This is the accepted version of the journal article: Urciuoli, Alessandro; Kubat, Jülide; Schisanowski, Lisa; [et al.]. «Cochlear morphology of Indonesian Homo erectus from Sangiran». Journal of human evolution, Vol. 165 (April 2022), art. 103163. DOI 10.1016/j.jhevol.2022.103163 This version is available at https://ddd.uab.cat/record/257222 under the terms of the license 1 Cochlear morphology of Indonesian Homo erectus from Sangiran 2 3 Abstract 4 Homo erectus s.l. is key for deciphering the origin and subsequent evolution of genus 5 Homo. However, the characterization of this species is hindered by the existence of multiple 6 variants in both mainland and insular Asia, as a result of divergent chronogeographical 7 evolutionary trends, genetic isolation, and interbreeding with other human species. Previous 8 research has shown that cochlear morphology embeds taxonomic and phylogenetic 9 information that may help infer the phylogenetic relationships among hominin species. Here 10 we describe the cochlear morphology of two Indonesian H. erectus individuals (Sangiran 2 11 and 4), and compare it with a sample of australopiths, Middle to Late Pleistocene humans, 12 and extant humans by means of linear measurements and both principal components and 13 canonical variates analyses performed on shape ratios. Our results indicate that H. erectus 14 displays a mosaic morphology that combines plesiomorphic (australopith-like) features, 15 (such as a chimp-like round cochlear cross section and low cochlear thickness) with derived 16 characters of later humans (a voluminous and long cochlea, possibly related to hearing 17 abilities)—consistent with the more basal position of H. erectus. Our results also denote 18 substantial variation between the two studied individuals, particularly in the length and radius 19 of the first turn, as well as cross-sectional shape. Given the small size of the available 20 sample, it is not possible to discern whether such differences merely reflect intraspecific 21 variation among roughly coeval H. erectus individuals or whether they might result from 22 greater age differences between them than currently considered. However, our results 23 demonstrate that most characters found in later humans were already present in Indonesian 24 H. erectus, with the exception of Neanderthals, which display an autapomorphic condition 25 relative to other Homo species. 26 27 Keywords: Fossil humans; Inner ear; Cochlea; Morphometrics; Human Evolution; 28 Southeastern Asia. 29 1. Introduction 30 1.1. Homo erectus from Indonesia 31 In 1891, a ‘pithecoid’ calotte and a ‘humanoid’ femur allowing for a bipedal stance were 32 discovered at the Trinil site in the island of Java. The following description of 33 Pithecanthropus erectus by Dubois (1892) represented a drastic paradigm shift in human 34 evolutionary studies, until then centered in Europe. Subsequent surveys in southeastern 35 Asia by von Koenigswald during the late 1930s led to the discovery of several hominin- 36 bearing sites within the Solo River Basin. The Sangiran Dome, found along a tributary of the 37 Solo River, is one of the richest fossil sites, having yielded several dozen hominin 38 specimens, comprising craniodental and postcranial material and including some remarkably 39 complete crania and numerous calvariae1 (Koenigswald and Weidenreich, 1939; 40 Weidenreich, 1945; Jacob, 1973; Day, 1986). The dome is formed by a partly eroded 41 anticlinal holding Neogene and Quaternary lacustrine and fluviatile deposits, with two 42 hominin-bearing stratigraphic units: the Lower Pleistocene Pucangan (Sangiran) and the 43 Middle Pleistocene Kabuh (Bapang) Formations. Despite considerable efforts, the lower 44 boundary of the Pucangan Formation, where the oldest hominin remains were found, and 45 thus that of the arrival of H. erectus at Java, is still contentious, being dated to either ≥1.5 Ma 46 (Larick et al., 2001; Antón and Swisher, 2004; Zhu et al., 2004; Kaifu et al., 2011), ~1.4 Ma 47 (Matsu’ura et al., 2005; 2020), or ~0.9 Ma (Sémah et al., 2000; Hyodo, 2001). While the 48 specimens recovered from the Kabuh Formation clearly belong to H. erectus due to their 49 derived human morphology (Kaifu et al., 2005), the taxonomic allocation of the fossils from 50 the Pucangan Formation is still debated (e.g., Weidenreich, 1945; Koenigswald, 1951; 51 Krantz, 1975; Franzen, 1985a,b; Schwartz and Tattersal, 2003; Schwartz, 2016). 1 Here we preferred to use ‘calvaria’, rather than ‘calvarium’, for referring to the bones constituting the neurocranium. While both wordings are viable singular form alternatives, we decided to follow the main definition by White et al. (2012): “The calvaria (or calvarium) [emphasis in the original] is the cranium without the face.”, which presents calvarium as a secondary option. Furthermore, we considered more advisable to rely on the original Latin word ‘calvaria’ (feminine, plural: calvariae), rather than on the New Latin term ‘calvarium’ (neuter, plural: calvaria) that derived from the former later in time. 52 Over the years, taxonomic opinions about Early Pleistocene Javanese hominins have 53 changed due to different interpretations of the considerable morphological variation 54 expressed by the Pucangan specimens and the presence of homoplastic characters in the 55 dentition of Early Pleistocene Homo and Pongo (Smith et al., 2009, 2018; Zanolli et al., 56 2019). At present, following Mayr’s (1950) lumping influence, almost all Sangiran hominin 57 specimens are customarily referred to H. erectus (albeit not without criticism; Schwartz 58 2016). However, initially they were variously attributed to multiple taxa (such as 59 Pithecanthropus robustus, Pithecanthropus dubius, H. erectus modjokertensis, 60 Meganthropus paleojavanicus, or even Pongo) and the allocation of many has changed over 61 time or is still pending (Koenigswald and Weidenreich, 1939; Weidenreich, 1945; 62 Koenigswald, 1950; Krantz, 1975, 1994; Tyler, 2003, 2004, 2006). The great variability of H. 63 erectus and the taxonomic uncertainties about this species probably stem from the complex 64 evolutionary context posed by the physiography of the Indonesian archipelago. Eustatic sea- 65 level changes during the Plio-Pleistocene likely led to several isolation phases that would 66 have had a major role in shaping the faunal diversity of this region, including H. erectus 67 (Djubiantono and Sémah, 1991; Antón, 2003; Schwartz and Tattersall, 2003). This is further 68 complicated by the possible inclusion of nonhominin specimens within the H. erectus 69 hypodigm, as suggested by the resurrection by Zanolli et al. (2019) of Meganthropus 70 palaeojavanicus as a valid taxon distinct from H. erectus. 71 The heterogeneity of the Sangiran hominin sample is best exemplified by the partial 72 calvaria Sangiran 2 (S2) and the fragmentary cranium Sangiran 4 (S4; Fig. 1a, b). These 73 specimens represent opposite extremes in terms of gracile and robust cranial proportions, 74 respectively (Santa Luca, 1980). The S2 calvaria shows clear similarities with the H. erectus 75 holotype (Trinil 2; Schwartz and Tattersall, 2003), both in size (being fairly small) and 76 morphological details (e.g., the shape of the superior, lateral, and posterior profiles, the low 77 bregmatic swelling, and the slight frontal ‘keel’), thus warranting a straightforward attribution 78 to H. erectus (e.g., Koenigswald and Weidenreich, 1939). Conversely, S4 has always been 79 subject to controversy, being originally designated as the holotype of Pithecanthropus 80 robustus, and subsequently considered by Weidenreich (1945, 1956) to be an intermediate 81 form between a giant hominid species (such as Gigantopithecus) and more gracile H. 82 erectus. S4 is indeed thick-boned, being also distinctive by the large and medially located 83 mastoid process and a peculiar drainage pattern in the sigmoid sinus (Schwartz, 2016). The 84 retention of some primitive features in the palate (e.g., the precanine diastemata and the 85 slightly anteriorly projecting canines) even led some authors to question its hominin status 86 (Krantz, 1975, 1994; Tyler, 2003, 2004, 2006), although such a claim seems no longer 87 tenable (Durband, 2008). 88 89 1.2. Cochlear morphology and human evolution 90 Given all these persisting uncertainties about Indonesian H. erectus, the analysis of inner 91 anatomical structures that might provide additional taxonomic and phylogenetic information 92 is of utmost significance. This is the case of the bony labyrinth of the inner ear, which has 93 been extensively investigated in hominids (Spoor and Zonneveld, 1994; 1998; Rook et al., 94 2004; Braga et al., 2015; Le Maître et al., 2017; Urciuoli et al., 2021a) and, especially, 95 hominins (Spoor et al., 1994, 2003; Ni et al., 2010; Braga et al., 2013; Osipov et al., 2013; 96 Wu et al., 2014; Quam et al., 2016; Conde-Valverde et al., 2018; Crevecoeur et al., 2016; 97 Ponce de León et al., 2018; Beaudet, 2019; Beaudet et al., 2019; Ward et al., 2020). The 98 bony labyrinth is composed of the semicircular canals and the vestibule, which encase the 99 vestibular system (in charge of perceiving linear and angular accelerations), as well as the 100 cochlea, which houses the hearing organ. Recent studies have demonstrated that 101 semicircular canals are useful for inferring hominin taxonomic affinities (Braga et al., 2013; 102 Wu et al., 2014; Crevecoeur et al., 2016; Conde-Valverde et al., 2018; Beaudet, 2019; 103 Beaudet et al., 2019) and modern human dispersal distances from Africa (Ponce de León et 104 al., 2018), as well as for phylogenetic reconstruction (Urciuoli et al., 2020; 2021b). In 105 contrast, most studies dealing with the cochlea have investigated the relation between its 106 morphology and hearing function (e.g., Manoussaki et al., 2008; Kirk and Gosselin-Ildari, 107 2009; Braga et al., 2021), although recent analyses have shown that cochlear shape further 108 embeds phylogenetically informative characters (Braga et al., 2015; Conde-Valverde et al., 109 2018; 2019; 2020; Beaudet, 2019; Beaudet et al., 2019). 110 As a result of the larger number of available Neanderthal specimens, their cochlear 111 morphology has been studied in more detail (Spoor et al., 2003; Beals et al. 2016; Conde- 112 Valverde et al., 2019). Previous studies have reported multiple similarities (some interpreted 113 as homologous and others as homoplastic) in cochlear morphology between Neanderthals 114 and modern humans. In particular, both species share a similarly long first turn of the 115 cochlea (Conde-Valverde et al., 2019), a large cochlear volume (Beals et al., 2016), and a 116 thickened cochlear coiling—all of them interpreted as derived relative to early hominins and 117 chimpanzees (Conde-Valverde et al., 2019). In addition, similarities in cochlear volume have 118 been used to infer a comparable hearing range for modern humans and Neanderthals (Beals 119 et al., 2016), while evidence coming from the outer and middle ear has allowed to 120 reconstruct a similar sound power transmission and occupied bandwidth in both taxa 121 (Conde-Valverde et al., 2021). Finally, Conde-Valverde et al. (2019) identified an 122 evolutionary morphocline among Neanderthals, with younger specimens showing some 123 autapomorphic features (e.g., very low number of cochlear turns and extremely short third 124 turn) that are absent from older specimens such as those from Krapina in Croatia (Beals et 125 al., 2016). 126 Information about cochlear morphology in Middle Pleistocene humans is limited to the 127 data available only for the sample from Sima de los Huesos in Spain (Conde-Valverde et al., 128 2019), the Aroeira 3 cranium from Portugal (Conde-Valverde et al., 2020), and Hexian 1 129 from China (Wu et al., 2014). A detailed analysis of cochlear shape in the Iberian hominins 130 showed that they differ from both Neanderthals and modern humans, by possessing an 131 intermediate morphology between Pan troglodytes and Late Pleistocene and Holocene 132 humans. In particular, Middle Pleistocene humans from the Iberian Peninsula display a 133 mosaic of some plesiomorphic hominin features (relatively short first cochlear turn, 134 somewhat rounder cross section, narrow coiling thickness, and fairly small volume) coupled 135 with characters displayed by later humans (long cochlea relative to their body mass, short 136 third turn, and reduced number of turns; Conde-Valverde et al., 2019, 2020). 137 Cochlear shape remains largely unexplored in several hominin species, such as H. 138 erectus. Aside from Spoor’s (1993) seminal work, Braga et al. (2015) first provided 139 information about early hominin cochlear morphology when describing the remains of South 140 African australopiths (Australopithecus and Paranthropus) and Early Pleistocene Homo. 141 These authors highlighted the presence of a phylogenetic signal in the cochlear variables 142 analyzed by them, and suggested that Homo species are apomorphic relative to earlier 143 hominins in the possession of a longer cochlea relative to their body size when the length is 144 standardized by the number of cochlear turns—possibly an adaptation for increased hearing 145 sensitivity at low frequencies (Braga et al., 2015). South African hominins have also been 146 inspected by means of 3D geometric morphometrics techniques (Beaudet, 2019; Beaudet et 147 al., 2019), which allowed to identify a unique coiling morphology (with a very tight 148 spiralization, unlike that of other australopiths and modern humans) in early Homo 149 specimens (especially SK 27) and a few Paranthropus individuals. Unfortunately, this feature 150 has not been analyzed in other early Homo specimens, and the evolution of hominin 151 cochlear morphology during the Early Pleistocene is unknown to a large extent, except for 152 brief reports on the H. erectus s.l. specimens OH 9 from Olduvai, Tanzania (Spoor, 1993; 153 Spoor and Zonneveld, 1994) and Lantian 1 from China (Wu et al., 2014; Wu and Zhang, 154 2016). 155 156 1.3. Aims of this study 157 To fill in the gap of knowledge on the cochlear shape morphology of Early Pleistocene 158 humans and explore its implications for the evolution of this anatomical structure in the 159 origins and early evolution of genus Homo, here we describe and analyze the cochlear 160 morphology of two H. erectus individuals from Sangiran (S2 and S4). This work represents 161 the first detailed description of the cochlear morphology for this species. First, we inspect the 162 cochlear morphology of both specimens by means of linear measurements and shape ratios, 163 and compare it to that of other Plio-Pleistocene hominins and extant humans using different 164 multivariate approaches. Second, we rely on cochlear morphology to evaluate several 165 competing taxonomic attributions proposed for the analyzed specimens as well as to refine 166 previous evolutionary inferences about this structure regarding the genus Homo. Finally, we 167 further infer the hearing ranges of the two H. erectus individuals and compare them with 168 those of modern humans and other hominins. 169 170 2. Materials and methods 171 2.1. Studied material 172 The fossil material analyzed in the present study comprises the S2 calvaria (Koenigswald 173 and Weidenreich, 1939) and S4 fragmentary cranium (partial calvaria and maxilla; 174 Weidenreich, 1945; Schwartz and Tattersall, 2000), both attributed to H. erectus (e.g., 175 Schwartz and Tattersall, 2000, 2003). Both specimens were collected from an area that 176 ranges from the Grenzbank zone and the upper levels of the Upper Sangiran formation (~1.4 177 Ma; Larick et al., 2001; or ~0.9 Ma; Matsu’ura et al., 2020) and are currently housed at the 178 Senckenberg Research Institute and Natural History Museum Frankfurt in Germany as part 179 of the Gustav Heinrich Ralph von Koenigswald collection. 180 181 2.2. Comparative sample 182 The comparative sample (n = 42; Table 1) is composed of modern humans (n = 10), 183 Neanderthals (n = 7), Sima de los Huesos humans (n = 10), Australopithecus spp. (n= 10), 184 Paranthropus spp. (n = 3), South African early Homo (n = 2), and Middle Pleistocene Homo 185 sp. (n = 1). The australopith and early Homo specimens come from different South African 186 sites (see Supplementary Online Material [SOM] Table S1), which span from ~4.02–3.67 Ma 187 to 2.01 Ma in the case of Australopithecus (Pickering and Kramers, 2010; Granger et al., 188 2015), and date to ~2.19–1.80 Ma regarding Paranthropus and early Homo (McKee et al., 189 1995; Gibbon et al., 2014). The comparative sample of Middle and Late Pleistocene 190 hominins includes individuals from the Sima de los Huesos site (0.43 Ma, Spain; Arsuaga et 191 al., 2014) together with Neanderthals from France and Israel (71–60 ka; Schwarcz et al., 192 1989; Rink et al., 2001; Rendu et al., 2014; Guérin et al., 2015; Frouin et al., 2017), as well 193 as the Aroeira 3 cranium from Gruta da Aroeira, Portugal, dated to 0.436–0.389 Ma (Daura 194 et al., 2017). Three-dimensional geometric morphometric techniques could not be applied 195 because 3D models are not available for the Sima de los Huesos and Neanderthal 196 specimens, and hence we relied on linear measurements. As a cautionary note, it must be 197 stressed that some individuals from the comparative sample are subadults (StW 329, SK 27, 198 La Ferrassie 8, Cr. 3, Cr. 7, AT-421, AT-1907) and that some previous analyses have 199 identified ontogenetic changes in the relative orientation of the cochlea relative to the middle 200 and outer ear regions (McRackan et al., 2012). However, the bony labyrinth fully ossifies 201 between 17–19 weeks of gestation and reaches its adult size before birth (Scheuer and 202 Black, 2000; Jeffery and Spoor, 2004), while the variables here employed are unrelated to 203 the orientation of other cranial structures. Therefore, the inclusion of juveniles in the 204 comparative sample—as it is often the case in the literature (e.g., Conde-Valverde et al., 205 2019; Beaudet et al., 2019)—seems warranted. 206 207 2.3. Sample preparation 208 The S2 and S4 partial crania were µCT-scanned at the Steinmann-Institute of Geology, 209 Mineralogy und Palaeontology at the University of Bonn, using a phoenix V|tome|x S240. 210 For S2, a total of 1300 projections were obtained using voltage = 200 kV and current = 200 211 μA, with magnification = 4.58, resulting in a voxel size of 0.087 mm after reconstruction. A 212 total of 1500 projections were obtained for S4 with voltage = 180 kV and current = 180 μA, 213 and magnification = 4.54, achieving a voxel size of 0.088 mm after reconstruction. Due to 214 their preservation state, only the right bony labyrinth of each individual was used. The µCT 215 scans of the australopith, early Homo, and modern human samples were performed at 216 different facilities (see SOM Table S1 for further information), and reconstructed with voxel 217 sizes ranging between 0.0135 and 0.088 mm. The reconstructed slice stacks were 218 segmented in Avizo v. 7.0 (FEI Visualization Sciences Group, Houston) following the 219 protocol defined by Kirk and Gosselin-Ildari (2009), which considers the variation in bone 220 density throughout the petrosal. The high-density threshold (HT) has been identified at the 221 first cochlear turn between the lumen and the cortical bone by averaging the maximum and 222 minimum gray values (Spoor and Zonneveld, 1995). Similarly, we identified a low-density 223 threshold (LT) at the boundary between the less dense modiolar bone and the cochlear 224 lumen. Finally, we obtained a high-adjusted threshold (HAT) by averaging LT and HT. The 225 volume of each individual was then obtained as the average of the volumes measured with 226 the LT and HAT. Because the cochlear cavity of some of the fossil specimens was partly or 227 completely filled with matrix (S4, MLD 37/38, STS5, STS19, and SK27), the segmentation 228 was carried out using the HAT only, as already performed by Conde-Valverde et al. (2019) in 229 the case of sediment filled cochlear lumen, with additional manual corrections when there 230 was not enough contrast at the boundary between the bone and the sediment. In the case of 231 the Aroeira 3 individual, we used the reconstructed surface provided by Conde-Valverde et 232 al. (2018: SOM File S1). The 3D models of the cochlea were obtained by separating it from 233 the rest of the bony labyrinth in Avizo, using the position of the round window and of the 234 basilar gap (the gap between the primary and secondary osseous spiral laminae) as 235 reference (Kirk and Gosselin-Ildari, 2009). The meshes were then imported into Geomagic 236 Studio v. 2014.3.0 (3D Systems, Rock Hill) to remove possible self-intersections and non- 237 manifold faces, to fill the holes produced by the cutting with a flat surface, and eventually 238 measure cochlear volume. The μCT scans and 3D models of S2 and S4 are stored in the 239 Senckenberg institutional digital repository and are accessible for the scientific community 240 upon request. 241 242 2.4. Data collection and statistical analyses 243 With the exception Neanderthals and Sima de los Huesos individuals, whose 244 measurements were taken from the literature (Conde-Valverde et al., 2019: SOM Table S1), 245 all the included specimens were directly measured in Avizo using the same landmarks, 246 reference planes, and linear measurements defined by Conde-Valverde et al. (2019). For the 247 present analysis we relied on a subset of the variables described by Conde-Valverde et al. 248 (2019), including: cochlear length (CL, mm), length of the first turn (L1, mm), length of the 249 third turn (L3, mm), radius of the first turn (R1, mm), radius of the third turn (R3, mm), 250 cochlear volume (Vol, mm3), angle of the third turn (A3), number of turns (NT; defined as the 251 number of the complete turns + A3/360°), width (Sw, mm), height (Sh, mm), and area (ABT, 252 mm2) of the first turn cross section, and cochlear thickness (CTh, mm). We also computed 253 the turn proportions (%L1, %L2, and %L3) defined as the length of each turn divided by total 254 cochlear length. All the variables were measured on the 3D model generated using the HAT. 255 Conversely, to ensure direct comparisons with previously published results for primates and 256 hominins (Kirk and Gosselin-Ildari, 2009; Braga et al., 2015; Conde-Valverde et al., 2019), 257 cochlear volume was defined as the average between the volumes obtained using the LT 258 and HAT during segmentation. The ABT was measured in ImageJ (Schneider et al., 2012) 259 on an image of the cross-section taken from a view perpendicular to the cross-sectional 260 plane. For each of the aforementioned variables (except A3, which is used to compute NT), 261 we relied on box-and-whisker plots to assess their variability within the various groups 262 defined in the comparative sample, as well as to compare them with individual fossil 263 specimens. 264 To investigate the proportions of the cross section at the first cochlear turn, we inspected 265 the correlation between Sh and Sw by means of an ordinary least squares (OLS) allometric 266 regression (using decimal logarithms) performed in R Studio v. 1.1.453 for R v. 3.5.0 (R 267 Core Team, 2019), using the ‘stats’ package. We also relied on a box-and-whisker plot to 268 assess the variation in the Sw/Sh ratio. To test for significant differences among extant 269 groups, we used an analysis of variance (ANOVA) with Bonferroni-corrected Mann-Whitney 270 post hoc pairwise comparisons performed in R. Furthermore, we computed z-scores (z = 271 [measurement – mean] / SD) for individual fossil specimens to test whether they significantly 272 differ (z > |1.96|) from the a priori defined groups of the reference sample 273 We converted the metrical variables into dimensionless shape indices (denoted by an 274 asterisk following variable abbreviations) by dividing the former by the geometric mean of all 275 the variables in each individual (Mosimann, 1970; Jungers et al., 1995). We then performed 276 a principal components analysis (PCA) on ten shape variables (CL*, L1*, L3*, Vol1/3*, Sh*, 277 Sw*, R1*, R3*, CTh*, and NT*; SOM Table S2) to inspect patterns of shape variation. The 278 PCA was performed in R using the ‘ade4’ v. 1.7-16 package (Dray and Dufour, 2007). We 279 also performed a canonical variate analysis (CVA) on the same shape variables included in 280 the PCA, using the ‘Morpho’ v. 2.6 package (Schlager, 2017) for R, to assess the closest 281 morphometric affinities of individual fossil specimens by classifying them into various groups 282 defined a priori within the comparative sample. We used modern humans, Neanderthals, 283 australopiths (Australopithecus + Paranthropus), and Sima de los Huesos humans as a 284 priori defined groups, assuming equally distributed prior probabilities. The H. erectus S2 and 285 S4, the early Homo SK 27 and SK 847, and the Middle Pleistocene Aroeira 3 specimens 286 were projected a posteriori onto the morphospace identified by the CVA to classify them 287 based on the Mahalanobis squared distances between their projected scores and those of 288 group centroids. For these fossil specimens, we computed cross-validated posterior 289 probabilities as well as typicality probabilities (Albrecht, 1992) of group membership; the 290 latter were obtained using the ‘typprobClass’ fuction of the ‘Morpho’ package. Posterior and 291 typicality probabilities rely on different assumptions and are used in a complementary 292 manner. The former assumes that the analyzed individuals must belong to one of the a priori 293 defined group and thus add up to 100% for all the groups together. Conversely, the typicality 294 probabilities allow to evaluate the similarity of an individual to each group separately, based 295 on the distribution of the scores in each group. The classification of individual specimens into 296 one of the a priori defined groups relies on the highest probability in both cases, but the 297 typicality probabilities further represent the p-value to test the null hypothesis of membership 298 for each group. In other words, the highest posterior probability denotes the highest 299 morphometric affinities, but a significant typicality probability (p < 0.05) might still denote that 300 the specimen does not fit with the variation of the group with the highest posterior probability. 301 302 2.5. Estimation of hearing capabilities 303 To assess hearing capabilities in fossil hominins, we relied on the correlation between the 304 log-transformed cochlear volume and both low and high hearing frequency limits for primates 305 as a whole (Kirk and Gosselin-Ildari, 2009). Conde-Valverde et al. (2019) highlighted some 306 discrepancies between the estimates—based on the regression obtained for the 307 aforementioned correlation —and the empirical values reported by audiogram data for extant 308 humans (Heffner, 2004), thus indicating that the estimated values must be interpreted with 309 caution. Particularly, the empirical values obtained from audiograms of modern humans 310 (Heffner, 2004) demonstrate that this correlation is especially accurate in the case of the low 311 frequency limit (LFL), while it overestimates the high frequency limit (HFL; Conde-Valverde 312 et al., 2019). Hence, besides directly employing the regressions published by Kirk and 313 Gosselin-Ildari (2009) to estimate the hearing frequency limits, we also computed an 314 adjusted high frequency limit value (AHFL). This adjustment was made by multiplying the 315 HFL estimates by a scaling factor corresponding to the quotient between the empirical mean 316 value known for extant humans (17.6 kHz; Heffner, 2004) and the mean estimated for extant 317 humans in the present study. Even if modern humans represent the best option available, as 318 they are the closest extant species to the analyzed fossil taxa, caution should be used when 319 interpreting the adjusted estimates, as the scaling factor is being used across taxa. The 320 variation in estimated LFL and AHFL for the groups included in the comparative sample was 321 inspected by means of box-and-whisker plots. Finally, we computed the z-scores for 322 individual fossil specimens to test whether they significantly differ from the variation of AHFL 323 and LFL displayed by the aforementioned groups. 324 325 3. Results 326 3.1. Description 327 The right bony labyrinths of S2 and S4 are completely preserved, even if the cochlea of 328 S4 is almost completely filled with sediment, and thus allowed us to readily extract the 3D 329 models of the cochlea (Fig. 1). While both specimens resemble each other in some features, 330 such as the low CTh and intermediate NT, they also display some differences, particularly in 331 the shape of the first turn cross section (slightly elliptical in S2, round in S4) and in its ABT 332 (larger in S4), the volumetric proportions (smaller Vol in S2 despite a similar CL), and the 333 size of the R1 relative to the R3 (larger in S4; Table 1). The cochlea of S2 shows a narrow 334 CTh (Fig. 1a). It is long and moderately coiled (NT = 2.54), with a not very extended L1 as 335 compared with the somewhat longer L3 relative to L1 and CL. The Vol is moderately large 336 and the first turn cross section has a small area, and appears slightly elliptical (Sw/Sh = 337 0.86; Fig. 2). The R1 is not very long, being slightly more than twice the size of the R3 338 (R1/R3 = 2.09). In turn, S4 shows an even narrower cochlea (very low CTh; Fig. 1b) than 339 S2. The CL is long and moderately coiled, quite similar to that of S2. Conversely, the Vol is 340 large and the proportion between L1 and L3 is much higher. The cross section at the first 341 turn is much larger than in S2 and almost perfectly round (Sw/Sh = 0.98; Fig. 2) and the R1 342 is much longer than the R3 (R1/R3 = 2.75). 343 344 3.2. Comparisons 345 When compared with other fossil hominins and extant humans, the cochlear morphology 346 of S2 appears more primitive (australopith-like) than that of S4 (Fig 3; Table 1). Particularly, 347 the CTh of S2 is very narrow, falling outside the ranges of modern human, Sima de los 348 Huesos and Neanderthals, and being even narrower than in most australopiths (Fig. 3h). 349 The small R1 and L1 of S2 do not fit either with the ranges of variation of modern humans 350 and Neanderthals, and rather resemble those of australopiths and Sima de los Huesos (Fig. 351 3d,f). When the relative length of the turns is considered, S2 falls instead within the range of 352 both modern humans and australopiths (Fig. 4). The relatively small ABT falls within the 353 range of all groups except modern humans (Fig. 3k), being farther than one SD from the 354 mean of Neanderthals and modern humans only (Table 1). The medium-sized Vol is 355 intermediate between the ranges of humans and australopiths, falling in the overlapping 356 zone between them (below the range of modern humans) and within one SD from the means 357 of all groups except modern humans (Fig 3b). Conversely, the cochlea of S2 shows 358 similarities with modern humans, Neanderthals and Sima de los Huesos humans in the 359 possession of a long CL, which is longer than in most australopiths (Fig. 3c). 360 The combination of features found in S4 is unlike that of the other analyzed specimens 361 (Figs. 2 and 3; Table 1). The CTh is very narrow, as in S2 and most australopiths. At the 362 same time, S4 shows closer similarities than S2 with modern humans and fossils attributed 363 to the genus Homo (Fig. 1). The CL and L1 are long (longer than in australopiths), similar to 364 those of Neanderthals and modern humans (Fig. 3c, d). Notably, the turn proportions of S4 365 appear more similar to those of Neanderthals and of Sima de los Huesos humans, while 366 modern humans show a lower %L1 and higher %L3 (Fig. 4). The R1 is long as well, only 367 falling within the range of variation of modern humans (Fig. 3f), and the R1/R3 ratio is larger 368 (R1/R3 = 2.75) than in most of the analyzed specimens. The Vol of S4 is very large, and 369 overlaps only with the range of modern humans, falling at less than one SD from the modern 370 human mean (Fig. 3b; Table 1). The ABT is also large and falls at the high end of the 371 modern human and Neanderthal range (Fig. 3k). Altogether, the values displayed by S4 for 372 Vol, L1, ABT, and R1 resemble those of Homo and fall well outside the range of 373 australopiths (Fig. 3b; Table 1). Finally, S4 possesses a round cross section of the first turn, 374 more so than in S2 and in any other analyzed specimen (Fig. 2). Due to the large overlap 375 between the different hominin groups analyzed, the remaining variables (L3 and NT) are 376 slightly less informative, yet they rule out close similarities between Indonesian H. erectus 377 and Neanderthals (Fig. 3a, e). 378 Given that S2 and, especially, S4 differ from extant and fossil humans by the rounder first 379 turn cross section, we investigated the correlation between Sh and Sw by means of an 380 allometric regression (Fig. 2a). Data used for computing the regressions are given in SOM 381 Table S2. The results indicate a significant (p < 0. 001) isometric correlation between the 382 variables (slope = 0.877 ± 0.251 95% CI; intercept = 0.360 ± 0.147 95% CI; Fig. 2a), albeit 383 with some dispersion (r2 = 0.62), with most australopiths falling below and humans above the 384 best-fit line. The two Sangiran specimens both fall below the regression line, like the early 385 Homo SK 27 and the Aroeira 3 specimens. Conversely, SK 847 falls slightly above the 386 regression line, like later and extant humans. Differences in the cross-sectional shape of the 387 first turn are even clearer based on the Sw/Sh ratio. The pairwise comparisons for this ratio 388 denote significant differences between australopiths and all other human groups (p < 0.001), 389 while the latter do not significantly differ among each other. In addition, SK847 more closely 390 resembles humans, yet it does not significantly differ from any of the groups based on the z- 391 scores, as neither does SK27, falling in the overlap zone between the upper range of 392 humans and australopiths (Table 2). Similarly, Aroeira 3 occupies an intermediate position 393 between these groups, although it significantly differs from Neanderthals (Table 2). 394 Regarding the studied specimens, S2 most closely resembles australopiths (and does not 395 overlap with humans; Table 2), while S4 possesses a much rounder cross section than any 396 other analyzed specimen (Fig. 2b; SOM Table S2) and falls outside the range of any group 397 (Table 2). 398 399 3.3. Multivariate shape analyses 400 Principal components analysis The PCA performed on cochlear shape variables shows a 401 considerable amount of overlap between the analyzed hominin groups and only enables a 402 clear discrimination between australopiths and Neanderthals, as well as between modern 403 humans and Sima de los Huesos (Fig. 5). The shape variation captured by the first principal 404 component (PC1, 48.8% of variance) is mainly driven by differences in R1*, L1*, Vol1/3*, CL*, 405 and Sw*, toward negative scores, as well as by L3*, with a positive loading (Fig. 5; Table 3). 406 Hence, the huge amount of overlap along this axis does not allow to discern clearly between 407 the groups, with only a few Neanderthals specimens (most negative values)that could be 408 pulled apart from rest of the comparative sample (intermediate and positive values, except 409 for Paranthropus). When the two Indonesian H. erectus are considered, S4 falls on 410 moderately negative scores for PC1, outside the range of Australopithecus and at the lower 411 range of modern humans and Paranthropus, due to its larger Vol* and L1*, as well as the 412 larger radius of the first turn, which mainly resemble the condition found in some 413 Neanderthal individuals. Conversely, S2 fits within the range of variation of all the groups 414 defined in the comparative sample along PC1, as it shows a long CL* coupled with 415 moderately large Vol* and R1*, together with a long L3*. Among the fossil comparative 416 sample, early Homo SK 847 shows a morphology that is intermediate between those of S2 417 and S4, while neither the other early Homo (SK 27) nor the Aroeira 3 specimens display 418 close morphometric affinities with the studied specimens (Fig. 5). 419 The PC2 (16.8% of variance) is driven by differences in few variables—Sh*, CTh* toward 420 positive and NT* toward negative scores (Table 3). Like for PC1, the overlap between the 421 various groups along PC2 is extensive (Fig. 5). Nonetheless, this axis separates most 422 australopith specimens (moderate to very negative scores) from humans (moderate to very 423 positive scores), due to the broader CTh*, more elliptical cochlear cross section (low Sw/Sh; 424 Fig. 2b) and less coiled cochlea (lower NT*) in the latter. Along PC2, S2 falls clearly outside 425 that of humans (both extant and fossil) and shows closest morphometric affinities with 426 australopiths, especially due to their very narrow CTh*. For the same reason, also S4 and 427 SK 27 fall at the lowest end of the human range on somewhat less negative scores than S2. 428 The other early Homo specimen, SK 847, together with the Aroeira 3 individual are found on 429 intermediate PC2 scores, more clearly overlapping with fossil and extant humans due to 430 their larger CTh* (particularly in SK 847) and smaller Sw/Sh ratio (Fig. 2b). 431 When PC1 and PC2 are considered simultaneously, the similarities between S2 and 432 australopiths appear even clearer as the former falls well within the range of the latter and 433 outside the variation of Homo. Conversely, S4 does not fit well with the variability of either 434 Homo and australopiths, falling outside their convex hulls, albeit very close to Sima de los 435 Huesos and, to a slightly lesser extent, both Neanderthals and australopiths. Similarly, 436 despite some closer affinities with later occurring humans, Aroeira 3 falls outside the convex 437 hulls of modern humans, Sima de los Huesos, and Neanderthal variation, and closer to that 438 of australopiths. In turn, the two early Homo individuals also fall apart from one another, with 439 SK 27 markedly departing from the range of variation of the analyzed groups and SK 847 440 falling in the overlap zone between Neanderthals and Sima de los Huesos, not far from 441 modern humans. 442 Canonical variates analysis The CVA performed on the same ten shape variables used in 443 the PCA correctly classified 97.5% of the specimens (82.5% after cross-validation; Table 4). 444 After cross-validation, all modern humans were correctly classified, while Sima de los 445 Huesos humans show the highest number of misclassified individuals (30%), with two 446 individuals classified as Neanderthals (AT-421 and Cr.13) and one as modern human (AT- 447 1907). In the case of Neanderthals, two specimens were classified as Sima de los Huesos 448 humans (La Ferrassie 8 and Kebara 1), resulting in 71% of correctly classified specimens, a 449 result very close to the one obtained for the Sima de los Huesos humans. Similarly, two 450 australopith specimens (84% correct classification) were misclassified as modern humans 451 (MLD 37/38 and StW 573). 452 The CVA discriminates well among the four a priori defined groups (Fig. 6), with only 453 minimal overlap between Sima de los Huesos humans and Neanderthals, despite the fact 454 that the three out of four groups belong to genus Homo. The first canonical variate (CV1, 455 64% of variance) discriminates Neanderthals and Sima de los Huesos humans from 456 australopiths and modern humans, and is driven mostly by Vol1/3* and R1* and, to a lesser 457 extent, Sw*, all toward negative scores. The Sh* is the only variable possessing a 458 moderately positive effect on the CV1 (Table 5). The CV2 (31% of variance) pulls apart 459 modern humans from australopiths and Sima de los Huesos humans, while Neanderthals 460 occupy an intermediate position. The Vol1/3* is the main factor positively influencing CV2 461 scores, followed by Sh*, L3*, and R1* with similar lower loadings. In contrast, Sw* and NT* 462 drive CV2 toward negative scores with a similar loading (Table 5). Hence, in combination, 463 CV1 and CV2 discriminate well modern humans (with negative CV1 and positive CV2 464 scores) from other groups due to the larger Vol1/3*, longer R1*, more elliptical cross section 465 (low Sw/Sh ratio; Fig. 2b), and moderately higher NT* of the former. Australopiths display 466 negative CV1 and CV2 scores due to the high NT*, moderate Vol1/3*, long R1* and L3*, and 467 similarly-sized Sw* and Sh*. Both Sima de los Huesos humans and Neanderthals differ from 468 australopiths and modern humans due to a shorter L3* and lower NT*, combined also with a 469 slightly longer CL*, thus displaying positive scores along CV1. At the same time, 470 Neanderthals differ from Sima de los Huesos humans by the larger Vol1/3* and broader 471 CTh*, thus being somewhat closer to modern humans along CV2 (the moderate positive 472 scores are due to the low NT* in Neanderthals), while Sima de los Huesos humans appear 473 similar to the australopiths condition, characterized by a moderately small Vol1/3*, narrow 474 CTh*, and fairly round cochlear cross section (hence falling on negative CV2 scores). 475 Sangiran 2 falls on moderate negative scores for both CV1 and CV2, at the edge of the 476 convex hull of australopiths and close to the Paranthropus specimens. This is due to its 477 moderate Vol1/3*, coupled with a low CTh*, fairly long L3*, and slightly elliptical cochlear 478 cross section. Sangiran 4 falls outside the variation of the comparative groups, close to the 479 hull of australopiths on negative CV2 and, particularly, CV1 scores (Fig. 6). Indeed, while the 480 Vol1/3* and R1* of S4 are larger than in most australopiths, the Sw/Sh ≈ 1 and narrow CTh* 481 are most similar to those of australopiths rather than Homo species. Both SK 847 and 482 Aroeira 3 display moderate negative values for both CV1 and CV2, falling outside the 483 variation of the inspected groups—with SK 847 being almost equidistant from modern 484 humans, Sima de los Huesos humans and australopiths—while Aroeira 3 approaches the 485 convex hull of Sima de los Huesos humans. Finally, SK 27 falls within the convex hull of 486 Sima de los Huesos humans, due to the low NT* combined with a small Vol1/3*. 487 The posterior probabilities of group membership (Table 6) classify the Sangiran 488 specimens as australopiths with very high probabilities even after cross-validation (99.5% 489 and 96.3% for S2 and S4, respectively), thereby denoting closer morphometric affinities in 490 cochlear shape with australopiths. However, while the typicality probabilities of group 491 membership (Table 6) further indicate that S2 significantly differs from the three human 492 groups but not australopiths (and thus confirm its overall australopith-like morphology), S4 493 represents an outlier for all the groups, suggesting that its cochlear shape does not fit well 494 with the variability displayed by any group. Among the fossil comparative sample (Table 6), 495 SK 847 is classified as an australopith with a high posterior probability (95.5%), as further 496 supported by the typicality probabilities, which do not recover it as an outlier for this group. In 497 turn, Aroeira 3 shows closest affinities with Sima de los Huesos and is classified as such 498 with a high posterior probability (82.8%), while its typicality probabilities reject the null 499 hypothesis of group membership for all the other groups (Table 6). The early Homo SK 27 500 specimen is the only fossil individual classified as a Neanderthal with a high posterior 501 probability (82.8%). However, as for S4, the typicality probabilities demonstrates that it falls 502 outside the range of variation of any group (Table 6). 503 504 3.3. Estimates of auditory capabilities 505 The LFL and HFL estimates obtained for the studied and comparative samples are 506 reported in Figure 7 and Table 7. The estimated hearing range for modern humans ranges 507 from 24.0 Hz to 20.6 kHz. Our HFL estimates (mean = 20.79 kHz) are on average 15.3% 508 higher than the empirical mean derived for modern humans (17.6 kHz; Heffner, 2004), 509 thereby suggesting that the regression might be overestimating HFL and strengthening the 510 view that HFL estimates should be adjusted. 511 The results show that S2 matches the range of variation of all the investigated hominin 512 groups except that of modern humans (showing lower values for both AHFL and LFL; Fig. 7; 513 Tables 7 and 8). Nonetheless, the AHFL estimate for S2 (19.85 kHz) is compatible with the 514 nominal range for extant humans, and most similar to the condition estimated for the early 515 Homo specimen SK 847 (Fig. 7; Table 7). Sangiran 4, on the other hand, is most similar to 516 modern humans, falling into their distribution for the AHFL and LFL (Fig. 7; Table 7), while it 517 only significantly differs from Sima de los Huesos humans (Table 8). Among the analyzed 518 specimens from the comparative fossil sample, Aroeira 3 is the only one that matches the 519 hearing range of modern humans, while both SK 847 and SK 27 significantly differ from the 520 latter (and SK 27 also from Sima de los Huesos humans and Neanderthals; Fig. 7; Table 8). 521 522 4. Discussion 523 4.1. Cochlear morphology in human evolution 524 Although our analysis is focused on the cochlear morphology of H. erectus from 525 Sangiran, the inspection of the modern human sample adds information about the 526 intraspecific variability of this species. In particular, our results indicate a higher maximum 527 volume than previous analyses, whereas the minimum value is similar (Beals et al., 2016; 528 Conde-Valverde et al., 2019). Given that all studies have relied on the same segmentation 529 protocol, this result is most likely related to individual variation and suggests a higher 530 intraspecific variability than previously reported for modern humans. Furthermore, the 531 inclusion of other extinct hominins in the fossil comparative sample further allows us to 532 provide insight on the evolution of this anatomical structure from a broader perspective— 533 which is essential for correctly inferring the polarity of change and hence adequately 534 interpreting the cochlear morphology of the investigated specimens. Our results first reveal 535 important cochlear characters of australopiths and South African early Homo (Table 9) that 536 remained unnoticed due to the approaches used in previous analyses (Braga et al., 2015; 537 Beaudet, 2019; Beaudet et al., 2019), which did not capture differences in cochlear volume, 538 cross section, and thickness. In particular, our results show that australopiths differ from 539 Middle/Late Pleistocene and extant humans by a rounder cochlear cross section (Sw/Sh 540 ratio closer to 1), shorter L1, and smaller Vol. All three characters were previously identified 541 as hominine symplesiomorphies based on the chimpanzee condition (Conde-Valverde et al., 542 2019), but our analyses reveal that the plesiomorphic condition for these features was still 543 retained in early hominins (australopiths). In contrast, the two analyzed early Homo 544 specimens only appear plesiomorphic in the small Vol and short L1, while more closely 545 resembling later humans (especially SK 847) in the possession of an oval cross section. Our 546 results further indicate that both Neanderthals and modern humans are derived in the long 547 R1, although the comparatively shorter R1 of Sima de los Huesos suggests that such a 548 similarity might be homoplastic feature, as in the case of several other cochlear features 549 (Conde-Valverde et al., 2019)—maybe owing to similar architectural constraints imposed by 550 the course of the facial nerve, which is similar in both species (Spoor et al., 2003) and known 551 to influence the expansion of the cochlear coiling within the petrosal in modern humans 552 (Pietsch et al., 2017). 553 Thanks to the inclusion of a broader comparative sample than in previous studies, our 554 results further imply that some other previous inferences must be reconsidered. Conde- 555 Valverde et al. (2019) suggested that the possession of an oval cross section might be 556 homoplastic between Neanderthals and modern humans, given the retention of a 557 plesiomorphic round cross section in the Sima de los Huesos humans (considered to 558 represent an early stage of the Neanderthal lineage; Arsuaga et al., 2014; Meyer et al., 559 2016). However, an inspection of the australopith morphology indicates instead that the 560 Sima de los Huesos humans (sharing a similar Sw length with the australopiths, yet 561 possessing a longer Sh) should be regarded as possessing an oval morphology, similar to 562 that of later occurring humans, thereby implying that this feature is likely synapomorphic 563 between Neanderthals and modern humans, at least to some extent. Indeed, the similar 564 morphology displayed by the early Homo SK 847, coupled with the intermediate condition 565 found in Aroeira 3 and SK 27, further supports that a somewhat oval cross section might 566 have emerged earlier than previously thought along hominin evolution. Similarly, while it has 567 proposed that both Neanderthals and modern humans are derived in the thicker CTh 568 (Conde-Valverde et al., 2018), the comparison with australopiths indicates that Neanderthals 569 should be rather regarded as less derived than modern humans. Finally, and contrary to 570 previous suggestions, our results for L3, %L3, and NT show that these features vary 571 considerably among hominins. Particularly, Paranthropus and early Homo display very low 572 values for NT (Braga et al., 2015; Beaudet et al., 2019), L3, and %L3, showing closest 573 similarities with Neanderthals, which are considered autapomorphic for these characters 574 (Conde-Valverde et al., 2019). Accordingly, homoplasy might have also played a role in the 575 evolution of these features. Overall, our results confirm that the ancestral condition for the 576 genus Homo most likely included a short CL (both relative to body mass and in absolute 577 value) and L1, a small Vol and R1, low CTh, and a round cross section, while similarities 578 between Neanderthals and extant humans might be homoplastic to a large extent. 579 580 4.2. Evolutionary and paleobiological implications of Sangiran 2 and 4 cochlear morphology 581 Our results demonstrate that the two analyzed H. erectus specimens, S2 and S4, are 582 quite different from one another, regardless of some similarities in the long CL—a 583 synapomorphy of humans except its earliest representatives—as well as in the narrow CTh 584 and round cochlear cross section—which are hominine plesiomorphic characters retained by 585 australopiths and some early Homo individuals. Such dissimilarities between S2 and S4 may 586 have different explanations. It is generally assumed that S2 and S4 are roughly 587 contemporaneous (Watanabe and Kadar, 1985; Brasseur et al., 2015), but given some 588 uncertainties about the chronostratigraphical provenance of S2 and S4 and the collection 589 methods employed by von Koenigswald’s local collaborators (e.g., Tyler and Sartono, 2001), 590 the analyzed Sangiran specimens might come from populations more distant in time from 591 one another than previously assumed. Under the latter assumption, the more derived 592 cochlear morphology of S4 would be suggestive of a younger (late Early/Middle Pleistocene) 593 age relative to S2. However, this is at odds with the clear similarities between S2 and the 594 Trinil 2 specimen (Schwartz and Tattersall, 2003; Schwartz, 2016), dated to the Middle 595 Pleistocene (Berghuis et al., 2021). 596 Irrespective of age uncertainties, it remains to be ascertained whether the differences 597 between S2 and S4 reflect intraspecific variation (as generally assumed) or are indicative of 598 taxonomic differences at the genus and/or species rank—as formerly concluded by previous 599 authors that assigned S4 to a different species (Weidenreich, 1945) or even suggested 600 pongine affinities (Krantz, 1975, 1994; Tyler, 2003, 2004, 2006). The cochlear morphology of 601 S4, together with its palate and dental morphology (Durband, 2008), allow us to rule out the 602 latter claim. Indeed, despite the plesiomorphic cross section of the first turn (falling close to 603 the high end of the chimpanzee distribution; Conde-Valverde et al., 2019), the larger Vol and 604 R1 and longer L1 of S4 suggest closer affinities with Neanderthals and modern humans than 605 in the case of S2. Given the small sample available, the differences between S2 and S4 606 might simply reflect opposite ends of the intraspecific range of cochlear morphology variation 607 of H. erectus. Despite considerable differences in robusticity between S2 and S4 (Santa 608 Luca, 1980), their conspecific status is supported by their similarities in external cranial 609 morphology to the H. erectus holotype (Koenigswald, 1940; Schwartz and Tattersal, 2003; 610 Schwartz, 2016), as well as by the overall marked variation within the H. erectus hypodigm 611 as currently conceived (Antón, 2003; Rightmire, 1990; Thorne and Wolpoff, 1982; Wolpoff, 612 1996; Wolpoff et al., 2001). Under such an assumption, it may be inferred that Indonesian H. 613 erectus would have probably displayed, on average, an intermediate cochlear morphology 614 between S2 and S4, and thus more derived than in australopiths and early Homo, but less 615 so than in later humans. Unfortunately, the scanty information on the cochlear morphology of 616 other Early and Middle Pleistocene H. erectus (Wu et al., 2014; Wu and Zhang, 2016) does 617 not favor either interpretation. Additional inner ear material of H. erectus s.l. from both Africa 618 and Asia should be inspected to test between these competing hypotheses in the future. 619 From a paleobiological viewpoint—and irrespective of the taxonomic and evolutionary 620 interpretation of differences in cochlear morphology between S2 and S4—the large Vol 621 found in S4 (and the hearing range estimated from it) closely resembles that of Neanderthals 622 and modern humans. Even if potentially variable, this is suggestive of a lowered high- 623 frequency cutoff, a possible homoplastic character for both taxa (Conde-Valverde et al., 624 2019). This result is broadly compatible with an increased sensitivity to low frequencies, 625 which is a modern human autapomorphy relative to other primates (Heffner, 2004). Hence, 626 the S4 individual, and perhaps H. erectus more generally, currently represents the earliest 627 occurrence of a modern human-like hearing range, although this by itself cannot be used as 628 direct evidence suggestive of advanced speech capabilities. 629 Indeed, H. erectus s.l. is the first hominin species to clearly display a suite of human-like 630 adaptations, including not only a modern brain organization despite smaller cranial capacity 631 (Holloway et al., 2009; Ponce de León et al., 2021), but further encompassing a very 632 committed terrestrial bipedalism (Harcourt-Smith, 2015), a more advanced (albeit not 633 completely modern human-like) pattern of dental development and inferred life-history profile 634 (Kelley and Bolter, 2013), and especially a diet including significant amounts of meat (Milton, 635 1999; Ben-Dor et al., 2021) and systematic stone tool-making abilities (Joordens et al., 636 2015), both associated with the so-called human predatory pattern (Thompson et al., 2019). 637 All these characteristics suggests that H. erectus s.l. represented the emergence of a new 638 hominin adaptive type (Wood and Collard, 1999), not fundamentally different from that 639 further elaborated by later humans. There is strong evidence in favor of the coevolution 640 between behavioral complexity and enhanced vocal communication abilities (Conde- 641 Valverde et al., 2021). Therefore, the possession—as hinted by the estimates obtained here 642 for S4—of a more modern-human-like hearing range in H. erectus than in any of its 643 predecessors may be hypothesized and should be subjected to more detailed investigation 644 in the future. 645 646 5. Conclusions 647 Our analyses indicate that cochlear morphology adequately distinguishes australopiths 648 from humans and further discriminates different human groups despite the presence of 649 homoplasies between modern humans and Neanderthals. Australopiths display overall a 650 plesiomorphic condition that confirms the ancestral morphotype previously inferred for 651 hominins—characterized by a short cochlear length (both in absolute and relative terms), 652 small volume, round cross-sectional shape, and short first cochlear turn. In turn, early Homo 653 specimens appear slightly more derived toward the modern human condition but less so 654 than later fossil humans. The inclusion of australopiths in the comparative sample allowed us 655 to refine the polarity of some characters, as well as to identify additional derived features for 656 humans. Based on these considerations, the studied specimens of Indonesian H. erectus 657 from Sangiran display a mosaic of plesiomorphic (australopith-like) and derived features, 658 consistent with their more basal status than other humans analyzed except for early Homo. 659 Both Sangiran specimens are largely plesiomorphic in cochlear thickness and cross- 660 sectional shape. Nevertheless, they also differ from one another in some other features. 661 Namely, S4 appears more derived toward later humans in its larger volume and longer and 662 larger first cochlear turn, while at the same time it retains an extremely round, chimpanzee- 663 like cross section. It cannot be completely ruled out that such differences result from S2 and 664 S4 belonging to different species and/or populations more distant in time than customarily 665 assumed. However, based on similarities in cranial morphology—between them and relative 666 to other Indonesian specimens—the most plausible interpretation is that the differences 667 between S2 and S4 just represent opposite ends of the range of variation displayed by H. 668 erectus in cochlear morphology. The inspection of additional specimens of H. erectus s.l. 669 from both Africa and Asia will hopefully shed further light on this matter in the future, with 670 implications for understanding the entangled evolutionary history of hominins in southeastern 671 Asia and elsewhere. 672 In the meantime, the remarkable similarities between the hearing range estimated for S4 673 and those of Neanderthals and modern humans deserve further attention. In agreement with 674 previous inferences for Neanderthals, this suggests that H. erectus might have been more 675 modern-human-like than its hominin forerunners (including early Homo representatives) in 676 terms of hearing capabilities—consistent with other morphological and behavioral 677 innovations indicating that H. erectus s.l. was the first hominin to clearly display a human-like 678 adaptive strategy. 679 680 Acknowledgments 681 This work has been funded by the Agencia Estatal de Investigación (PID2020-117289GB- 682 I00 and PID2020-116908GB-I00, AEI/FEDER, EU) and the Generalitat de Catalunya 683 (CERCA Programme and Consolidated Research Group 2017 SGR SGR 116). We express 684 our gratitude to the Werner Reimers Foundation in Bad Homburg, which provides the Gustav 685 Heinrich Ralph von Koenigswald collection as a permanent loan for scientific research to the 686 Senckenberg Research Institute and Natural History Museum Frankfurt. We would like to 687 thank C. Zanolli for technical and scientific discussion, as well as the following people and 688 institutions for granting access to and/or performing the CT scans of the comparative 689 material: the University of the Witwatersrand, the Ditsong National Museum of Natural 690 History, the Max Plank Institute for Evolutionary Anthropology, Steinmann-Institute of 691 Geology, Mineralogy und Palaeontology at the University Bonn, M. Tawane, J. Braga, J. 692 Kudakwashe Jakata, and G. Chinamatira. The support of the AESOP+ program is hereby 693 acknowledged. We are also thankful to the DST-NRF for allowing access to the Micro-XCT 694 facility at Necsa. We thank the Guest Editors (C. Zanolli, S. Xing, and Y. Zhang) of the 695 special issue “Asian Pleistocene Hominids” for inviting us to contribute. 696 697 References 698 Albrecht, G., 1992. Assessing the affinities of fossils using canonical variates and 699 generalized distances. Hum. Evol. 7, 49–69. 700 Antón, S.C., 2003. Natural history of Homo erectus. Yearb. Phys. Anthropol. 46, 126–70. 701 Antón, S.C., Swisher, C.C. III, 2004. Early dispersals of Homo from Africa. Annu. Rev. 702 Anthropol. 33, 271–296. 703 Armstrong, S.D., Bloch, J.I., Houde, P., Silcox, M.T., 2011. Cochlear labyrinth volume in 704 euarchontoglirans: implications for the evolution of hearing in Primates. Anat. Rec. 294, 705 263–266. 706 Arsuaga, J.L., Martínez, I., Arnold, L., Aranburu, A., Gracia-Téllez, A., Sharp, W., Quam, R., 707 Falguères, C., Pantoja-Pérez, A., Bischoff, J., Poza-Rey, E., Parés, J., Carretero, J., 708 Demuro, M., Lorenzo, C., Sala, N., Martinon-Torres, M., García, N., Alcázar de Velasco, 709 A., Cuenca-Bescós, G., Gómez-Olivencia, A., Moreno, D., Pablos, A., Shen, C., 710 Rodríguez, L., Ortega, A., García, R., Bonmatí, A., Bermúdez de Castro, J.M., Carbonell, 711 E., 2014. Neandertal roots: Cranial and chronological evidence from Sima de los Huesos. 712 Science 344, 1358–1363. 713 Beals, M.E., Frayer, D.W., Radovčić, J., Hill, C.A., 2016. Cochlear labyrinth volume in 714 Krapina Neandertals. J. Hum. Evol. 90, 176–182. 715 Beaudet, A., 2019. The inner ear of the Paranthropus specimen DNH 22 from Drimolen, 716 South Africa. Am. J. Phys. Anthropol. 170, 439–446. 717 Beaudet, A., Clarke, R. J., Bruxelles, L., Carlson, K. J., Crompton, R., de Beer, F., Dhaene, 718 J., Heaton, J. L., Jakata, K., Jashashvili, T., Kuman, K., McClymont, J., Pickering, T. R., 719 Stratford, D., 2019. The bony labyrinth of StW 573 (“Little Foot”): Implications for early 720 hominin evolution and paleobiology. J. Hum. Evol. 127, 67–80. 721 Ben‐Dor, M., Sirtoli, R., Barkai, R., 2021. The evolution of the human trophic level during the 722 Pleistocene. Yearb. Phys. Anthropol. 175, 27–56. 723 Berghuis, H.W.K., Veldkamp, A., Adhityatama, S., Hilgen, S.L., Sutisna, I., Barianto, D.H., 724 Pop, E.A., Reimann, T., Yurnaldi, D., Ekowati, D.R., Vonhof, H.B., van Kolfschoten, T., 725 Simanjuntak, T., Schoorl, J.M., Joordens, J.C.A., 2021. Hominin homelands of East Java: 726 Revised stratigraphy and landscape reconstructions for Plio-Pleistocene Trinil. Quat. Sci. 727 Rev. 260, 106912. 728 Braga, J., Thackeray, J.F., Dumoncel, J., Descouens, D., Bruxelles, L., Loubes, J.M., Kahn, 729 J.L., Stampanoni, M., Bam, L., Hoffman, J., de Beer, F., 2013. A new partial temporal 730 bone of a juvenile hominin from the site of Kromdraai B (South Africa). J. Hum. Evol. 65, 731 447–456. 732 Braga, J., Loubes, J.-M., Descouens, D., Dumoncel, J., Thackeray, J.F., Kahn, J.-L., de 733 Beer, F., Riberon, A., Hoffman, K., Balaresque, P., Gilissen, E., 2015. Disproportionate 734 cochlear length in genus Homo shows a high phylogenetic signal during apes' hearing 735 evolution. PLoS One 10, e0127780. 736 Braga, J., Samir, C., Fradi, A., Feunteun, Y., Jakata, K., Zimmer, V.A., Zipfel, B., Thackeray, 737 J.F., Macé, M., Wood, B.A., Grine, F.E., 2021. Cochlear shape distinguishes southern 738 African early hominin taxa with unique auditory ecologies. Sci. Rep. 11, 17018. 739 Brasseur, B., Sémah, F., Sémah, A.-M., Djubiantono, T., 2015. Pedo-sedimentary dynamics 740 of the Sangiran dome hominid bearing layers (Early to Middle Pleistocene, central Java, 741 Indonesia): A palaeopedological approach for reconstructing ‘Pithecanthropus’ (Javanese 742 Homo erectus) palaeoenvironment. Quat. Int. 376, 84–100. 743 Conde-Valverde, M., Quam, R., Martínez, I., Arsuaga, J.L., Daura, J., Sanz, M., Zilhão, J., 744 2018. The bony labyrinth in the Aroeira 3 Middle Pleistocene cranium. J. Hum. Evol. 124, 745 105–116. 746 Conde-Valverde, M., Martínez, I., Quam, R., Bonmatí, A., Lorenzo, C., Velez, A.D., 747 Martínez-Calvo, C., Arsuaga, J.L., 2019. The cochlea of the Sima de los Huesos 748 hominins (Sierra de Atapuerca, Spain): New insights into cochlear evolution in the genus 749 Homo J. Hum. Evol. 136, 102641. 750 Conde-Valverde, M., Martínez, I., Quam, R., Arsuaga, J.L., Daura, J., Sanz, M., Zilhão, J., 751 2020. The cochlea of the Aroeira 3 Middle Pleistocene cranium—a comparative study. J. 752 Hum. Evol. 148, 102887. 753 Conde-Valverde, M., Martínez, I., Quam, R.M., Rosa, M., Velez, A.D., Lorenzo, C., Jarabo, 754 P., de Castro, J.M.B., Carbonell, E., Arsuaga, J.L., 2021. Neanderthals and Homo 755 sapiens had similar auditory and speech capacities. Nat. Ecol. Evol. 5, 609–615. 756 Crevecoeur, I., Brooks, A., Ribot, I., Cornelissen, E., Semal, P., 2016. Late Stone Age 757 human remains from Ishango (Democratic Republic of Congo): New insights on Late 758 Pleistocene modern human diversity in Africa. J. Hum. Evol. 96, 35–57. 759 Daura, J., Sanz, M., Arsuaga, J.L., Hoffmann, D.L., Quam, R.M., Ortega, M.C., Santos, E., 760 Gómez, S., Rubio, A., Villaescusa, L., Souto, P., Mauricio, J., Rodrigues, F., Ferreira, A., 761 Godinho, P., Trinkaus, E., Zilhão, J., 2017. New Middle Pleistocene hominin cranium from 762 Gruta da Aroeira (Portugal). Proc. Natl. Acad. Sci. USA 114, 3397–3402. 763 Day, M., 1986. Guide to Fossil Man, 4th ed. University of Chicago Press, Chicago. 764 Djubiantono, T., Sémah, F., 1991. Lower Pleistocene marine-continental transitional beds in 765 the Solo depression and their relation to the environment of the Pucangan hominids. Bull. 766 Indo-Pac. Prehist. Assoc. 11, 7–13. 767 Dray, S., Dufour, A., 2007. The ade4 package: Implementing the duality diagram for 768 ecologists. J. Stat. Softw. 22, 1–20. 769 Dubois, E., 1892. Palaeontologische andrezoekingen op Java. Versl. Mijnw. Batavia 3, 10– 770 14. 771 Durband, A.C., 2008. The enigma of the Sangiran 4 palate revisited. HOMO 59, 111–122. 772 Ferring, R., Oms, O., Agustí, J., Berna, F., Nioradze, M., Shelia, T., Tappen, M., Vekua, A., 773 Zhvania, D., Lordkipanidze, D., 2011. Earliest human occupations at Dmanisi (Georgian 774 Caucasus) dated to 1.85–1.78 Ma. Proc. Natl. Acad. Sci. USA 108, 10432–10436. 775 Franzen, J.L., 1985a. Asian australopithecines? In: Tobias, P.V. (Ed.), Hominid Evolution: 776 Past, Present, and Future. Wiley-Liss, New York, pp. 255–263. 777 Franzen, J.L., 1985b. What is ‘‘Pithecanthropus dubius Koenigswald, 1950’’? In: Delson, E. 778 (Ed.), Ancestors: The Hard Evidence. Alan R Liss, New York, pp. 221–226. 779 Frouin, M., Lahaye, C., Valladas, H., Higham, T., Debénath, A., Delagnes, A., Mercier, N., 780 2017. Dating the middle Paleolithic deposits of La Quina Amont (Charente, France) using 781 luminescence methods. J. Hum. Evol 109, 30–45. 782 Gibbon, R.J., Pickering, T.R., Sutton, M.B., Heaton, J.L., Kuman, K., Clarke, R.J., Brain, 783 C.K., Granger, D.E., 2014. Cosmogenic nuclide burial dating of homininbearing 784 Pleistocene cave deposits at Swartkrans, South Africa. Quat. Geochronol. 24, 10–15. 785 Granger, D.E., Gibbon, R.J., Kuman, K., Clarke, R.J., Bruxelles, L., Caffee, M.W., 2015. 786 New cosmogenic burial ages for Sterkfontein Member 2 Australopithecus and Member 5 787 Oldowan. Nature 522, 85–88. 788 Guérin, G., Frouin, M., Talamo, S., Aldeias, V., Bruxelles, L., Chiotti, L., Dibble, H.L., 789 Goldberg, P., Hublin, J.-J., Jain, M., 2015. A multi-method luminescence dating of the 790 Palaeolithic sequence of La Ferrassie based on new excavations adjacent to the La 791 Ferrassie 1 and 2 skeletons. J. Archaeol. Sci. 58, 147–166. 792 Harcourt-Smith, W.E.H., 2015. The origins of bipedal locomotion. In: Henke, W., Tattersall, I. 793 (Eds.), Handbook of Paleoanthropology, 2nd ed. Springer Verlag, Heidelberg, pp. 1919– 794 1959. 795 Heffner, R.S., 2004. Primate hearing from a mammalian perspective. Anat. Rec. 281, 796 1111e1122. 797 Holloway R.L., 1981. The indonesian Homo erectus brain endocasts revisited. Am. J. Phy. 798 Anthropol. 55, 503–521. 799 Holloway, R.L., Sherwood, C.C., Hof, P.R., Rilling, J.K., 2009. Evolution of the brain in 800 humans – Paleoneurology. In: The New Encyclopedia of Neuroscience: Springer, pp. 801 1326–1334. 802 Hyodo, M., 2001. The Sangiran geomagnetic excursion and its chronological contribution to 803 the Quaternary geology in Java. In: Simanjuntak, H.T., Prasetyo, B., Handini, R. (Eds.), 804 Sangiran: Man, Culture and Environment in Pleistocene Times. Yayasan Obor Indonesia, 805 Jakarta, pp. 320–335. 806 Jacob, T., 1973. Paleoanthropological discoveries in Indonesia with special reference to the 807 finds of the last two decades. J. Hum. Evol. 2, 473–485. 808 Jeffery, N., Spoor, F., 2004. Prenatal growth and development of the modern human 809 labyrinth. J. Anat. 204, 71–92. 810 Joordens, J.C.A., d’Errico, F., Wesselingh, F.P., Munro, S., de Vos, J., Wallinga, J., 811 Ankjærgaard, C., Reimann, T., Wijbrans, J.R., Kuiper, K.F., Mücher, H.J., Coqueugniot, 812 H., Prié, V., Joosten, I., van Os, B., Schulp, A.S., Panuel, M., van der Haas, V., 813 Lustenhouwer, W., Reijmer, J.J.G., Roebroeks, W., 2015. Homo erectus at Trinil on Java 814 used shells for tool production and engraving. Nature 518, 228–231. 815 Jungers, W.L., Falsetti, A.B., Wall, C.E., 1998. Shape, relative size, and size-adjustments in 816 morphometrics. Yearb. Phys. Anthropol. 38, 137–161. 817 Kaifu, Y., Zaim, Y., Baba, H., Kurniawan, I., Kubo, D., Rizal, Y., Arif, J., Aziz, F., 2011. New 818 reconstruction and morphological description of a Homo erectus cranium Skill IX (Tjg- 819 1993.05) from Sangiran, Central Java. J. Hum. Evol. 61, 270–294. 820 Kelley, J., Bolter, D., 2013. Growth, development, and life history in hominin evolution. In: 821 Begun, D.R. (Ed.), A Companion to Paleoanthropology. Blackwell Publishing, Chichester, 822 pp. 97–117. 823 Kirk, E.C., Gosselin-Ildari, A.D., 2009. Cochlear labyrinth volume and hearing abilities in 824 primates. Anat Rec. 292, 765–776. 825 Koenigswald, G.H.R. von, 1940. Neue Pithecanthropus-Funde 1936-1938. Ein Beitrag zur 826 Kenntnis der Praehominiden. Wetenschappelijke Mededeelingen Dienst Mijnbouw in 827 Nederlandsch-Indië 28, 1–232. 828 Koenigswald, G.H.R. von, 1951. Introduction. In: Weidenreich, F. (Ed.), Morphology of Solo 829 Man. Anthropol. Pap. Am. Mus. Nat. Hist. 43, 205–290. 830 Koenigswald, G.H.R. von, Weidenreich, F., 1939. The relationships between 831 Pithecanthropus and Sinanthropus. Nature 144, 926–929. 832 Krantz, G.S., 1975. An explanation for the diastema of Javan erectus skull IV. In: Tuttle, R.H. 833 (Ed.), Paleoanthropology, Morphology and Paleoecology. Mouton Publishers, The Hague, 834 pp. 361–372. 835 Krantz, G.S., 1994. The palate of skull Sangiran 4 from Java. Cour. Forsch.-inst. Senck. 171, 836 69–74. 837 Larick, R., Ciochon, R.L., Zaim, Y., Sudijono, Suminto, Rizal, Y., Aziz, F., Reagan, M., 838 Heizler, M., 2001. Early Pleistocene 40Ar/39Ar ages for Kabuh Formation hominins, Central 839 Java, Indonesia. Proc. Natl. Acad. Sci. USA 98, 4866–4871. 840 Le Maître, A., Schuetz, P., Vignaud, P., Brunet, M., 2017. New data about semicircular canal 841 morphology and locomotion in modern hominoids. J. Anat. 231, 95–109. 842 Manoussaki, D., Chadwick, R.S., Ketten, D.R., Arruda, J., Dimitriadis, E.K., O'Malley, J.T., 843 2008. The influence of cochlear shape on low-frequency hearing. Proc. Natl. Acad. Sci. 844 USA 105, 6162–6166. 845 Matsu'ura, S., Kondo, M., Takeshita, Y., Kumai, H., Hyodo, M., Kamishima, Y., Kaneda, T., 846 Aziz, F., 2005. Probable source horizon of the Pithecanthropus II (Sangiran 2) calotte and 847 its possible palaeoanthropological implications. Anthropol. Sci. 113, 323. 848 Matsu’ura, S., Kondo, M., Danhara, T., Sakata, S., Iwano, H., Hirata, T., Kurniawan, I., 849 Setiyabudi, E., Takeshita, Y., Hyodo, M., Kitaba, I., Sudo, M., Danhara, Y., Aziz, F., 2020. 850 Age control of the first appearance datum for Javanese Homo erectus in the Sangiran 851 area. Science 367, 210–214. 852 Mayr, E., 1950. Taxonomic categories in fossil hominids. Cold Spr. Harb. Symp. Quant. Biol. 853 15,109–118. 854 McKee, J.K., Thackeray, J.F., Berger, L.R., 1995. Faunal assemblage seriation of Southern 855 African Pliocene and Pleistocene fossil deposits. Am. J. Phys. Anthropol. 96, 235–250. 856 McRackan, T.R., Reda, F.A., Rivas, A., Noble, J.H., Deitrich, M.M., Dawant, B.M., Labadie, 857 R.F., 2012. Comparison of cochlear implant relevant anatomy in children versus adults. 858 Otol. Neurotol. 33, 328–334. 859 Meyer, M., Arsuaga, J.L., de Filippo, C., Nagel, S., Aximu-Petri, A., Nickel, B., Martínez, I., 860 Gracia, A., Bermúdez de Castro, J.M., Carbonell, E., Viola, B., Kelso, J., Prüfer, K., 861 Pääbo, S., 2016. Nuclear DNA sequences from the Middle Pleistocene Sima de los 862 Huesos hominins. Nature 531, 504–507. 863 Milton, K., 1999. A hypothesis to explain the role of meat-eating in human evolution. Evol. 864 Anthropol. 8, 11–21. 865 Mosimann, J.E., 1970. Size allometry: Size and shape variables with characterizations of the 866 log-normal and gamma distributions. J. Am. Stat. Assoc. 56, 930–-945. 867 Ni, X., Flynn, J.J., Wyss, A. R., 2010. The bony labyrinth of the early platyrrhine primate 868 Chilecebus. J. Hum. Evol. 59, 595–607. 869 Osipov, B., Harvati, K., Nathena, D., Spanakis, K., Karantanas, A., Kranioti, E. F., 2013. 870 Sexual dimorphism of the bony labyrinth: A new age-independent method. Am. J. Phys. 871 Anthropol. 151, 290–301. 872 Pickering, R., Kramers, J.D., 2010. Reappraisal of the stratigraphy and determination of new 873 U-Pb dates for the Sterkfontein hominid site, South Africa. J. Hum. Evol. 59, 70–86. 874 Pietsch, M., Davila, L.A., Erfurt, P., Avci, E., Lenarz, T., Kral, A., 2017. Spiral form of the 875 human cochlea results from spatial constraints. Sci. Rep. 7, 7500. 876 Ponce de León, M.S., Koesbardiati, T., Weissmann, J.D., Milella, M., Reyna-Blanco, C.S., 877 Suwa, G., Kondo, O., Malaspinas, A.-S., White, T.D., Zollikofer, C.P.E., 2018. Human 878 bony labyrinth is an indicator of population history and dispersal from Africa. Proc. Natl. 879 Acad. Sci. USA 115, 4128–4133. 880 Ponce de León, M.S., Bienvenu, T., Marom, A., Engel, S., Tafforeau, P., Warren, J.L.A., 881 Lordkipanidze, D., Kurniawan, I., Murti, D.B., Suriyanto, R.A. Koesbardiati, T., Zollikofer, 882 C.P.E., 2021. The primitive brain of early Homo. Science 372, 165–171. 883 Quam, R.M., Martínez, I., Rosa, M., Bonmatí, A., Lorenzo, C., de Ruiter, D.J., Moggi-Cecchi, 884 J., Conde-Valverde, M., Jarabo, P., Menter, C.G., Thackeray, J.F., Arsuaga, J.L., 2015. 885 Early hominin auditory capacities. Sci. Adv. 1, e1500355. 886 Quam, R., Lorenzo, C., Martínez, I., Gracie-Téllez, A., Arsuaga, J.L., 2016. The bony 887 labyrinth of the middle Pleistocene Sima de los Huesos hominins (Sierra de Atapuerca, 888 Spain). J. Hum. Evol. 90, 1–15. 889 R Core Team, 2019. R: A language and environment for statistical computing. R Foundation 890 for Statistical Computing, Vienna. 891 Rendu, W., Beauval, C., Crevecoeur, I., Bayle, P., Balzeau, A., Bismuth, T., Bourguignon, L., 892 Delfour, G., Faivre, J.-P., Lacrampe-Cuyaubère, F., 2014. Evidence supporting an 893 intentional Neandertal burial at La Chapelle-aux-Saints. Proc. Natl. Acad. Sci. USA 111, 894 81–86. 895 Rightmire, P., 1990. The Evolution of Homo erectus. Cambridge University Press, 896 Cambridge. 897 Rink, W., Schwarcz, H., Lee, H., Rees-Jones, J., Rabinovich, R., Hovers, E., 2001. Electron 898 spin resonance (ESR) and thermal ionization mass spectrometric (TIMS) 230Th/234U dating 899 of teeth in Middle Paleolithic layers at Amud Cave, Israel. Geoarchaeology 16, 701–717. 900 Rook, L., Bondioli, L., Casali, F., Rossi, M., Köhler, M., Moyá-Solá, S., Macchiarelli, R., 901 2004. The bony labyrinth of Oreopithecus bambolii. J. Hum. Evol. 46, 349–356. 902 Santa Luca, A.P., 1980. The Ngandong Fossil Hominids: A Comparative Study of a Far 903 Eastern Homo erectus Group. Yale University Press, New Haven. 904 Scheuer, L., Black, S., 2000. Developmental Juvenile Osteology. Academic Press, San 905 Diego. 906 Schlager, S., 2017. Morpho and Rvcg-shape analysis in R: R-packages for geometric 907 morphometrics, shape analysis and surface manipulations. In: Zheng, G., Li, S., Székely, 908 G. (Eds.), Statistical Shape and Deformation Analysis. Methods, Implementation and 909 Applications. Academic Press, London, pp. 217–256. 910 Schneider, C.A., Rasband, W.S., Eliceiri, K.W., 2012. NIH Image to ImageJ: 25 years of 911 image analysis. Nat. Methods 9, 671–675. 912 Schwarcz, H., Buhay, W., Grün, R., Valladas, H., Tchernov, E., Bar-Yosef, O., 913 Vandermeersch, B., 1989. ESR dating of the Neanderthal site, Kebara Cave, Israel. J. 914 Archaeol. Sci. 16, 653–659. 915 Schwartz, J., 2016. Beyond Homo erectus: Sangiran is key to deciphering the Asian human 916 fossil record and re-evaluating Homo. In: Ribot Trafí, F. (Ed.), Homenaje al Dr. José 917 Gibert Clols. Una Vida Dedicada a la Ciencia y al Conocimiento de los Primeros 918 Europeos. Publicaciones Diputación de Granada, Granada, pp. 93-110. 919 Schwartz, J.H., Tattersall, I. 2000. What constitutes Homo erectus? Acta Anthropol. Sin. 920 Suppl. 19, 18-22. 921 Schwartz, J.H., Tattersall, I., 2003. The Human Fossil Record, Vol. 2. Craniodental 922 Morphology of Genus Homo (Africa and Asia). Wiley-Liss, Hoboken. 923 Sémah, F., Saleki, H., Falguères, C., Féraud, G., Djubiantono, T., 2000. Did Early Man 924 reach Java during the Late Pleistocene? J. Archaeol. Sci. 27, 763–769. 925 Smith, T.M., Olejniczak, A.J., Kupczik, K., Lazzari, V., de Vos, J., Kullmer, O., Schrenk, F., 926 Hublin, J.-J., Jacob, T., Tafforeau, P., 2009. Taxonomic assessment of the Trinil molars 927 using non-destructive 3D structural and development analysis. PaleoAnthropology 2009, 928 117–129. 929 Smith, T.M., Houssaye, A., Kullmer, O., Le Cabec, A., Olejniczak, A.J., Schrenk, F., de Vos, 930 J., Tafforeau, P., 2018. Disentangling isolated dental remains of Asian Pleistocene 931 hominins and pongines. PloS One 13, e0204737. 932 Spoor, F., 1993. The comparative morphology and phylogeny of the human bony labyrinth. 933 Ph.D. Dissertation, University of Utrecht. 934 Spoor, F., Zonneveld, F., 1994. The bony labyrinth in Homo erectus: A preliminary report. 935 Cour. Forsch.-Inst. Senck. 171, 251–256. 936 Spoor, F., Zonneveld., F., 1995. Morphometry of the primate bony labyrinth: a new method 937 based on high-resolution computed tomography. J. Anat. 186, 271–286. 938 Spoor, F., Zonneveld., F., 1998. Comparative review of the human bony labyrinth. Yearb. 939 Phys. Anthropol. 41, 211–251. 940 Spoor, F., Wood, B., Zonneveld, F., 1994. Implications of early hominid labyrinthine 941 morphology for evolution of human bipedal locomotion. Nature 369, 645–648. 942 Spoor, F., Hublin, J.-J., Braun, M., Zonneveld, F., 2003. The bony labyrinth of Neanderthals. 943 J. Hum. Evol. 44, 141–165. 944 Thompson, J.C., Carvalho, S., Marean, C.W., Alemseged, Z., 2019. Origins of the human 945 predatory pattern. The transition to large-animal exploitation by early hominins. Curr. 946 Anthropol. 60, 1–23. 947 Thorne, A., Wolpoff, M., 1982. Regional continuity in Australasian Pleistocene hominid 948 evolution. Am. J. Phys. Anthropol. 55, 337–49. 949 Tyler, D.E., 2001. Meganthropus: Cranial fossils from Java. Hum. Evol. 16, 81–101. 950 Tyler, D.E., 2003. Sangiran 5, (‘‘Pithecanthropus dubius’’), Homo erectus, ‘‘Meganthropus,’’ 951 or Pongo? Hum. Evol. 18, 229–242. 952 Tyler, D.E., 2004. An examination of the taxonomic status of the fragmentary mandible 953 Sangiran 5, (Pithecanthropus dubius), Homo erectus, ‘‘Meganthropus’’, or Pongo? Quat. 954 Int. 117, 125–130. 955 Tyler, D.E., Sartono, S., 2001. A new Homo erectus cranium from Sangiran, Java. Hum. 956 Evol. 16, 13–25. 957 Urciuoli, A., Zanolli, C., Beaudet, A., Dumoncel, J., Santos, F., Moyà-Solà, S., Alba, D.M., 958 2020. The evolution of the vestibular apparatus in apes and humans. eLife 9, e51261. 959 Urciuoli, A., Zanolli, C., Almécija, S., Beaudet, A., Dumoncel, J., Morimoto, N., Nakatsukasa, 960 M., Moyà-Solà, S., Begun, D.R., Alba, D.M., 2021a. Reassessment of the phylogenetic 961 relationships of the late Miocene apes Hispanopithecus and Rudapithecus based on 962 vestibular morphology. Proc. Natl. Acad. Sci. USA 118, e2015215118. 963 Urciuoli, A., Zanolli, C., Beaudet, A., Pina, M., Almécija, S., Moyà-Solà, S., Alba, D.M., 964 2021b. A comparative analysis of the vestibular apparatus in Epipliopithecus 965 vindobonensis: Phylogenetic implications. J. Hum. Evol. 151, 102930. 966 Ward, D. L., Pomeroy, E., Schroeder, L., Viola, T. B., Silcox, M.T., Stock, J.T., 2020. Can 967 bony labyrinth dimensions predict biological sex in archaeological samples? J. Archaeol. 968 Sci. Rep. 31, 102354. 969 Watanabe, N., Kadar, D., 1985.Quaternary Geology of the Hominid Fossil Bearing 970 Formations in Java. Geological Research and Development Centre, Bandung. 971 Weidenreich, F., 1945. Giant early man from Java and South China. Anthropol. Pap. Am. 972 Mus. Nat. Hist. 40, 1–134. 973 Weidenreich, F., 1956. Apes, Giants, and Man. University of Chicago Press, Chicago. 974 Wolpoff, M.H., 1996. Human Evolution. McGraw-Hill, New York. 975 Wolpoff, M.H., Hawks, J., Frayer, D.W, Huntley, K., 2001. Modern human ancestry at the 976 peripheries: A test of the replacement theory. Science 291, 293–297. 977 Wood, B., Collard, M., 1999. The changing face of genus Homo. Evol. Anthropol. 8, 195– 978 207. 979 Wu, X.-J., Zhang, Y., 2016. The temporal bony labyrinthine morphology of Lantian Homo 980 erectus from Gongwangling, Shaanxi province. Acta Anthropol. Sin. 35, 14–23. 981 Wu, X.-J., Crevecoeur, I., Liu, W., Xing, S., Trinkaus, E., 2014. Temporal labyrinths of 982 eastern Pleistocene humans. Proc. Natl. Acad. Sci. USA, 111, 10509–10513. 983 Zanolli, C., Kullmer, O., Kelley, J., Bacon, A.-M., Demeter, F., Dumoncel, J., Fiorenza, L., 984 Grine, F.E., Hublin, J.-J., Nguyen, A.T., Nguyen, T.M.H., Pan, L., Schillinger, B., Schrenk, 985 F., Skinner, M.M., Ji, X., Macchiarelli, R., 2019. Evidence for increased hominid diversity 986 in the Early to Middle Pleistocene of Indonesia. Nat. Ecol. Evol. 3, 755–764. 987 Zhu, R.X., Potts, R., Pan, Y.X., Yao, H.T., Lü, L.Q., Zhao, X., Gao, X., Chen, L.W., Gao, F., 988 Deng, C.L., 2004. Early evidence of the genus Homo in East Asia. J. Hum. Evol. 55, 989 1075–1085. 990 991 Figure captions 992 993 Figure 1. Lateral view of Sangiran 2 calvaria (a) and Sangiran 4 partial cranium (b). Not to 994 scale. 3D model renderings of the Sangiran 2 (c, in red) and Sangiran 4 (d, in blue) cochlea 995 in lateral (top) and inferior (bottom) views. Scale bar = 5 mm. 996 997 Figure 2. Proportions of the first cochlear turn cross section. a) Bivariate ordinary least 998 square regression between log-transformed cross section height (log Sh) and width (log Sw). 999 b) Box-and-whisker plot of cochlear cross section aspect ratio (Sw/Sh). Horizontal lines 1000 correspond to the median, boxes depict interquartile range, whiskers represent maximum 1001 and minimum values within 1.5 times the interquartile range, and black dots are outliers. The 1002 3D models at the low and high end of the distribution represent extreme morphologies of 1003 cross-sectional shape (shaded in red): round (above) vs. oval (below). 1004 1005 Figure 3. Box-and-whisker plots of the cochlear variables analyzed in the present study and 1006 derived from Conde-Valverde et al. (2018): a) number of turns (NT); b) cochlear volume (Vol, 1007 in mm3); c) cochlear length (CL, in mm); d) length of the first cochlear turn (L1, in mm); e) 1008 length of the third cochlear turn (L3, in mm); f) size of the first turn radius (R1, in mm); g) size 1009 of the third turn radius (R3, in mm); h) cochlear thickness (CTh, in mm); i) height of the cross 1010 section of the first cochlear turn (Sh, in mm); j) width of the cross section of the first cochlear 1011 turn (Sw, in mm). Horizontal lines correspond to the median, boxes depict interquartile 1012 range, whiskers represent maximum and minimum values within 1.5 times the interquartile 1013 range, and black dots are outliers. 1014 1015 Figure 4. Box-and-whisker plots of cochlear turn proportions. Vertical lines correspond to the 1016 median, boxes depict interquartile range, whiskers represent maximum and minimum values 1017 within 1.5 times the interquartile range, and black dots are outliers. The wireframe on the 1018 right depicts the three cochlear turns with red (L1), green (L2), and blue (L3) spheres. 1019 1020 Figure 5. Patterns of cochlear shape variation among hominins based on the results of a 1021 principal component analysis, as depicted by bivariate plots between the first two principal 1022 components: PC2 vs. PC1. The variance explained by each PC is given along the 1023 corresponding axis. Four groups (modern humans, australopiths [Australopithecus + 1024 Paranthropus], Neanderthals, and Sima de los Huesos) were defined a priori. The convex 1025 hulls corresponding to the range of variation of each group are colored as follows: modern 1026 humans = blue; australopiths = brown; Neanderthals = lilac; Sima de los Huesos = green. 1027 1028 Figure 6. Patterns of cochlear shape variation among hominins based on the results of a 1029 canonical variates analysis, as depicted by bivariate plots between the first two canonical 1030 variates (CV): CV2 vs. CV1. The variance explained by each CV is given along the 1031 corresponding axis. Four groups (modern humans, australopiths [Australopithecus + 1032 Paranthropus], Neanderthals, and Sima de los Huesos) were defined a priori, while 1033 individual fossil specimens (the H. erectus S2 and S4, the early Homo SK 27 and SK 847, 1034 and the Middle Pleistocene human Aroeira 3) were projected post hoc onto the 1035 morphospace. The convex hulls corresponding to the range of variation of each group are 1036 colored as follows: modern humans = blue; australopiths = brown; Neanderthals = lilac; 1037 Sima de los Huesos = green. 1038 1039 Figure 7. Box-and-whisker plots of the hearing range limit estimates obtained for the studied 1040 and comparative samples: a) adjusted high frequency limit (in kHz; at 60db); b) low 1041 frequency limit (in Hz; at 60db). Horizontal lines correspond to the median, boxes depict 1042 interquartile range, whiskers represent maximum and minimum values within 1.5 times the 1043 interquartile range, and black dots are outliers. The plots are presented in logarithmic scale 1044 to facilitate the comparison between the results obtained for the high and low frequency limit. 1045 Abbreviations: AU = Australopithecus + Paranthropus; MH = modern humans; NE = 1046 Neanderthals; SH = Sima de los Huesos humans 1047 Figure 1_rev a b Figure 2 Sangiran 4 1.0 0.9 SK847 Sangiran 4 Sangiran 2 0.8 Aroeira 3 0.8 log Sh SK27 Sangiran 2 Aroeira 3 SK847 0.7 0.6 SK27 Australopithecus Sangiran 2 Paranthropus Sangiran 4 Sima de los Huesos SK847 Australopithecus (n = 10) Paranthropus (n = 3) Neanderthals SK27 Neanderthals (n = 7) Sima de los Huesos (n = 10) 0.4 Modern humans Aroeira 3 Modern humans (n = 10) 0.4 0.6 0.8 log Sw Sw/Sh Figure 3_rev 2.9 a NT b Vol 45 c CL d L1 100 25.0 2.7 40 80 22.5 2.5 60 35 20.0 2.3 40 17.5 1.6 e L3 4.5 f R1 g R3 3.0 h CTh 2.00 2.5 1.2 4.0 1.75 2.0 1.50 0.8 3.5 1.25 1.5 i Sh j Sw k ABT 3.0 5.0 Australopithecus (n = 10) 2.25 Paranthropus (n = 3) 4.0 2.5 Modern humans (n = 10) 2.00 Neanderthals (n = 7) 3.0 Sima de los Huesos (n = 10) 1.75 2.0 S2 SK27 S4 Aroeira 3 1.50 2.0 SK847 1.5 Figure 4_rev Proportional length 0% 10% 20% 30% 40% 50% 60% 70% Australopithecus (n = 10) L1 Paranthropus (n = 3) Modern humans (n = 10) Neanderthals (n = 7) L2 Sima de los Huesos (n = 10) Sangiran 2 Sangiran 4 SK27 L3 SK847 Aroeira 3 L3 L2 L1 Figure 5_rev La Quina H5 2 La Ferrassie 1 La Chapelle-aux-Saints StW 573 Amud 1 Kebara 1 La Ferrassie 2 PC2 (16.8%) AT-1907 SK847 AT-421 Cr.4 Cr.7 Cr.13 SK 879 0 Cr.12 Cr.3 La Ferrassie 8 MLD 37/38 StS 5 Aroeira 3 StW 98 StW 255 Sangiran 4 AT-643 StW 53g Cr.8 StW 578 SK27 Cr.15 DNH 22 StW 329 -2 SKW 18 Sangiran 2 StW 151c Australopithecus Sangiran 2 Paranthropus Sangiran 4 Sima de los Huesos SK847 Neanderthals SK27 StS 19 Modern humans Aroeira 3 -6 -4 -2 0 2 4 PC1 (48.8%) Figure 6_rev Australopithecus Sangiran 2 Paranthropus Sangiran 4 4 Sima de los Huesos SK847 Neanderthals SK27 Modern humans Aroeira 3 La Ferrassie 2 2 La Ferrassie 1 La Quina H5 La Chapelle-aux-Saints StW 573 Kebara 1 CV2 (31.2%) Amud 1 StW 255 AT-1907 0 SK 847 MLD 37/38 Cr.8 Sangiran 4 Cr.13 SK 879 Aroeira 3 Cr.4 Cr.7 DNH 22 La Ferrassie 8 StW 329 Sangiran 2 Cr.15 AT-421 SK 27 StW 578 Cr.3 −2 StS 5 Cr.12 AT-643 StS 19 StW 53g StW 151c StW 98 SKW 18 −2.5 0.0 2.5 5.0 CV1 (63.9%) Figure_7_R2 a b Adjusted high frequency limit (60db) Low frequency limit (60db) SK27 40 Hz SK27 24 kHz 22 kHz 35 Hz SK847 30 Hz SK847 20 kHz Sangiran 2 Sangiran 2 SH SH (n = 10) Aroeira 3 (n = 10) Aroeira 3 18 kHz 25 Hz Sangiran 4 Sangiran 4 AU AU (n = 13) (n = 13) NE NE MH (n = 7) MH (n = 7) (n = 10) (n = 10) Table 1 Cochlear variables measured for Indonesian Homo erectus (Sangiran 2 and 4) and the comparative sample. Specimens/samples CL L1 L3 Vol Sw Sh R1 R3 CTh NT ABT Indonesian H. erectus Sangiran 2 39.04 21.20 5.66 66.80 1.81 2.14 3.79 1.81 1.58 2.54 2.93 Sangiran 4 41.72 24.20 5.17 89.13 2.39 2.45 4.32 1.57 1.70 2.47 4.39 Early Homo: SK27 32.11 19.65 2.02 41.34 1.4 1.81 3.2225 1.26 1.54 2.15 1.9 SK847 35.73 20.70 3.52 63.74 1.75 2.39 3.7225 1.69 2.06 2.333 2.99 Middle Pleistocene Homo Aroeira 3 40.94 21.47 7.01 76.46 1.72 2.20 3.84 1.94 2.54 2.64 2.80 Australopithecus (n = 10) Mean (SD) 35.86 (2.62) 19.70 (1.27) 5.23 (0.85) 58.97 (15.75) 1.70 (0.22) 2.10 (0.27) 3.53 (0.23) 1.46 (0.21) 2.01 (0.39) 2.63 (0.15) 2.63 (0.63) Range 31.31–40.30 17.88–22.03 3.59–6.36 36.02–94.82 1.31–1.98 1.51–2.41 3.21–3.96 1.15–1.89 1.39–2.77 2.41–2.90 1.51-3.57 Paranthropus (n = 3) Mean (SD) 36.46 (1.48) 21.17 (0.20) 3.89 (0.99) 63.62 (6.13) 1.90 (0.06) 2.31 (0.18) 3.73 (0.13) 1.75 (0.06) 1.65 (0.25) 2.47 (0.10) 3.21 (0.46) Range 35.01–38.48 20.90–21.37 3.16–5.30 58.11–72.17 1.82–1.97 2.17–2.57 3.67–3.80 1.53–2.03 1.35–1.96 2.36–2.60 2.85-3.86 Neanderthals (n = 7)a Mean (SD) 41.11 (2.89) 24.13 (1.53) 3.90 (1.57) 78.56 (13.27) 1.84 (0.13) 2.70 (0.21) 3.99 (0.24) 1.61 (0.26) 2.30 (0.35) 2.37 (0.13) 3.63 (0.53) Range 36.30–45.00 21.60–26.80 2.00–6.80 54.70–102.00 1.70–2.10 2.30–2.90 3.50–4.30 1.30–2.10 1.70–2.80 2.18–2.59 2.70-4.40 Sima de los Huesos (n = 10)a Mean (SD) 39.41 (1.96) 22.1 (0.95) 5.10 (1.06) 60.31 (8.85) 1.71 (0.15) 2.33 (0.20) 3.73 (0.12) 1.45 (0.15) 2.03 (0.24) 2.53 (0.15) 3.11 (0.49) Range 35.30–42.10 20.50–23.60 2.90–6.60 44.90–71.60 1.40–1.90 1.80–2.50 3.50–4.30 1.30–2.10 1.70–2.80 2.18–2.59 1.90-3.70 Modern humans (n = 10) Mean (SD) 41.81 (2.01) 22.91 (1.06) 6.46 (0.94) 91.01 (9.65) 1.88 (0.17) 2.68 (0.17) 4.13 (0.18) 1.75 (0.22) 2.59 (0.26) 2.62 (0.08) 3.81 (0.61) Range 38.62–45.47 21.61–24.74 4.12–7.61 72.64–107.98 1.65–2.20 2.33–2.95 3.86–4.48 1.48–2.18 2.11–3.01 2.43–2.7 3.05-5.01 Abbreviations: CL = cochlear length (mm); L1 = length of the first turn (mm); L3 = length of the third turn (mm); Vol = cochlear volume (mm3); Sw = width of the first turn cross section (mm); Sh = height of the first turn cross section (mm); R1 = radius of the 1st (basal) turn (mm); R3 = radius of the 3rd (apical) turn (mm); CTh = cochlear thickness (mm); NT = number of turns; ABT = cross-sectional area at the first basal turn. a Data taken from Conde-Valverde et al. (2019). Table 2 Z-scores computed for individual fossil specimens based on the distribution (mean and SD) of Sw/Sh in the various groups included in the comparative samplea. Fossil specimens AU NE SH MH Sangiran 2 0.46 4.56 4.11 3.01 Sangiran 4 2.43 8.20 8.91 5.70 SK27 -0.64 2.52 1.44 1.50 SK847 -1.26 1.36 -0.09 0.65 Aroeira 3 -0.51 2.76 1.75 1.68 Abbreviations: AU = Australopiths (Australopithecus + Paranthropus); NE = Neanderthals; SH = Sima de los Huesos; MH = modern humans; Sw/Sh = ratio between the width and height of the cochlear first turn cross section. a Z-scores greater than |1.96|, indicating significant differences relative to the reference distributions, are bolded. Table 3 Shape variable loadings for the first two principal components (PCs) of the principal components analysis depicted in Figure 5. Variable PC1 PC2 NT* 0.0500 -0.5959 CL* -0.3196 -0.0835 L1* -0.4151 -0.0073 L3* 0.4265 -0.0667 Sw* -0.3192 -0.0588 Sh* -0.2754 0.5068 Vol1/3* -0.3921 0.0156 R1* -0.4217 -0.1172 R3* -0.0414 -0.2095 CTh* 0.1734 0.5615 Abbreviations: NT* = number of turns; CL* = cochlear length; L1* = length of the first turn; L3* = length of the third turn; Sw* = width of the first turn cross section; Sh* = height of the first turn cross section; Vol1/3* = cube root of cochlear volume; R1* = radius of the first turn; R3* = radius of the third turn; CTh* = cochlear thickness. Table 4 Classification results of a canonical variate analysis based on cochlear variables and performed using the four groups distinguished among the comparative sample. For each group distinguished a priori, the number of individuals classified in each group and the percentage of correctly classified individuals is provided. Cross-validated results are given within parentheses. n Group % AU MH NE SH Australopiths 92.3% (84.6%) 12 (11) 1 (2) 0 (0) 0 (0) MH 100% (100%) 0 (0) 10 (10) 0 (0) 0 (0) NE 100% (71.4%) 0 (0) 0 (0) 7 (5) 0 (2) SH 100% (70.0%) 0 (1) 0 (0) 0 (2) 10 (7) Abbreviations: AU = Australopiths (Australopithecus + Paranthropus); MH = modern humans; NE = Neanderthals; SH = Sima de los Huesos. Table 5 Shape variable loadings for the first two canonical variates (CVs) of the canonical variate analysis depicted in Figure 6. Variable CV1 CV2 NT* -8.316 -19.446 CL* 6.078 -5.542 L1* -0.562 7.637 L3* -8.699 12.260 Sw* -18.854 -23.407 Sh* 11.286 16.110 Vol1/3* -30.097 30.358 R1* -24.654 11.975 R3* -4.467 7.143 CTh* -0.069 6.393 Abbreviations: NT* = number of turns; CL* = cochlear length; L1* = length of the first turn; L3* = length of the third turn; Sw* = width of the first turn cross section; Sh* = height of the first turn cross section; Vol1/3* = cube root of cochlear volume; R1* = radius of the first turn; R3* = radius of the third turn; CTh* = cochlear thickness. Table 6 Cross-validated classification results for individual fossil specimens based on a canonical variate analysis of cochlear variables performed using four groups distinguished a priori among the comparative sample. Fossil specimens AU MH NE SH Posterior probabilitiesa Sangiran 2 99.5% 0.3% 0.0% 0.2% Sangiran 4 96.3% 3.7% 0.0% 0.1% SK27 0.0% 0.0% 92.0% 8.0% SK847 95.5% 2.7% 0.0% 1.8% Aroeira 3 15.7% 1.5% 0.2% 82.8% Typicality probabilitiesb Sangiran 2 0.141 0.001 <0.001 0.001 Sangiran 4 <0.001 <0.001 <0.001 <0.001 SK27 <0.001 <0.001 0.001 0.001 SK847 0.107 0.004 <0.001 0.003 Aroeira 3 0.012 0.001 <0.001 0.054 Abbreviations: AU = Australopiths (Australopithecus + Paranthropus); MH = modern humans; NE = Neanderthals; SH = Sima de los Huesos. a These probabilities assume that each fossil specimen belongs to one of the four groups defined a priori, so that the sum of probabilities equals 100%. The highest nonsignificant probability (if any) is bolded. b These probabilities denote the probability of having the score of each fossil specimen given membership in a particular group, not the likelihood of group membership in each of the a priori defined groups given a particular score. p < 0.05 thus denotes outliers to the distribution of each a priori group. The highest probability for each fossil specimen is bolded. Table 7 Inferred high and low frequency limits based on cochlear parameters (Kirk and Gosselin- Ildari, 2009). Specimens/samples HFL AHFL LFL Indonesian Homo erectus Sangiran 2 23.45 19.85 29.42 Sangiran 4 20.89 17.69 24.60 Early Homo SK27 28.41 24.05 39.61 SK847 23.89 20.23 30.29 Middle Pleistocene Homo Aroeira 3 22.21 18.81 27.06 Australopithecus (n = 10) Mean (SD) 25.13 (2.64) 21.28 (2.24) 32.91 (5.37) Range 20.38–30.02 17.26–25.42 23.68–43.15 Paranthropus (n = 3) Mean (SD) 23.97(0.89) 20.29 (0.75) 30.46 (1.74) Range 22.73–24.79 19.25–20.99 28.04–32.07 Neanderthals (n = 7) Mean (SD) 22.16 (1.59) 18.77 (1.35) 27.02 (3.04) Range 19.80–25.40 16.76–21.50 22.63–33.30 Sima de los Huesos (n = 10) Mean (SD) 24.58 (1.53) 20.82 (1.29) 31.71 (3.07) Range 22.81–27.49 19.31–23.27 28.18–37.64 Modern humans (n = 10) Mean (SD) 20.79 (0.91) 17.60 (0.77) 24.43 (1.66) Range 19.35–22.67 16.38–19.20 21.84–27.93 Abbreviations: HFL = high frequency limit (kHz) at 60 dB; AHFL = adjusted high frequency limit (kHz) at 60 dB; LFL = low frequency limit (Hz) at 60 dB. Table 8 Z-scores computed for individual fossil specimens based on the distribution (mean and SD) of AHFL and LFL in the various groups included in the comparative sample. Fossil specimens AU NE SH MH AHFL Z-scores: Sangiran 2 -0.59 0.81 -0.74 2.93 Sangiran 4 -1.65 -0.80 -2.42 0.12 SK27 1.47 3.92 2.51 8.40 SK847 -0.40 1.09 -0.45 3.42 Aroeira 3 -1.10 0.03 -1.56 1.57 LFL Z-scores: Sangiran 2 -0.60 0.79 -0.75 3.00 Sangiran 4 -1.58 -0.79 -2.31 0.11 SK27 1.49 4.14 2.57 9.14 SK847 -0.42 1.07 -0.46 3.53 Aroeira 3 -1.08 0.01 -1.51 1.58 Abbreviations: AHFL = Adjusted high frequency limit (kHz) at 60 dB; LFL = low frequency limit (kHz) at 60 dB; AU = Australopiths (Australopithecus + Paranthropus); NE = Neanderthals; SH = Sima de los Huesos; MH = modern humans Table 9 Updated character state summary combining the information provided in the present study with previously published dataa. Taxon CL %L1 %L3 NT ABT Sw/Sh Vol CTh Pan troglodytes Long Short Long High Small Round Small Low Australopiths Short Short Long Low Small Round Small Low Early Homo Short Long Short Low/Very low Small Oval-shaped Small Low Homo erectus Long Intermediate/Long Intermediate/Short Low Small/Large Round Intermedia Low te/Large Sima de los Huesos Hominins Long Short Short Low Small Oval-shaped Small Low Neanderthals Long Long Very short Very low Large Oval-shaped Large High Modern humans Long Long Short Low Large Oval-shaped Large High Abbreviations: CL = cochlear length; L1% = proportional length of the first turn; %L3 = proportional length of the third turn; NT = number of turns; ABT = cross-sectional area at the first basal turn; Sw/Sh = ratio between the cochlear first turn cross section width; Vol = cochlear volume; CTh = cochlear thickness. a States reported here for the first time or modified relative to Conde-Valverde et al. (2019) are in bold.