisive foramen; INTE, 1247 smallest interorbital breadth; LNAS, maximum length of the nasals; PALA, smallest 1248 palatal breadth; RB, maximum rostrum breadth; RH, mediosagittal projection of rostrum 1249 height at the anterior border of the M1; SL, maximum length of the skull. All 1250 measurements are in mm. Estimated measurements (due to breakage or distortion) are 1251 given in parentheses, while ‘>’ indicates that the measurement could not be reliably taken 1252 but certainly exceeded the reported value. 1253 60 Species and specimen SL HEBA HEPA IF DIA INTE ZYGO PALA CTL ZYPL BNAS LNAS BULL BRCA RH RB Miopetaurista (67.7) (59.4) (36.0) – >14.7 (9.7) – 7.3 17.0 8.6 – – – (27.6) >19.4 >12.1 crusafonti (1978 V1) Miopetaurista neogrivensis (72.4) (58.6) (38.8) 6.8 16.1 (20.1) – (9.0) 18.4 (8.6) – >25.0 19.0 (30.3) (22.7) (20.0) (IPS56468h) Miopetaurista neogrivensis (73.6) (61.8) 38.1 6.9 17.8 – (54.6) (11.2) 17.4 11.3 – >22.0 (18.9) (37.2) (24.1) (21.7) (IPS88677) Miopetaurista neogrivensis 69.8 (58.8) 39.4 6.8 16.4 19.6 46.4 9.5 18.4 10.4 – >19.2 19.0 37.2 25.1 (23.8) (virtual recons.) Petaurista petaurista 64.5 54.1 30.7 3.3 12.7 16.7 44.8 7.3 15.5 5.9 12.0 19.4 13.2 29.7 20.2 14.5 (ZMA131418) Eupetaurus cinereus 72.3 58.5 35.7 6.2 14.7 18.7 44.1 12.0 19.6 9.0 13.0 24.7 14.7 26.7 23.9 13.8 (19524) Aeromys tephromelas 66.0 53.8 27.6 5.3 15.6 14.2 40.7 8.4 12.3 7.7 11.3 17.2 15.1 31.0 22.0 15.2 (24076) Belomys pearsonii 43.2 34.3 20.8 2.9 9.3 8.5 26.0 4.8 9.9 4.4 6.4 13.1 10.2 20.1 12.7 8.4 (56.046) Pteromys volans 36.5 29.4 15.9 4.3 7.5 9.9 22.6 5.4 6.6 4.6 6.1 11.6 10.5 16.4 11.8 6.4 (40035) Hylopetes sagitta 35.8 20.4 15.8 2.2 7.9 11.3 20.3 4.9 7.3 4.2 5.4 9.7 8.6 16.9 10.9 7.7 (15512) Glaucomys volans 34.4 28.3 16.0 2.0 7.3 8.3 20.8 4.7 6.3 4.4 5.0 10.0 9.8 15.7 10.4 7.4 (19786) Glaucomys sabrinus 39.2 31.3 19.1 2.1 9.5 7.6 23.7 4.5 7.5 4.4 5.7 11.8 9.5 18.3 12.0 7.9 (IPS60584) Iomys horsfieldii 30.5 23.8 15.7 1.1 5.7 8.3 19.1 5.1 7.2 2.8 4.8 8.3 7.4 15.7 8.6 7.2 (15937) 61 1254 1255 TABLE 3. Endocranial volume, olfactory bulb, petrosal lobules, neocortical ratios and encephalization quotients for Miopetaurista crusafonti and 1256 published extant Pteromyini (Bertrand et al. 2017). To calculate the neocortical surface area ratio, we used the following formula NS1*2/TS. 1257 Volume and surface area ratios are expressed as percentages. The encephalization quotients were calculated using the equations of Eisenberg 1258 (1981) and Pilleri et al. (1984). Neocortical Olfactory Petrosal Total EQ surface area bulb lobule EQ Pilleri et Subtribe Species and specimen endocast (Eisenberg, ratio (%) volume volume ratio al. (1984) volume (cm3) 1981) NS1*2/TS ratio (%) (%) Pteromyina Miopetaurista crusafonti (1978 V1) 10.82 30.19 1.21 1.92 1.01 1.13 Pteromyina Aeromys tephromelas (USNM 481190) 11.46 35.19 2.85 1.45 1.35 1.45 Pteromyina Petaurista petaurista (USNM 589079) 12.32 35.28 1.64 1.62 1.25 1.38 Pteromyscus pulverulentus (USNM Pteromyina 481178) 3.62 33.82 2.81 1.37 1.32 1.23 Pteromyina Pteromys volans (USNM 172622) 2.33 34.11 1.75 1.61 1.33 1.17 Glaucomyina Glaucomys volans (AMNH 240290) 2.01 35.69 3.49 1.68 1.68 1.40 Glaucomyina Hylopetes spadiceus (USNM 488639) 2.12 36.73 3.30 0.85 1.44 1.24 Glaucomyina Petinomys setosus (USNM 488674) 1.51 35.76 3.10 - 1.73 1.38 1259 62 1260 TABLE 4. Phylogenetic generalized least-square linear regressions (PGLS) for endocast measurements and for total endocranial volume vs. body 1261 size in our sample (see Figs 9–10). For each regression, the values for the intercept, slope and associated significance, Pagel’s λ, the pooled estimate 1262 of the residual standard error (RSE), degrees of freedom (df) for the model and residuals, 95 % confidence intervals for the slope (CI), and 1263 correlation coefficient (R2) obtained from predicted and residual values. Residuals associated with the different regressions are provided in Grau- 1264 Camats et al. (2021, tables S6–S7). Body mass, olfactory bulb, petrosal lobule and endocast volume (in mm3) as well as neocortical surface area 1265 for Miopetaurista crusafonti and other extinct and extant sciurid and ischyromyid rodents are given in Grau-Camats et al. (2021, table S4). p-value R2 R2 Equation intercept slope λ RSE df 95% CI (slope) (predicted) (residuals) Log (olfactory bulb volume) / Log (endocranial 1.462 0.928 0.000 0.937 0.181 36, 34 [0.808, 1.049] 0.909 0.885 volume) Log (olfactory bulb volume) / Log (body mass) 0.852 0.509 0.000 0.611 0.150 36, 34 [0.427, 0.592] 0.865 0.856 Log (neocortical surface area) / Log (endocranial -0.405 0.981 0.000 1.014 0.111 36, 34 [0.923, 1.039] 0.939 0.905 surface area) Log (neocortical surface area) / Log (endocranial -0.430 0.998 0.000 -0.094 0.018 24, 22 [0.960, 1.036] 0.993 0.992 surface area) [excluding Ischyromyidae] Log (petrosal lobule volume) / Log (endocranial -0.666 0.708 0.000 0.352 0.181 35, 33 [0.482, 0.933] 0.590 0.557 volume) Log (petrosal lobule volume) / Log (body mass) 0.3461 0.292 0.006 0.396 0.221 35, 33 [0.151, 0.434] 0.401 0.356 Log (endocranial volume) / Log (body mass) -0.531 0.488 0.000 0.429 0.104 36, 34 [0.427, 0.549] 0.897 0.889 1266 63 1267 64