This is the accepted version of the article: Martín-Perea, David M.; Domingo, Maria S.; Cantero, Enrique; [et al.]. «Re- curring taphonomic processes in the carnivoran-dominated Late Miocene assem- blages of Batallones-3, Madrid Basin, Spain». Lethaia, (September 2021). DOI 10.1111/let.12445 This version is available at https://ddd.uab.cat/record/248992 under the terms of the license 1   Recurring taphonomic processes in the carnivoran-dominated assemblages of 2   Batallones-3 (Late Miocene, Madrid Basin, Spain) 3   4   David M. Martín-Pereaa, b, c*, M. Soledad Domingod, Enrique Canteroa, Lloyd A. 5   Courtenaye, Alberto Valencianof, Lucía R. Sualdeab, Juan Abellag, h, Jorge Moralesa 6   a 7   Palaeobiology Department, National Natural Sciences Museum – CSIC, Madrid, 8   Spain b 9   Geodynamics, Stratigraphy and Palaeontology Department, Faculty of Geology, 10   Complutense University of Madrid, Madrid, Spain c 11   Institute of Evolution in Africa – IDEA, Madrid, Spain d 12   Sciences, Social Sciences and Mathematics Education Department, Faculty of 13   Education, Complutense University of Madrid, Madrid, Spain e 14   Department of Cartographic and Land Engineering, Higher Polytechnic School of 15   Avila, University of Salamanca, Avila, Spain f 16   Earth Sciences Department, Paleontology Area and Environmental Sciences 17   Research Institute (IUCA), Zaragoza University, Zaragoza, Spain g 18   Institut Català de Paleontologia Miquel Crusafont, Cerdanyola del Vallès, Spain h 19   Biodiversity National Institute (INABIO), Quito, Ecuador 20   21   *Corresponding author: davidmam@ucm.es 22   23   Abstract 24   25   Carnivoran-dominated fossil sites are scarce in the fossil record but provide precious 26   information on the diversity and ecology of past carnivoran guilds. The Cerro de los 27   Batallones sites host the oldest carnivoran-dominated assemblages, with the highest 28   carnivoran abundances observed in the fossil record. Batallones-3 (Late Miocene, 29   Madrid Basin, Spain) hosts three discrete, carnivoran-dominated fossiliferous levels 30   deposited in a 15 m diameter, 4 m high pseudokarstic cavity with 1 m high talus cone 31   located beneath the mouth of the cavity on the roof. Levels I, II and III are multitaxic 32   multidominant assemblages, with the sabretooth cats Promegantereon ogygia and 33   Machairodus aphanistus and the ursid Indarctos arctoides being the most abundant 34   species. These carnivoran-dominated assemblages are autochthonous and show 35   diagenetically fractured but well-preserved remains (complete, little to no weathering or 36   abrasion). Root marks and manganese oxide precipitation are common and more 37   abundant in Level III, due to modern pedogenic processes. There are also a few 38   allochthonous, badly preserved (weathered and abraded) herbivore remains that were 39   washed into the cavity. The taphonomic homogeneity of all three levels suggest 40   recurring taphonomic and geologic processes throughout the accumulation of infill in 41   the cave. The trap-like nature of the cave, unbroken, and largely unweathered 42   carnivoran bones suggest these predators intentionally jumped into the cave but were 43   unable to escape. 44   45   Keywords: Taphonomy, Mammals, Carnivoran, Vallesian, Pseudokarst 46   47   Introduction 48   49   Carnivore fossil remains commonly outnumber herbivore remains in mammalian fossil 50   sites, with carnivorans usually represented by less than 10% of the total Number of 51   Identified Specimens (NISP) and Minimum Number of Individuals (MNI; Klein & Cruz- 52   Uribe 1984). This amount of carnivores in fossil sites is a result of the position of this 53   group of animals at the top of the trophic pyramid in living communities, where they 54   constitute a small portion of the ecosystem biomass, with herbivore to carnivore ratios 55   estimated at 50:1 in modern ecosystems (Eisenberg 1981; Stock & Harris 1992; Farlow 56   1993). 57   58   For this reason, there are few known carnivoran-dominated fossil sites, most of them 59   Pleistocene or Holocene in age. The best examples are the Pleistocene localities of 60   Rancho La Brea (Stock 1972; Marcus & Berger 1984; Stock & Harris 1992; Binder et 61   al. 2002; Feranec et al. 2009), Talara (Lindsey & Seymour 2015), La Carolina 62   (Hoffstetter 1952), Sima de los Huesos (Arsuaga et al. 1997; García et al. 1997), 63   Moonshine Cave (White et al. 1984) and Middle Butte Cave (White et al. 1984). At 64   Rancho La Brea, Talara and La Carolina, it is proposed that herbivores trapped in 65   asphalt attracted carnivores, which also became trapped. Similar taphonomic histories 66   are inferred for the other carnivoran-dominated localities, although in caves. In these 67   cases, either karstic caves (Sima de los Huesos) or Lava Blisters (Moonshine Cave 68   and Middle Butte) acted as natural traps, where carnivores entered the cave to feed on 69   entrapped herbivores but became trapped themselves (Arsuaga et al. 1997; García et 70   al. 1997; White et al. 1984). 71   72   Not all fossil sites interpreted as the result of natural trapping host carnivoran- 73   dominated assemblages. Natural Trap Cave (Gilbert 1978; Martin & Gilbert 1978; 74   Chomko & Gilbert 1987; Wang & Martin 1993), Rantis Cave (Marder et al. 2011) and 75   Trinchera-Galería (García & Arsuaga 1998) in the Pleistocene, and Cupcake Cave 76   shaft (Lord et al. 2016), McEachern’s Deathtrap Cave (Kos 2003) and Shield Trap 77   Cave (Oliver 1986; Oliver et al. 1986) in the Holocene are all examples of naturally 78   trapped, herbivore-dominated assemblages in karstic caves. The differences between 79   these herbivore-dominated and carnivore-dominated sites are attributed to different 80   entrance modes, with animals being trapped after active entrance in carnivoran- 81   dominated assemblages and due to accidental entrance in herbivore-dominated 82   assemblages (White et al. 1984). 83   84   Since the discovery of the Cerro de los Batallones (Batallones Butte) sites in 1991, two 85   carnivoran-dominated sites, Batallones-1 and Batallones-3, have provided precious 86   information on the diversity of carnivorans in the past. These sites are unique for 87   numerous reasons: 1) dating back to the Miocene, these sites host the oldest known 88   carnivoran-dominated assemblages, 2) both assemblages host the highest carnivoran 89   abundances observed in the fossil record, 3) the excellent nature of preservation and 90   skeletal completeness of the remains allows the description of new or poorly-known 91   species, and 4) the peculiar, piping pseudokarst nature of the cavities constitutes a 92   new geological setting for fossil assemblages. These sites constitute an exceptional 93   archive that expands our knowledge of the diversity and ecology of these ancient 94   carnivoran guilds and the taphonomic history of these uncommon fossil localities. 95   96   The Cerro de los Batallones complex 97   98   The Cerro de los Batallones complex is located 30km south of Madrid (Spain, Figure 99   1A), and 1 km east of Valdemoro (Figure 1B) and belongs to the municipality of 100   Torrejón de Velasco. The paleontological complex is comprised of nine sites, which 101   host rich and diverse faunal assemblages (Morales 2017) including a wide array of 102   carnivorans, such as sabretooth cats (Antón et al. 2004; Salesa et al. 2006, 2008, 103   2010; Monescillo et al. 2014; Siliceo et al. 2014; Antón et al. 2020), other felids (Salesa 104   et al. 2012), amphicyonids (Peigné et al. 2008; Siliceo et al. 2015; Morales et al. 2017; 105   Siliceo et al. 2017, 2020; Morales et al. 2021), ailurids (Peigné et al. 2005; Antón et al. 106   2006; Salesa et al. 2006; Salesa et al. 2017), ailuropod ursids (Abella et al. 2012, 107   2013, 2015; Abella & Valenciano 2017), hyaenids (Fraile 2016, 2017), mustelids and 108   mephitids (Valenciano et al. 2015; Valenciano 2017; Valenciano et al. 2020; 109   Valenciano & Govender 2020), herbivorous mammals, including proboscideans 110   (Alberdi et al. 2017), perissodactyls (Romano et al. 2017; Sanisidro & Cantalapiedra 111   2017; Domingo et al. 2018), and artiodactyls (Sánchez et al. 2009; Sánchez et al. 112   2011; Pickford 2015; Ríos et al. 2017; Ríos & Morales 2019). These sites have also 113   yielded a rich assemblage of small mammals, including rodents, insectivores and 114   lagomorphs (López-Antoñanzas et al. 2010; 2014; Álvarez-Sierra et al. 2017; Medina- 115   Chavarrías et al. 2019). 116   117   Based on their faunal association, the Cerro de los Batallones sites are inferred to have 118   a late Vallesian age (ca. 9.1 Ma; early Late Miocene; Domingo et al. 2007; Morales et 119   al. 2008; López-Antoñanzas et al. 2010; Gómez Cano et al. 2011). However, the 120   rodent fauna indicates the sites differ in age, with Batallones-10 being the oldest, 121   followed by Batallones-1, with Batallones-3 being the youngest site in the Cerro de los 122   Batallones complex (López-Antoñanzas et al. 2010). In any event, given that López- 123   Antoñanzas et al. (2010) placed all the Batallones localities either in the local subzone 124   J2 (dated 9.71-9.48 Ma) or J3 (9.34-9.05 Ma), the age difference between the oldest 125   and youngest of these localities is at most of 0.3 Myr. 126   127   The Cerro de los Batallones sites are vertical, hourglass-shaped cavities (Figure 1C) 128   formed in the Late Miocene due to soil piping pseudokarstic processes (Pozo et al. 129   2004; Calvo et al. 2013) in a woodland palaeoenvironment with patches of wooded 130   grassland (Domingo et al. 2013b, 2016). In most of these caves, the basal strata are 131   dominated by carnivoran remains which formed when the cavities were natural traps 132   (Domingo et al. 2011, 2012, 2013a) whereas the upper strata host more typical 133   herbivore-dominated assemblages, probably formed around ponds that corresponded 134   to the upper part of the cavities as they were filled with sediment. However, at 135   Batallones-3, located in the slope of the Batallones butte, only one stratum is 136   preserved, and it contains the carnivoran-dominated assemblage. It is not known 137   whether an upper stratum existed and if it did, it could have been lost to erosion.. 138   139   . The cavity has a diameter of ~15 metres with an 8-metre-wide, ~1-metre-high debris 140   cone below the vertical entrance. The debris cone is composed of collapsed cavity 141   blocks embedded in a detritic matrix, which hosts plant remains and abundant 142   freshwater fauna (gastropods, river crabs, amphibians and some beaver remains). 143   These materials are inferred to come from the upper part of the hourglass-shaped 144   cavity, which in this case would have acted as a small pond or a permanent water body 145   before roof collapse (Abella et al. 2017). 146   147   Recent studies have confirmed the existence of discrete fossiliferous levels in both the 148   upper (at Batallones-10) and lower parts (at Batallones-3) of the cavities (Martín-Perea 149   et al. 2020). Three discrete fossiliferous levels were found in the lower part of the 150   hourglass structure at Batallones-3 (Figure 2). 151   152   The identification of discrete fossiliferous levels makes it necessary to analyse each of 153   them independently to better characterise the taphonomic history of the site which, in 154   turn, will allow more accurate future palaeoecological and palaeoenvironmental 155   inferences of such similar assemblages. In this work, a detailed taphonomic and 156   comparative study is carried out for each of the levels present at Batallones-3 to 157   understand the biostratinomic and diagenetic processes of this carnivoran-dominated 158   assemblage and to assess whether or not these different assemblages differ in their 159   taphonomy. 160   161   Materials and Methods 162   163   Batallones-3 has undergone numerous excavations, including a campaign in 2001 and 164   continuous annual interventions from 2005 through 2017, ranging in duration from 165   three weeks to six weeks. Standard fossil vertebrate excavation protocols were 166   followed Eberth et al. 2007). The extraction of fossils was documented on-site, 167   including taxonomic features, element identification, mapping of bones (x, y and z 168   coordinates) and the trend and plunge of elongated bones. Unidentified bone 169   fragments and splinters smaller than 4 cm were not mapped. 170   171   Artificial intelligent systems (AI) were used to identify discrete fossiliferous levels within 172   Batallones-3. AI methods followed Martín-Perea et al. (2020) and consisted of: (1) 173   unsupervised Machine Learning for density-based clustering, (2) expert-in-the-loop 174   Collaborative Intelligence Learning for the integration of geological data and (3) 175   supervised learning for fine-tuning of fossiliferous level models. Once the levels had 176   been identified across several sections of the site, supervised Random Forest (RF) and 177   Support Vector Machines (SVM) models were used to classify remaining fossils into 178   their corresponding fossiliferous levels. These algorithms were trained using a 179   traditional machine learning approach, employing k-fold cross-validation (k = 10), a 180   70:30% train:test split, and random searches for hyperparameter optimization. An 80% 181   probability threshold was established for each prediction to determine whether a fossil 182   should be classed as indeterminable (<80%) or not (>80%). In cases where SVM and 183   RF disagree on a class label, the model with the greatest confidence in their prediction 184   was used for the final classification. 185   186   Each fossil was then provided with a final class probability. Overall confidence for the 187   final classified site was established and calculated using robust statistical metrics, 188   reporting the median confidence as well as a +/- margin of error using non-symmetric 189   inter-quartile ranges marking 68.3% of overall variability (Höhle & Höhle 2009). Finally, 190   agreement between the two machine learning algorithms was determined measuring 191   for inter-rater reliability through the calculation of Cohen’s Kappa (k) coefficient (Cohen 192   1960). 193   194   Trend and plunge of elongated fossils in each level were plotted on equal area 195   (Schmidt) stereographic projections with inverse area smoothing. A stereographic 196   projection was plotted for each of 8 theoretical 2 m2 grids around the debris cone for 197   each fossil level. A detailed spatial analysis was carried out using R software (Spatstat 198   library; Baddeley et al. 2015) to represent fossil densities per taxon per level. 199   200   A taphonomical analysis was carried out for each of the identified discrete fossiliferous 201   levels. Number of Identified Specimens (NISP) and Minimum Number of Elements 202   (MNE) were calculated following Lyman (1994). To calculate MNE, age, side and the 203   superposition of anatomical markers have been taken into account. Since NISP may 204   differentially exaggerate sample sizes across taxa (Lyman 2008), Minimum Number of 205   Individuals (MNI) was also calculated. Relative Abundance (Ri) of particular skeletal 206   elements for the dominant taxa (Machairodus aphanistus, Promegantereon ogygia and 207   Indarctos arctoides) were calculated with the formula by Andrews (1990): MNEi Ri   = x  100 MNI  x  Ei 208   where Ri is the relative abundance of element i, MNEi is the minimum number of 209   element I in the sample, MNI is the minimum number of individuals of the taxon, and Ei 210   is the number of times that element i appears in a complete skeleton. In order to 211   determine whether or not there were significant bias in the skeletal representation of 212   different taxa, Kolmogorov-Smirnov tests were performed comparing the observed 213   skeletal representation to the expected skeletal representation according to the MNI 214   estimated for each taxon. 215   216   Age at death of carnivoran individuals was calculated based the eruption and 217   replacement of teeth, degree of wear of occlusal surfaces and bone epiphysiary fusion. 218   These were grouped into juveniles, prime adults and old individuals, following Stiner 219   (1990). 220   221   A systematic examination of bone surface modifications was carried out with an 8X- 222   15X hand lens following Blumenschine (1995). Given the taphonomic homogeneity of 223   the assemblage and the large number of remains, 10% of the elements in level were 224   randomly selected for taphonomic evaluation (479 for Level I, 405 for Level II, 171 for 225   Level III). Although unidentified bone fragments with an elongated axis smaller than 4 226   cm and splinters were not mapped, a random sample of 1004 of these remains was 227   also taphonomically analysed. 228   229   Breakage analyses followed the criteria of Villa & Mahieu (1991) and Lyman (1994). 230   Perpendicular and smooth (dry) fractures often occur in recrystallized or permineralized 231   bones due to diagenetic processes, whereas spiral, irregular and saw-toothed (green) 232   fractures mainly occur in fresh, collagen-rich bones, usually produced by carnivoran or 233   anthropic activity and trampling (Lyman 1994). Bone completeness was analysed 234   following Pesquero et al. (2013) using the following categories: complete, almost 235   complete (bones only missing a chip), more than one-half complete, and one-half 236   complete or less. 237   238   Weathering was estimated according to Behrensmeyer’s (1978) scale from 0 (not 239   weathered, exposed for less than one year before burial) to Stage 5 (extremely 240   weathered, exposed for 15-30 years before burial). Note that other authors, such as 241   Lyman & Fox (1989) have warned that direct correlations between weathering stages 242   and time should be considered with caution as other taphonomic factors 243   (microenvironment, skeletal element, taxon, agents of exposure and burial) might 244   produce deviations. 245   246   Abrasion is indicative of the erosion caused to the remains by means of friction with 247   sedimentary particles. Stages proposed by Alcalá (1994) were used to analyse 248   abrasion: intact bone (Stage 1), rounded bone (Stage 2) and polished and smoothed 249   bone (Stage 3). 250   251   Other surface modifications such as corrosion, digestion, root markings, presence of 252   manganese oxide, trampling, rodent marks and carnivore marks were also recorded, 253   following Fisher (1995) and Fernández-Jalvo & Andrews (2016). Diagnostic criteria 254   defined by Blumenschine (1988, 1995) were used to identify carnivore marks. 255   256   Statistical analyses (chi-square tests) were carried out to compare the three discrete 257   fossiliferous levels at Batallones-3. Additionally, these were also compared to the 258   Batallones-1 assemblage. In order to avoid type I errors, we set our significance level 259   at p = 0.01. In the Supporting Information, we provide the adjusted residuals of those 260   analyses that were statistically significant in order to shed light on the specific variables 261   that contribute more strongly to the observed differences (values outside the two-sigma 262   interval [-1.96, 1.96). 263   264   Results 265   266   Batallones-3 discrete fossiliferous levels 267   268   The three discrete fossiliferous levels previously described at Batallones-3 (Martín- 269   Perea et al. 2020) could be observed throughout the site (Figure 2). Out of the 18,922 270   fossil remains recovered from the excavations, 16,178 had complete coordinate data 271   (x, y and z values). A total of 10,553 fossil remains were classified with over 80% 272   confidence into one of the three fossil levels: 4794 into Level I, 4048 into Level II and 273   1711 into Level III. 274   275   Both machine learning algorithms agreed on 89.2% of classifications, with RF generally 276   considered the most confident. Overall probability of level association revealed models 277   where 99.6 +/- [0.004, 0.006] % confident of the final classifications for the Batallones-3 278   site levels I, II and III. Furthermore, calculations reveal models have a 75-1% likelihood 279   of agreement according to Cohen’s k. This implies substantial agreement between 280   models, thus supporting the use of both RF and SVM for the systematic analyses of 281   this site. 282   283   Assemblage data 284   285   A total of 16 large-mammal species have been identified at Batallones-3, out of which 286   12 are carnivoran species (Table 1). The felids Promegantereon ogygia (NISP = 956, 287   20.13%) and Machairodus aphanistus (NISP = 752, 15.83%) as well as the ursid 288   Indarctos arctoides (NISP = 918, 19.33%) dominate the Level I assemblage (Table 1). 289   Similarly, Level II is dominated by Promegantereon ogygia (NISP = 1117, 27.67%), 290   Machairodus aphanistus (NISP = 844, 20.91%) and Indarctos arctoides (NISP = 517, 291   12.81%; Table 1). The same is true for the Level III assemblage, which is 292   multidominant but with Machairodus aphanistus remains (NISP = 537, 31.44%) being 293   more abundant than Promegantereon ogygia (NISP = 110, 6.44%) and Indarctos 294   arctoides (NISP = 149, 8.72%; Table 1). The remaining carnivores are represented in 295   notably lower proportions. The statistical test points to a significant difference in the 296   carnivoran composition between Level I and Level III (Table 2), with Machairodus 297   aphanistus (more abundant in Level III than in Level I) and Protictitherium crassum 298   (more abundant in Level I than in Level III) having more weight in this comparison (see 299   Supplementary Information). 300   301   Table 3 shows the skeletal element proportions in Batallones-3 levels. Vertebrae 302   constitute the most abundant skeletal element of Level I (15.05%), Level II (15.52%) 303   and Level III (18.09%) assemblages. Moderately abundant elements (over 2% of the 304   assemblage) include incisors, ribs, femurs, metapodials, phalanges, sesamoids, carpal 305   and tarsal bones. A total of 461 (9.62%) remains were anatomically unidentifiable in 306   Level I, 210 (5.17%) in Level II and 91 (5.33%) in Level III (Table 3). 307   308   Skeletal completeness percentages for Machairodus aphanistus, Promegantereon 309   ogygia and Indarctos arctoides are high in Levels I and II (Figures 3A-F), but decrease 310   in Level III (Figures 3G-I). Even though vertebrae constitute the most abundant 311   elements for the total sample (Table 3), vertebrae and ribs are underrepresented for 312   any taxon in any level (Figure 3). 313   314   Spatial data 315   316   Batallones-3 has an excavated area of 160 m2. Bedrock floor has been reached 317   throughout the site, while the lowermost cave walls are also observable (Figure 2). The 318   uppermost section of the cave was lost due to erosion on slopes along the northwest 319   side of the cave(Figure 2). The three fossiliferous levels dip away from the centre to the 320   walls of the cave and rest disconformably over a debris cone of collapsed cave 321   fragments (Figures 2B, 2C). 322   323   Trend and plunge data were represented on several stereographic projections per level 324   per 16 m2 grids around the debris cone (Figure 4). Stereographic projections indicate 325   bones generally dipping very gently (sub-horizontal) towards the northwest or towards 326   the wall of the cave. 327   328   Fossil densities were visualized per level per species in plan view (X-Y plain) using the 329   “spatstat” R package. The number of clusters per level closely resembles the MNI for 330   each species in that given level, as exemplified for Indarctos arctoides remains in 331   Levels I, II and III (Figure 5). 332   333   Bone modification data 334   335   Breakage, along with root etching, is the most pervasive taphonomic alteration in all 336   Batallones-3 fossiliferous levels (Figure 6). Although some bones are fractured, with 337   bones presenting more than 3 fractures making up 37.5% of the Level I assemblage, 338   43.1% of Level II and 43.8% of Level III (Figures 6A, B and C), the majority of remains 339   are complete: 84.4% complete bones in Level I, 76.4% in Level II and 53.4% in Level III 340   (Figures 6D, E and F). These fractures appear in different angles, types (Figure 6G, H 341   and I) and most have smooth fracture plane surfaces: 63.4% of fractures in Level I, 342   58.0%in Level II and 67.9% of fractures in Level III (Figures 6J, K and L). In general, 343   there is homogeneity among Batallones-3 levels when the fracture categories are 344   statistically compared. Nevertheless, as for bone completeness, there are significant 345   statistical differences between Level I and Level III, and between Level II and Level III 346   (Table 2). In Level III, bones that are less than one half complete are significantly more 347   abundant than in Level I and Level II, where complete bones are more abundant 348   (Figure 6; Supporting Information). 349   350   Unweathered bones (Stage 0) dominate in all Batallones-3 levels (Table 4). 351   Weathering Stage 1 is only observed in 0.6% of Level III fossil remains, whereas for 352   unidentified bones with no assigned level it is slightly higher (2.5%). Only three 353   unidentified remains (0.3 %) showed Weathering Stage 2. No bones displayed 354   Weathering Stages 3 or higher. Abrasion follows a similar pattern with a predominance 355   of Abrasion Stage 1 (intact, non-abraded bone; Table 4). Abrasion Stage 2 only affects 356   5.3% of the remains in Level III and 7.7% in unidentified remains whereas Stage 3 is 357   only observed in a few unidentified remains with no assigned level (0.4%). There are 358   no statistically significant differences between levels for weathering or abrasion (Table 359   2). 360   361   Root marks are common throughout Batallones-3 levels: 43.8% of remains are root 362   marked in Level I, 59.7% in Level II and 64.4% in Level III (Table 4). Trampling was 363   only observed in 1.6% of the unidentifiable remains, none of which were assigned to a 364   level. There is no evidence of rodent or carnivore marks or digestion alteration in any 365   Batallones-3 levels. 366   367   The vast majority of analysed remains exhibited manganese oxide precipitations: 368   73.5% in Level I, 84.7% in Level II and 91.8% in Level III. Manganese oxides are 369   common along fractures and widespread on bone surfaces. 370   371   Discussion 372   373   Using traditional methods, which often make it difficult to recognize discrete horizons, a 374   taphonomy study would have been conducted on the entire deposit, obfuscating 375   potential differences through the time averaged assemblage. This problem was 376   minimized using AI, which identified three discrete fossil levels at Batallones-3 and 377   assigned 65% of the fossil remains into these levels. This enabled a more accurate 378   understanding of the environmental history of the site. 379   380   The carnivoran guilds in all three Batallones-3 levels, which includes 12 species, is the 381   most diverse of the Cerro de los Batallones complex. Moreover, carnivoran species 382   dominate the assemblage, which, as has previously been stated, is uncommon in the 383   fossil record. Our taxonomic study of the Batallones-3 fossil material provides 384   unprecedented information on such rare carnivoran-dominated assemblages. The 385   similarities between the three levels indicate (1) all were subject to the same 386   taphonomic conditions and (2) indicates taphonomic conditions did not skew 387   assemblage data. 388   389   Batallones-3 Levels I, II and III comprise multitaxic and multidominant assemblages, 390   dominated by the sabretooth felids Promegantereon ogygia and Machairodus 391   aphanistus and the ursid Indarctos arctoides (Table 1). Statistical analyses of the MNI 392   of the carnivoran species composition of the three levels reveal that Levels I and II 393   exhibit a similar carnivoran composition, which differ from Level III (Table 2; Supporting 394   Information). 395   396   In Levels I and II, which are deeper and therefore older, Promegantereon ogygia is 397   more abundant than the other sabretooth felid, Machairodus aphanistus, which in turn 398   becomes the dominant carnivoran taxon in the youngest level, Level III. Other trends in 399   the carnivoran guild involve a decrease in the hyaenid Protictitherium crassum and the 400   mustelid Eomellivora piveteaui and a slight increase in the two amphicyonid species. 401   The ursid Indarctos arctoides does not show a clear trend: its MNI decreases from 402   Level I to Level II, the increases in Level III. 403   404   When comparing Batallones-3 with Batallones-1 in terms of carnivoran faunal 405   composition, statistically significant differences emerge related to the 406   presence/absence of some of the carnivoran species, as revealed by the values of the 407   adjusted residuals of the chi-square analysis (Supporting Information). These 408   differences were previously discussed by Domingo et al. (2016) and mainly pertain to 409   the presence of the ailurid Simocyon batalleri at Batallones-1 and its absence at 410   Batallones-3, and the presence of the ursid Indarctos arctoides and the mustelid 411   Eomellivora piveteaui at Batallones-3 and its absence from Batallones-1. Evaluation of 412   the residuals also reveals differences in the abundance of Magericyon anceps, which is 413   greater at Batallones-1 than in any Batallones-3 levels. Apart from the aforementioned 414   absences of taxa, the adjusted residuals also show that the frequencies of 415   Promegantereon ogygia (higher at Batallones-1), and Thaumastocyon sp. (greater at 416   Batallones-3 Level III), play an important role in the statistically significant differences 417   between Batallones-3 Level III and Batallones-1 (Supporting Information). 418   419   As indicated by Domingo et al. (2016), these species abundance differences could 420   represent biological trends, such as diversity turnover or changes in population 421   densities in the Madrid Basin over time. The accumulation of the Batallones-1 and 422   Batallones-3 assemblages was asynchronous; they were deposited in the following 423   order, from oldest to youngest: Batallones-1, Batallones-3 Level I, Batallones-3 Level II 424   and Batallones-3 Level III (López-Antoñanzas et al. 2010), and our analyses indicate 425   that Machairodus aphanistus, Indarctos arctoides and Thaumastocyon sp. became 426   more abundant through time, whereas the abundance of individuals of 427   Promegantereon ogygia, Protictitherium crassum and Simocyon batalleri decreased. 428   Further light will be shed on these trends in future studies, which will consider not only 429   Batallones-1 and Batallones-3, but other Cerro de los Batallones localitiesonce they 430   are excavated and studied. Furthermore, the identification of discrete fossiliferous 431   levels (following Martín-Perea et al. 2020) within the other Cerro de los Batallones 432   localities will provide more information on faunal compositional changes. 433   434   The low percentages of anatomically unidentified remains in Batallones-3 (9.62% in 435   Level I, 5.17% in Level II and 5.33% in Level III) are indicative of the good state of 436   preservation of these fossils. Although the remains are fractured (Figures 6A, B and C), 437   many of them are complete (Figures 6D, E and F) and well preserved (Figure 7). 438   Interestingly, there are statistically significant differences of bone completeness in the 439   Batallones-3 levels, between Level I and Level III and between Level II and Level III 440   (Table 2; Supporting Information). The bones are most complete in Levels I and II, 441   whereas in Level III 30.7% of the bones are one-half or less complete. Level III, the 442   highest level, was very close to the current surface and damage may be related to 443   rooting and slope erosion. The fractures at Batallones-3 are mostly perpendicular in 444   long bones (Figures 6G, H and I) and smooth (63.4% in Level I, 58.0% in Level II and 445   67.9% in Level III; Figures 6J, K and L), which is indicative of diagenetic breakage 446   (Figure 7, Figure 8B). The bone completeness at all three Batallones-3 levels 447   significantly differ from that at Batallones-1 (Table 2). This difference, on comparing 448   Batallones-3 Level I and Level II with Batallones-1, is due to the higher abundance of 449   complete bones, together with a lower abundance of almost complete bones at 450   Batallones-3 compared to Batallones-1 (Supporting Information). In any case, the 451   bones present a high degree of integrity, as complete and almost complete bones are 452   prevalent. Nevertheless, when comparing Batallones-3 Level III with Batallones-1, the 453   primary difference is the abundance of bones which are one-half or less complete at 454   Batallones-3 Level III in comparison with Batallones-1. 455   456   Although the most abundant skeletal elements are vertebrae (15.05% in Level I, 457   15.52% in Level II and 18.09% in Level III), all other skeletal elements are represented 458   (Table 3). The very low representation of vertebrae and ribs for all taxa (Figure 3) could 459   be due to the difficulty in assigning these to a particular species, given the lack of 460   diagnostic characters. Nonetheless, the percentages of skeletal completeness for 461   Machairodus aphanistus, Promegantereon ogygia and Indarctos arctoides are 462   generally high (Figure 3), and Kolmogorov-Smirnov tests indicate that there is no 463   significant anatomical bias relative to complete skeletons (Table 5). These results 464   suggest individuals were complete in the thanatocoenosis and taphocoenosis. 465   466   Furthermore, the fossil remains are found in monotaxic clusters, which when 467   represented per level, show approximately the same number of clusters as of MNI 468   (Figure 5). This appears to indicate that the remains have undergone little, if any, 469   biostratinomic or fossildiagenetic displacement. Conversely, the trend data for elongate 470   remains show that although they were not significantly displaced, they were slightly 471   reoriented (Figure 4). This partial and slight reorientation likely took place during the 472   episodic sedimentary infilling of the cavity, since the mean trend and plunges of the 473   fossil remains coincide with the inferred entry of alluvial sediments presenting a north- 474   western direction, following the slope dip of the Batallones butte rim (Figure 1). 475   476   Given the stable humidity and temperature levels in caves, as well as the lack of direct 477   sunlight, bones can remain unburied long periods of time without undergoing heavy 478   weathering (Wang & Martin 1993). For this reason, estimates of time before burial 479   according to weathering stages (Behrensmeyer 1978) are not applicable in this 480   scenario. Most of the bones at Batallones-3 are unweathered (Weathering Stage 0), 481   which is accounted for by the protective environment of the chamber. From a statistical 482   viewpoint, the three levels are similar in terms of weathering (Table 2). The same is 483   true when compared to Batallones-1, where unweathered bones were also prevalent 484   (Domingo et al. 2013a). Abrasion data imply that most of the bones were not 485   significantly impacted by sedimentary particles or by each other in any of the 486   Batallones-3 assemblages, a circumstance also observed at Batallones-1 (Tables 2 487   and 4). The prevalence of unabraded bones, alongside the monotaxic clusters, along 488   with the associated and articulated complete or partial skeletons (Figure 8), reinforces 489   the interpretation of an autochthonous taphocoenosis (Fernández-López 1990), 490   maintaining its in situ properties. 491   492   Bones displaying trampling, higher degrees of weathering or abrasion are very scarce 493   (Table 4), corresponding either to unidentified remains or to herbivore mammal species 494   (Figures 7F and G). These elements, which in most cases are not classified into any 495   discrete fossiliferous level, likely correspond to allochthonous taphocoenoses, 496   transported and deposited along with sediment into the cavity. In the case of the 497   herbivore remains, this allochthony appears to be reinforced by the almost total lack of 498   remains. 499   500   Plant roots constitute common biological agents that can biochemically damage bone 501   (Sarjeant 1975; Behrensmeyer 1978; Mikulás 1999, 2001; Fernández-Jalvo et al. 2002; 502   Tjelldén et al. 2016; Paes-Neto et al. 2018). These types of alterations are well known 503   in both archaeological and palaeontological contexts (e.g. Mikulás 2001; Fernández- 504   Jalvo & Andrews 2016). Root marks are common at Batallones-3 on the surface of 505   teeth (Figure 7B) and bones (Figure 7C). Root marks are present in 43.8% of the 506   remains from Level I, 59.7% from Level II and 64.4% from Level III (Table 4). This 507   increasing prevalence of root marks towards the shallow level of Batallones-3 is 508   evidenced by the statistical analysis: Level III shows a significant higher abundance of 509   root-marked bones than Level I (Table 2; Supporting Information); and Level II shows a 510   statistically marginal greater abundance (p = 0.02) of root-marked bones than Level I. 511   This points to a progressive increase in root-marking in the shallower assemblage of 512   Batallones-3. Although it is difficult to assess when these modifications took place, the 513   most plausible scenario is that these are present-day root marks, mainly the result of 514   artificial pine tree reforestation in the area sometime between 1994 and 2000 (damage 515   to the bones by modern roots was observed throughout the Batallones-3 excavations), 516   and to a lesser extent, of the effect of natural shrubs on the butte slopes before and 517   during this period. The three levels from Batallones-3 show significantly higher 518   abundances of bones with root marks than Batallones-1 (Table 2; Supporting 519   Information). As they are located on the butte slope, modern soil comes into contact 520   with the Batallones-3 fossiliferous deposits, whereas Batallones-1 is located towards 521   the centre of the butte and, therefore, too deep to be subjected to present-day edaphic 522   processes, including plant root activity. 523   524   Significant differences in the percentages of remains with manganese oxide coatings 525   appear to be associated with the abundance of root marked specimens, as they follow 526   the same trend (Tables 2 and 4). It has been suggested that biological processes and 527   soil humification expedite precipitation of manganese oxides on fossil remains (e.g. 528   Shahack-Gross et al. 1997; Marín-Arroyo et al. 2008). Statistically significant 529   differences are observed between all three Batallones-3 levels and Batallones-1 in 530   terms of the presence of bones presenting manganese oxides, which are much more 531   abundant at Batallones-3. Once again, this might be because the Batallones-1 fossil 532   bones are better protected from pedogenic processes, due to their deeper position in 533   the butte. 534   535   There are no carnivore-marked bones at Batallones-3. The pervasiveness of root 536   marks could be affecting other taphonomic modifications or even hindering correct 537   taxonomical or anatomical identification. Such could be the case for carnivore marks, 538   which might be obscured by root etching. Nonetheless, and given that at Batallones-1, 539   the other carnivoran-dominated assemblage from Cerro de los Batallones, carnivore 540   marks are practically non-existent (0.30%), we believe that there are few tooth marks 541   on the bones at Batallones-3, and that this is not an artefact of other superimposed 542   taphonomic processes. Although it might seem puzzling that a carnivoran-dominated 543   site displays such a low number of tooth-marked bones, different hypotheses can 544   account for this. For example, taphonomic analyses performed on felid-consumed 545   carcasses conclude that very few marks are left (Domínguez-Rodrigo & Barba 2006; 546   Domínguez-Rodrigo et al. 2007; Gidna et al. 2014; Parkinson et al. 2015; Organista et 547   al. 2016), which could explain the extremely low percentages of carnivore marks in the 548   Cerro de los Batallones sites, where sabre-toothed cats are the dominant carnivoran 549   species. In their study of the Batallones-1 assemblage, Domingo et al. (2013a) also 550   suggested that the low incidence of carnivore marks could be the result of a low level of 551   exploitation of carcasses because there was abundant supply of meat or because 552   carnivorans died quickly after entrapment (from exhaustion, hypothermia, illness, toxic 553   gases, poisoning from drinking water or injuries after entrance). Also, it is possible that 554   carnivores were searching for water during droughts, as opposed to just food. 555   556   Finally, trampling marks are very scarce at Batallones-3. Again, root marks might have 557   obscured evidence in this sense. However, trampling is also extremely scarce at 558   Batallones-1, which suggests trapped carnivorans did not move a lot inside the 559   chamber (due to the above mentioned conditions - exhaustion, injuries or illness). 560   561   Given the inference that the carnivoran-dominated assemblages in Levels I, II and III at 562   Batallones-3 are autochthonous, with a few scattered allochthonous remains, there are 563   only two likely scenarios for the formation of these deposits: these were created either 564   by denning activities or by the cavity acting as a natural trap. As for the hypothesis that 565   Batallones-3 acted as a den for carnivoran species, from a taxonomical point of view, 566   prey species (generally, herbivores) dominate bone assemblages in these scenarios. 567   Carnivoran MNI’s in den assemblages (from 6% to 53%; Brain 1981; Wang & Martin 568   1993; Marean & Ehrhardt 1995; Pokines & Kerbis Peterhans 2007) differ greatly from 569   the overwhelming majority of carnivoran remains in all Batallones-3 levels. 570   571   The geomorphology of the cavity in Batallones-3, exhibiting a vertical hourglass shape, 572   is inferred to only have one entrance in the middle of the cave roof, directly >3 metres 573   above the debris cone (Calvo et al. 2013; Martín-Perea et al. 2017). This opening likely 574   presented difficult entry and hindered exit. The absence of tooth marks or spiral 575   fractured bones resulting from consumption, supports the hypothesis that the 576   assemblages were not produced as a consequence of denning activities. 577   578   Ursids, such as Ursus deningeri and Ursus spelaeus, are one of the most common 579   dwellers of Pleistocene and Holocene caves, which usually result in fossil assemblages 580   abundant in carnivoran (mainly ursid) remains. Since these assemblages are frequently 581   found alongside human remains or artefacts, archaeologists have paid special attention 582   to this interaction. Study of data on ursid mortality is of particular interest with regard to 583   the taphonomic history of cave-bear assemblages (Webb 1988; Gargett 1996; Stiner et 584   al. 1996; García et al. 1997; Wolverton 2001, 2006) and to establishing whether it more 585   closely resembles a hibernation death assemblage (Stiner et al. 1996; Stiner 1998) or a 586   natural trap (Wolverton 2001, 2006). 587   588   Although Ailuropodinae ursids are not known to hibernate, Indarctos arctoides mortality 589   profiles (Figure 9) more closely resemble ursid prime-dominated natural-trap 590   assemblages (García et al. 1997; Wolverton 2001, 2006), compared to living-structure 591   assemblages resulting from natural hibernation deaths (Stiner 1998; Wolverton 2006). 592   Claw marks, which can be interpreted as resulting from the action of trapped live 593   animals, have been documented alongside ursid assemblages (Bednarik 1994, 2004). 594   These marks are normally found in karstic cavities with hard limestone bedrock walls. 595   In turn, the peculiar pseudokarstic formation of the Cerro de los Batallones cavities, 596   carved into a sepiolite-rich marl bedrock (Pozo et al. 2004; Calvo et al. 2013), would 597   preclude the formation of these claw markings. 598   599   Faunal concentrations in natural traps in caves can be either accidentally produced by 600   falling (passive accumulation) or intentional through purposefully entering these 601   cavities (active accumulation, Lyman 1994). Accidental passive accumulations are 602   unselective, both taxonomically and ontogenetically, and will result in a diverse, 603   unbiased assemblage (Wang & Martin 1993). This is clearly not the case for any of the 604   Batallones-3 fossiliferous levels, which are dominated by prime adult carnivorans, 605   presenting very few herbivore remains (Table 1, Figure 9). Consequently, a recurring, 606   intentional entrance of carnivores into the cavity is the most likely scenario (Figure 10). 607   Since most of the herbivore remains are allochthonous, it is safe to assume carnivores 608   were not lured into the cavity in order to scavenge herbivore remains, but rather were 609   searching for water. The absence of carnivore marks on the carnivoran remains 610   suggests these were not scavenged either. 611   612   Furthermore, as for the moment of death of the trapped animals, it is difficult to infer 613   whether these died as a consequence of the fall or a posteriori for other reasons. 614   Although the absence of trampling could suggest otherwise, the geomorphology of the 615   cave, with the opening inferred to be >3 meters above the ground, would not entail a 616   mortal fall. The absence of fragmented bones which would suggest falling-related 617   injuries, reinforces this idea. Other causes of death such as exhaustion or hypothermia, 618   as explained above, appear to constitute more logical options. Moreover, it seems 619   counterintuitive that carnivores would risk their lives intentionally to access the cavity. 620   The vertical distance between the cave opening and the cave floor would appear less 621   high due to the presence of the 1-metre-high debris cone beneath the cave entrance, 622   which would facilitate entry. However, once the carnivores intentionally entered the 623   cavity, they would be unable to escape due to the inward-slanting cave walls and the 624   insurmountable >3-meter-high cave opening. 625   626   The presence of the giant tortoise Titanochelon bolivari (Pérez-García & Vlachos 2014) 627   in all three levels is not surprising given this testudinid is exceptionally frequent in 628   continental facies in the Madrid Basin. It is uncertain, however, whether the remains of 629   this testudinid at Batallones-3 are autochthonous or allochthonous. On one hand, giant 630   tortoises could have actively and intentionally entered the cavity attracted to water, 631   looking for a place to breed or to bathe in mud in order to regulate body temperature, 632   as observed in extant giant tortoises (Walton et al. 2019). Although it seems 633   implausible that tortoises jumped into the cave, the phreatic level might have risen 634   intermittently, sufficing to partially fill the cavity, granting easy access to the tortoises 635   but difficult subsequent egress. These animals, unable to exit the partially filled cavity, 636   would have died of exhaustion. Well preserved (unabraded, unweathered) remains, 637   some of these articulated or associated, and the presence of these throughout all three 638   fossiliferous levels, makes this scenario feasible. It cannot, however, be ruled out that 639   the Titanochelon remains might be allochthonous, deriving from the older, freshwater 640   assemblage before the cave opening was formed (Figure 10A). 641   642   The few taphonomic differences between the Batallones-3 levels, and also between 643   Batallones-3 and Batallones-1, are related to post-mortem, pedogenic alterations, 644   which are possibly modern, such as root markings, manganese oxide precipitation and 645   a slightly lower bone completeness in Level III. The homogeneity of taphonomic 646   properties between levels and even between Batallones-1 and Batallones-3 (Table 2) 647   suggests recurring taphonomic and geologic processes throughout the infill process of 648   the cave, which implied the successive intentional entry of carnivorans and their 649   subsequent entrapment (Figure 10). Given that almost one third of the carnivoran 650   species known for this time period (~9.8 to 8.8 Ma) are exclusively from the Cerro de 651   los Batallones sites, there is a vital need for future palaeoecological and 652   palaeoenvironmental studies of each of the Batallones-3 levels. The present study 653   ensures that these studies can be conducted without risk of differential taphonomic 654   histories skewing the assemblage data. 655   656   Conclusions 657   658   The Batallones-3 site (Madrid Basin, Spain) hosts an unusual fossil assemblage, both 659   because of its location in a pseudokarstic cavity and its carnivoran-dominated 660   composition. Artificial intelligence systems enabled us to identify three fossiliferous 661   levels embedded within the massive Batallones-3 deposit. 662   663   Batallones-3 Levels I and II are multitaxic multidominant assemblages containing two 664   sabretooth cats, Promegantereon ogygia and Machairodus aphanistus, and an ursid, 665   Indarctos arctoides, which is the most abundant species. On the other hand, 666   Batallones-3 Level III is also a multitaxic multidominant assemblage, although it is 667   primarily dominated by Machairodus aphanistus. Prime adults dominate all three 668   assemblages. Statistically significant differences in the carnivoran faunal composition 669   may be a function of faunal changes in the populations of carnivoran species 670   throughout the accumulation of these assemblages. 671   672   These carnivoran-dominated assemblages are autochthonous, with diagenetically 673   fractured but well preserved remains (complete, little to no weathering or abrasion), 674   presenting abundant root marks and manganese oxide precipitation, with a very small 675   amount of allochthonous, badly preserved (weathered and abraded) herbivore remains. 676   Statistical analyses reveal homogeneity in most of the taphonomic variables of the 677   assemblages. A trend can be seen, however, which is associated with the higher 678   prevalence of pedogenic and erosional slope processes (more root marks, manganese 679   oxide precipitates and bone incompleteness) in the shallower Batallones-3 Level III. 680   681   We propose that all three levels were mainly formed by the active entrance of 682   carnivorans into the cavity. These were likely attracted by water and subsequently 683   became trapped. Many of the taphonomic characteristics of Batallones-3, including the 684   prevalence of carnivorans, coincide with those observed in the previously studied 685   neighbouring Batallones-1 site. Indeed, both sites present a similar formation scenario. 686   Future palaeoecological and palaeoenvironmental studies can therefore be conducted 687   in each level with the assurance that differential taphonomic histories have not biased 688   the assemblage data; this would provide a better understanding of the diversity and 689   ecology of these important Late Miocene carnivoran faunas. 690   691   Acknowledgements 692   693   We thank everyone involved in the excavation campaigns carried out at Batallones-3. 694   Excavations carried out in the Cerro de los Batallones are funded by the Dirección 695   General de Patrimonio Cultural (Comunidad de Madrid). David M. Martín-Perea was 696   funded by an FPI Predoctoral grant BES-2016-079460 from the Spanish Government 697   associated with Projects CGL2015-6833-P and PGC2018-094122-B-100. Alberto 698   Valenciano was funded by a “Juan de la Cierva Formación” program (FJC2018- 699   036669-I) from the Spanish Ministry of Science, Innovation and Universities and by the 700   Aragon Government (E33_17R). Juan Abella was funded by the Generalitat de 701   Catalunya (CERCA Programme and Beatriu de Pinós) contract 2017-BP-00223 from 702   AGAUR. This study is also a contribution of Projects PGC2018-094122-B100 703   (AEI/FEDER, UE), CGL2016-76431-P (AEI/FEDER, UE) and the research groups 704   CSIC-64-1538 and CAM-UCM-910607. The authors declare no conflict of interest. The 705   authors thank Antonio Rodríguez-Hidalgo, an anonymous reviewer and Associate 706   Editor Brooks Britt, whose helpful suggestions greatly improved this manuscript. 707   708   References 709   710   Abella, J., Alba, D.M., Robles, J.M., Valenciano, A., Rotgers, C., Carmona, R., 711   Montoya, P. & Morales., J. 2012: Kretzoiarctos gen. nov. the oldest member of the 712   giant panda clade. PLoS ONE 7, e48985. 713   714   Abella, J., Valenciano, A., Pérez-Ramos, A., Montoya, P. & Morales, J. 2013: On the 715   Socio-Sexual Behaviour of the Extinct Ursid Indarctos arctoides: An approach based 716   on its baculum size and morphology. PLoS ONE 8, e73711. 717   718   Abella, J., Pérez-Ramos, A., Valenciano, A., Alba, D.M., Ercoli, M.D., Hontecillas, D., 719   Montoya, P. & Morales, J. 2015: Tracing the origin of the panda’s thumb. The Science 720   of Nature 102, 1-13. 721   722   Abella, J. & Valenciano, A. 2017: Indarctos arctoides: Los ancestros de los grandes 723   osos panda. In Morales, J. (ed.), La Colina de los Tigres Dientes de Sable. Los 724   yacimientos miocenos del Cerro de los Batallones (Torrejón de Velasco, Comunidad 725   de Madrid), 302-316. Museo Arqueológico Regional, Madrid, Spain. 726   727   Abella, J., Valenciano, A., Martín-Perea, D.M., Montoya, P. & Morales, J. 2017: 728   Batallones-3 un paraíso de diversidad en carnívoros. In Morales, J. (ed.), La Colina de 729   los Tigres Dientes de Sable. Los yacimientos miocenos del Cerro de los Batallones 730   (Torrejón de Velasco, Comunidad de Madrid), 278-296. Museo Arqueológico Regional, 731   Madrid, Spain. 732   733   Alberdi, M.T., de la Iglesia, A., Montoya, P. & Morales, J. 2017: La ciénaga de los 734   mastodontes. In Morales, J. (ed.), La Colina de los Tigres Dientes de Sable. Los 735   yacimientos miocenos del Cerro de los Batallones (Torrejón de Velasco, Comunidad 736   de Madrid), 395-410. Museo Arqueológico Regional, Madrid, Spain. 737   738   Alcalá, L. 1994: Macromamíferos Neógenos de la Fosa de Alfambra-Teruel. Instituto 739   de Estudios Turolenses 554. 740   741   Álvarez-Sierra, M.A., García-Paredes, I., Gómez Cano, A.R., Hernández-Ballarín, V., 742   van den Hoek Ostende, L.W., López-Antoñanzas, R., López-Guerrero, P., Oliver, A. & 743   Peláez-Campomanes, P. 2017: Los micromamíferos del Cerro de los Batallones. In 744   Morales, J. (ed.), La Colina de los Tigres Dientes de Sable. Los yacimientos miocenos 745   del Cerro de los Batallones (Torrejón de Velasco, Comunidad de Madrid), 516-529. 746   Museo Arqueológico Regional, Madrid, Spain. 747   748   Andrews, P. 1990: Owls, caves and fossils. Chicago, The University of Chicago Press. 749   750   Antón, M., Salesa, M.J., Morales, J. & Turner, A. 2004: First known complete skulls of 751   the scimitar-toothed cat Machairodus aphanistus (Felidae, Carnivora) from the Spanish 752   late Miocene site of Batallones-1. Journal of Vertebrate Paleontology 24, 957-969. 753   754   Antón, M., Salesa, M.J., Pastor, J.F., Peigné, S. & Morales, J. 2006: Implications of the 755   functional anatomy of the hand and forearm of Ailurus fulgens (Carnivora, Ailuridae) for 756   the evolution of the “false-thumb” in pandas. Journal of Anatomy 209, 757-764. 757   758   Antón, M., Siliceo, G., Pastor, J.F., Morales, J. & Salesa, M.J. 2020: The early 759   evolution of the sabre-toothed felid killing bite: the significance of the cervical 760   morphology of Machairodus aphanistus (Carnivora: Felidae: Machairodontinae). 761   Zoological Journal of the Linnean Society 188, 319-342. 762   763   Arsuaga, J.L., Martínez, I., Gracia, A., Carretero, J.M., Lorenzo, C., García, N. & 764   Ortega, A.I. 1997: Sima de los Huesos (Sierra de Atapuerca, Spain): The site. Journal 765   of Human Evolution 33, 109-127. 766   767   Baddeley, A., Rubak, E. & Turner, R. 2015: Spatial Point Patterns: Methodology and 768   applications with R. Chapman and Hall/CRC Press. 769   770   Bednarik, R.G. 1994: Wall markings of the cave bear. Studies in Speleology 9, 51-70. 771   772   Bednarik, R.G. 2004: Rock markings of humans and other animals. Rock art research 773   21, 57-84. 774   775   Behrensmeyer, A.K. 1978: Taphonomic and ecologic information from bone 776   weathering. Paleobiology 4, 140-158. 777   778   Binder, W.J., Thompson, E.N. & van Valkenburgh, B. 2002: Temporal variation in tooth 779   fracture among Rancho La Brea dire wolves. Journal of Vertebrate Paleontology 22, 780   423-428. 781   782   Blumenschine, R.J. 1988: An experimental model of the timing of hominid and 783   carnivore influence on archaeological bone assemblages. Journal of Archaeological 784   Science 15, 483-502. 785   786   Blumenschine, R.J. 1995: Percussion marks, tooth marks and the experimental 787   determinations of the timing of hominid and carnivore access to long bones at FLK 788   Zinjanthropus, Olduvai Gorge, Tanzania. Journal of Human Evolution 29, 21-51. 789   790   Calvo, J.P., Pozo, M., Silva, P.G. & Morales, J. 2013: Pattern of sedimentary infilling of 791   fossil mammal traps formed in pseudokarst at Cerro de los Batallones, Madrid Basin, 792   central Spain. Sedimentology 60, 1681-1708. 793   794   Chomko, S. & Gilbert, B.M. 1987: The late Pleistocene/Holocene Faunal Record in the 795   Northern Bighorn Mountains, Wyoming. In Graham, R.W., Semken, H.A. & Graham, 796   M.A. (eds.), Late Quaternary Mammalian Biogeography and Environments of the Great 797   Plains and Prairies, 21-35. Illinois State Museum, Springfield. 798   799   Cohen, J. 1960: A coefficient of agreement for nominal scales. Educational and 800   Psychological Measurement 20, 37-46. 801   802   Domingo, M.S., Alberdi, M.T. & Azanza, B. 2007: A new quantitative biochronological 803   ordination for the Upper Neogene mammalian localities of Spain. Palaeogeography, 804   Palaeoclimatology, Palaeoecology 255, 361-376. 805   806   Domingo, M.S., Domingo, L., Sánchez, I.R., Alberdi, M.T., Azanza, B. & Morales, J. 807   2011: New insights on the taphonomy of the exceptional mammalian fossil sites of 808   Cerro de los Batallones (Late Miocene, Spain) based on rare earth element 809   geochemistry. Palaios 26, 55-65. 810   811   Domingo, M.S., Alberdi, M.T., Azanza, B. & Morales, J. 2012: Mortality patterns and 812   skeletal physical condition of the carnivorans from the Miocene assemblage of 813   Batallones-1 (Madrid Basin, Spain). Neues Jahrbuch fur Geologie und Palaeontologie- 814   Abhandlungen 265, 131-145. 815   816   Domingo, M.S., Alberdi, M.T., Azanza, B., Silva, P.G. & Morales, J. 2013a: Origin of an 817   assemblage massively dominated by carnivorans from the Miocene of Spain. PLOS 818   ONE 8, e63046. 819   820   Domingo, M.S., Domingo, L., Badgley, C., Sanisidro, O. & Morales, J. 2013b: Resource 821   partitioning among top predators in a Miocene food web. Proceedings of the Royal 822   SocietyB: Biological Sciences 280, 20122138. 823   824   Domingo, M.S., Domingo, L., Abella, J., Valenciano, A., Badgley, C. & Morales, J. 825   2016: Feeding ecology and habitat preferences of top predators from two Miocene 826   carnivore-rich assemblages. Paleobiology 42, 489-507. 827   828   Domingo, M.S., Cantero, E., García-Real, I., Chamorro Sancho, M.J., Martín-Perea, 829   D.M., Alberdi, M.T. & Morales, J. 2018: First radiological study of a complete dental 830   ontogeny sequence of an extinct equid: implications for equidae life history and 831   taphonomy. Scientific reports 8, 1-11. 832   833   Domínguez-Rodrigo, M. & Barba, R. 2006: New estimates of tooth mark and 834   percussion mark frequencies at the FLK Zinj site: the carnivore-hominid-carnivore 835   hypothesis falsified. Journal of Human Evolution 50, 170-194. 836   837   Domínguez-Rodrigo, M., Egeland, C.P. & Pickering, T.R. 2007: Equifinality in carnivore 838   tooth marks and the extended concept of archaeological palimpsests: implications for 839   models of passive scavenging in early hominids. Breathing life into fossils: taphonomic 840   studies in honor of CK (Bob) Brain, 255-267. 841   842   Eberth, D.A., Rogers, R.R. & Fiorillo, A.R. 2007. A practical approach to the study of 843   bonebeds. In Rogers, R.R., Eberth, D.A. & Fiorillo, A.R. (eds.), Bonebeds: Genesis, 844   analysis and paleobiological significance, 265-331. Chicago: The University of Chicago 845   Press. 846   847   Eisenberg, J. 1981: The Mammalian Radiation. Chicago, The University of Chicago 848   Press. 849   850   Farlow, J.O. 1993: On the rareness of big, fierce animals: speculations about the body 851   sizes, population densities, and geographic ranges of predatory mammals and large 852   carnivorous dinosaurs. American Journal of Science 293, 167-199. 853   854   Feranec, R.S., Hadly, E.A. & Paytan, A. 2009. Stable isotopes reveal seasonal 855   competition for resources between late Pleistocene bison (Bison) and horse (Equus) 856   from Rancho La Brea, southern California. Palaeogeography, Palaeoclimatology, 857   Palaeoecology 271, 153-160. 858   859   Fernández-Jalvo, Y., Sánchez-Chillón, B., Andrews, P. & Alcalá-Martínez, L. 2002: 860   Morphological taphonomic transformations of fossil bones in continental environments, 861   and repercussions on their chemical composition. Archaeometry 44, 353-361. 862   863   Fernández-Jalvo, Y. & Andrews, P. 2016: Atlas of taphonomic identifications: 1001+ 864   Images of Fossil and Recent Mammal Bone Modification. Springer, London. 865   866   Fernández-López, S. 1990: El significado de la autoctonía/aloctonía tafonómica. 867   Reunión de Tafonomía y Fosilización, 115-124. 868   869   Fisher, J.W. 1995: Bone surface modifications in zooarchaeology. Journal of 870   Archaeological Method and Theory 2, 7-68. 871   872   Fraile, S. 2016: Estudio de Protictitherium crassum del Cerro de los Batallones 873   (Torrejón de Velasco, Madrid): aportación a la filogenia y evolución de la familia 874   hyaenidae. PhD Thesis, Complutense University of Madrid, Spain. 875   876   Fraile, S. 2017: Protictitherium crassum, la pequeña hiena de 9 millones de años del 877   Cerro de los Batallones. In Morales, J. (ed.), La Colina de los Tigres Dientes de Sable. 878   Los yacimientos miocenos del Cerro de los Batallones (Torrejón de Velasco, 879   Comunidad de Madrid), 224-234. Museo Arqueológico Regional, Madrid, Spain. 880   881   García, N., Arsuaga, J.L. & Torres, T. 1997: The carnivore remains from the Sime de 882   los Huesos Middle Pleistocene site (Sierra de Atapuerca, Spain). Journal of Human 883   Evolution 33, 155-174. 884   885   García, N. & Arsuaga, J.L. 1998: The carnivore remains from the hominid-bearing 886   Trinchera-Galería, Sierra de Atapuerca, Middle Pleistocene site (Spain). Geobios 31, 887   659-674. 888   889   Gidna, A.O., Kisui, B., Mabulla, A., Musiba, C. & Domínguez-Rodrigo, M. 2014: An 890   ecological neo-taphonomic study of carcass consumption by lions in Tarangire National 891   Park (Tanzania) and its relevance for human evolutionary biology. Quaternary 892   International 322, 167-180. 893   894   Gilbert, B.M. 1978: Natural Trap Cave. Wyoming Wildlife 42, 10-13. 895   896   Gómez-Cano, A.R., Hernández Fernández, M. & Álvarez-Sierra, M.A. 2011: 897   Biogeographic provincialism in rodent faunas from the Iberoccitanian Region 898   (southwestern Europe) generates severe diachrony within the Mammalian Neogene 899   (MN) biochronologic scale during the Late Miocene. Palaeogeography, 900   Palaeoclimatology, Palaeoecology 307, 193-204. 901   902   Hoffstetter, R. 1952: Les Mammifères Pléistocènes de la République de l’Équateur. 903   Mémoires de la Société Géologique de France 66, 1-391. 904   905   Höhle, J. & Höhle, M. 2009: Accuracy assessment of digital elevation models by means 906   of robust statistical methods. ISPRS Journal of Photogrammetry and Remote Sensing 907   64, 398-406. 908   909   Klein, R.G. & Cruz-Uribe, K. 1984: The analysis of animal bones from archaeological 910   sites. Chicago, The University of Chicago Press. 911   912   Kos, A.M. 2003: Characterisation of post-depositional taphonomic processes in the 913   accumulation of mammals in a pitfall cave deposit from southeastern Australia. Journal 914   of Archaeological Science 30, 781-796. 915   916   López-Antoñanzas, R., Peláez-Campomanes, P., Álvarez-Sierra, M.A. & García- 917   Paredes, I. 2010: New species of Hispanomys (Rodentia, Cricetodontinae) from the 918   Upper Miocene of Batallones (Madrid, Spain). Zoological Journal of the Linnean 919   Society 160, 725-747. 920   921   López-Antoñanzas, R., Peláez-Campomanes, P. & Álvarez-Sierra, A. 2014: New 922   species of Rotundomys (Cricetinae) from the Late Miocene of Spain and its bearing on 923   the phylogeny of Cricetulodon and Rotundomys. PLoS ONE 9, e112704. 924   925   Lord, T.C., Thorp, J.A. & Wilson, P. 2016: A wild boar dominated ungulate assemblage 926   from an early Holocene natural pit fall trap: Cave shaft sediments in northwest England 927   associated with the 9.3 ka BP cold event. The Holocene 26, 147-153. 928   929   Lyman, R.L. & Fox, G.L. 1989: A critical evaluation of bone weathering as an indication 930   of bone assemblage formation. Journal of Archaeological Science 16, 293-317. 931   932   Lyman, R.L. 1994: Vertebrate Taphonomy. Cambridge University Press. 933   934   Lyman, R.L. 2008: Quantitative Paleozoology. Cambridge University Press. 935   936   Lyndsey, E.L. & Seymour, K.L. 2015: “Tar pits” of the western Neotropics: 937   paleoecology, taphonomy, and mammalian biogeography. Natural History Museum of 938   Los Angeles County Science Series 42, 111-123. 939   940   Marcus, L.F. & Berger, R. 1984: The significance of radiocarbon dates for Rancho La 941   Brea. In Martin, P.S. & Klein, R.G. (eds.), Quaternary Extinctions: A Prehistoric 942   Revolution, 159-188. University of Chicago Press, Chicago. 943   944   Marder, O., Yeshurun, R., Lupu, R., Bar-Oz, G., Belmaker, M., Porat, N., Ron, H. & 945   Frumkin, A. 2011: Mammal remains at Rantis Cave, Israel, and Middle-Late 946   Pleistocene human subsistence and ecology in the Southern Levant. Journal of 947   Quaternary Science 26, 769-780. 948   949   Marean, C.W. & Ehrhardt, C.L. 1995: Paleoanthropological and paleoecological 950   implications of the taphonomy of a sabertooth’s den. Journal of Human Evolution 29, 951   515-547. 952   953   Marín-Arroyo, A.B., Landete Ruiz, M.D., Vidal Bernabeu, G., Seva Román, R., 954   González Morales, M.R. & Straus, L.G. 2008: Archaeological implications of human- 955   derived manganese coatings: a study of blackened bones in El Mirón Cave, Cantabrian 956   Spain. Journal of Archaeological Science 35, 801-813. 957   958   Martin, L.D. & Gilbert, B.M. 1978: Excavations at Natural Trap Cave. Transactions of 959   the Nebraska Academy of Sciences 6, 107-116. 960   961   Martín-Perea, D.M., Aldama, J., Domingo, M.S., Valenciano, A., Abella, J. & Morales, 962   J. 2017: 3D reconstruction of the carnivoran-dominated assemblage of Batallones-3 963   (Late Miocene, Madrid Basin, Spain). Insight into Paleoart 1, 59-63. 964   965   Martín-Perea, D.M., Courtenay, L.A., Domingo, M.S. & Morales, J. 2020: Application of 966   artificially intelligent systems for the identification of discrete fossiliferous levels. PeerJ 967   8, e8767. 968   969   Medina-Chevarrías, V., Oliver, A., López-Guerrero, P., Peláez-Campomanes, P. & 970   Álvarez-Sierra, M.A. 2019: New insights on Hispanomys moralesi (Rodentia, 971   Mammalia) and its use as biostratigraphical indicator. Journal of Iberian Geology 45, 972   641-654. 973   974   Mikulás, R. 1999: Notes on the concept of plant trace fossils related to plant-generated 975   sedimentary structures. Vestník Ceského Geologického Ústavu 74, 39-42. 976   977   Mikulás, R. 2001: Modern and fossil traces in terrestrial lithic substrates. Ichnos 8, 177- 978   184. 979   980   Monescillo, M.F.G., Salesa, M.J., Antón, M., Siliceo, G. & Morales, J. 2014: 981   Machairodus aphanistus (Felidae, Machairodontinae, Homotherini) from the late 982   Miocene (Vallesian, MN10) site of Batallones-3 (Torrejón de Velasco, Madrid, Spain). 983   Journal of Vertebrate Paleontology 34, 699-709. 984   985   Morales, J., Pozo, M., Silva, P.G., Domingo, M.S., López-Antoñanzas, R., Álvarez- 986   Sierra, M.A., Antón, M., Martín Escorza, C., Quiralte, V., Salesa, M.J., Sánchez, I.M., 987   Azanza, B., Calvo, J.P., Carrasco, P., García-Paredes, I., Knoll, F., Hernández- 988   Fernández, M., van den Hoek Ostende, L., Merino, L., van der Meulen, A.J., Montoya, 989   P., Peigné, S., Peláez-Campomanes, P., Sánchez-Marco, A., Turner, A., Abella, J., 990   Alcalde, G.M., Andrés, M., DeMiguel, D., Cantalapiedra, J.L., Fraile, S., García Yelo, 991   B.A., Gómez Cano, A.R., López Guerrero, P., Oliver Pérez, A. & Siliceo, G. 2008: El 992   sistema de yacimientos de mamíferos miocenos del Cerro de los Batallones, Cuenca 993   de Madrid: estado actual y perspectivas. Palaeontologica Nova 8, 41-117. 994   995   Morales, J. 2017: Vertebrados miocenos de los yacimientos del Cerro de los 996   Batallones. In Morales, J. (ed.), La Colina de los Tigres Dientes de Sable. Los 997   yacimientos miocenos del Cerro de los Batallones (Torrejón de Velasco, Comunidad 998   de Madrid), 36-40. Museo Arqueológico Regional, Madrid, Spain. 999   1000   Morales, J., Abella, J. & Valenciano, A. 2017. Thaumastocyon: los extraños 1001   Amphicyonidae de Batallones-3. In Morales, J. (ed.), La Colina de los Tigres Dientes 1002   de Sable. Los yacimientos miocenos del Cerro de los Batallones (Torrejón de Velasco, 1003   Comunidad de Madrid), 338-351. Museo Arqueológico Regional, Madrid, Spain. 1004   1005   Morales, J., Abella, J., Sanisidro, O., Valenviaco, A. 2021. Ammitocyon kainos gen. et 1006   sp. nov., a chimerical amphicyonid (Mammalia, Carnivora) from the late Miocene 1007   carnivore traps of Cerro de los Batallones (Madrid, Spain). Journal of Systematic 1008   Palaeontology 20, 1-23. 1009   1010   Oliver, J.S. 1986: The taphonomy and paleoecology of Shield Trap Cave (24CB91), 1011   Carbon County, Montana. PhD dissertation, University of Maine. 1012   1013   Oliver, J.S., Bonnichsen, R. & Sorg, M.H. 1989: Analogues and site context: bone 1014   damages from Shield Trap Cave (24CB91), Carbon County, Montana, USA. University 1015   of Maine Center for the Study of First Americans. 1016   1017   Organista, E., Pernas-Hernández, M., Gidna, A., Yravedra, J. & Domínguez-Rodrigo, 1018   M. 2016: An experimental lion-to-hammerstone model and its relevance to understand 1019   hominin-carnivore interactions in the archeological record. Journal of Archaeological 1020   Science 66, 69-77. 1021   1022   Paes-Neto, V.D., Francischini, H., Martinelli, A.G., Marinho, T.S., Ribeiro, L.C.B., 1023   Soares, M.B. & Schultz, C.L. 2018: Bioerosion traces on titanosaurian sauropod bones 1024   from the Upper Cretaceous Marília Formation of Brazil. Alcheringa 42, 415-426. 1025   1026   Parkinson, J.A., Plummer, T. & Hartstone-Rose, A. 2015: Characterizing felid tooth 1027   marking and gross bone damage patterns using GIS image analysis: An experimental 1028   feeding study with large felids. Journal of Human Evolution 80, 114-134. 1029   1030   Peigné, S., Salesa, M.J., Antón, M. & Morales, J. 2005: Ailurid carnivoran mammal 1031   Simocyon from the late Miocene of Spain and the systematics of the genus. Acta 1032   Paleontologica Polonica 50, 219-238. 1033   1034   Peigné, S., Salesa, M.J., Antón, M. & Morales, J. 2008: A new Amphicyonine 1035   (Carnivora: Amphicyonidae) from the upper Miocene of Batallones-1, Madrid, Spain. 1036   Palaeontology 51, 943-965. 1037   1038   Pérez-García, A. & Vlachos, E. 2014: New generic proposal for the European Neogene 1039   large testudinids (Cryptodira) and the first phylogenetic hypothesis for the medium and 1040   large representatives of the European Cenozoic record. Zoological Journal of the 1041   Linnean Society 172, 653-719. 1042   1043   Pesquero, M.D., Alcalá, L. & Fernández-Jalvo, Y. 2013: Taphonomy of the reference 1044   Miocene vertebrate site of Cerro de la Garita, Spain. Lethaia 46, 378-398. 1045   1046   Pickford, M. 2015: Late Miocene Suidae from Eurasia: the Hippopotamodon and 1047   Microstonyx problem revisited. Münchner Geowissenschaftliche Abhandlungen 42, 1- 1048   124. 1049   1050   Pokines, J.T. & Kerbis Peterhans, J.C. 2007: Spotted hyena (Crocuta crocuta) den use 1051   and taphonomy in the Masai Mara National Reserve. Journal of Archaeological 1052   Science 34, 1914-1931. 1053   1054   Pozo, M., Calvo, J.P., Silva, P.G., Morales, J., Peláez-Campomanes, P. & Nieto, M. 1055   2004: Geología del sistema de yacimientos de mamíferos miocenos del Cerro de los 1056   Batallones, Cuenca de Madrid. Geogaceta 35, 143-146. 1057   1058   Ríos, M., Sánchez, I.M. & Morales, J. 2017: A new giraffid (Mammalia, Ruminantia, 1059   Pecora) from the late Miocene of Spain, and the evolution of the sivathere-samothere 1060   lineage. PLoS ONE 12, e0185378. 1061   1062   Ríos, M. & Morales, J. 2019. A new skull of Decennatherium rex Ríos, Sánchez and 1063   Morales, 2017 from Batallones-4 (upper Vallesian, MN10, Madrid, Spain). 1064   Palaeontologia Electronica, 22.2.pvc_1. 1065   1066   Romano, C.O., Pesquero, M.D. & Alberdi, M.T. 2017: Hipparion: los caballos de 1067   Batallones. In Morales, J. (ed.), La Colina de los Tigres Dientes de Sable. Los 1068   yacimientos miocenos del Cerro de los Batallones (Torrejón de Velasco, Comunidad 1069   de Madrid), 426-440. Museo Arqueológico Regional, Madrid, Spain. 1070   1071   Salesa, M.J., Antón, M., Peigné, S. & Morales, J. 2006: Evidence of a false thumb in a 1072   fossil carnivore clarifies the evolution of pandas. Proceedings of the National Academy 1073   of Science 103, 379-382. 1074   1075   Salesa, M.J., Antón, M., Peigné, S. & Morales, J. 2008: Functional anatomy and 1076   biomechanics of the postcranial skeleton of Simocyon batalleri (Viret, 1929) (Carnivora, 1077   Ailuridae) from the Late Miocene of Spain. Zoological Journal of the Linnean Society 1078   152, 593-621. 1079   1080   Salesa, M.J., Antón, M., Turner, A. & Morales, J. 2010: Functional anatomy of the 1081   forelimb in Promegantereon ogygia (Felidae, Machairodontinae, Smilodontini) from the 1082   Late Miocene of Spain and the origins of the sabre-toothed felid model. Journal of 1083   Anatomy 216, 381-396. 1084   1085   Salesa, M.J., Antón, M., Morales, J. & Peigné, S. 2012: Systematics and phylogeny of 1086   the small felines (Carnivora, Felidae) from the late Miocene of Europe: a new species 1087   of Felinae from the Vallesian of Batallones (MN10, Madrid, Spain). Journal of 1088   Systematic Palaeontology 10, 87-102. 1089   1090   Salesa, M.J., Antón, M., Siliceo, G. & Peigné, S. 2017: Los otros félidos de Batallones. 1091   In Morales, J. (ed.), La Colina de los Tigres Dientes de Sable. Los yacimientos 1092   miocenos del Cerro de los Batallones (Torrejón de Velasco, Comunidad de Madrid), 1093   322-336. Museo Arqueológico Regional, Madrid, Spain. 1094   1095   Sánchez, I.M., Domingo, M.S. & Morales, J. 2009: New data on the Moschidae 1096   (Mammalia, Ruminantia) from the upper Miocene of Spain (MN10-MN11). Journal of 1097   Vertebrate Paleontology 29, 567-575. 1098   1099   Sánchez, I.M., Quiralte, V. & Morales, J. 2011: Presence of the bovid Austroportax in 1100   the upper Miocene fossil site of Batallones-1 (MN10, Madrid Basin, Madrid). Estudios 1101   Geológicos 67, 637-642. 1102   1103   Sanisidro, O. & Cantalapiedra, J. 2017: Los rinocerontes del Cerro de los Batallones. 1104   In Morales, J. (ed.), La Colina de los Tigres Dientes de Sable. Los yacimientos 1105   miocenos del Cerro de los Batallones (Torrejón de Velasco, Comunidad de Madrid), 1106   410-421. Museo Arqueológico Regional, Madrid, Spain. 1107   1108   Sarjeant, W.A.S. 1975: Plant trace fossils. In Frey, R.W. (ed.), The study of trace 1109   fossils: a synthesis of principles, problems and procedures in ichnology, 163-179. 1110   Springer-Verlag, New York. 1111   1112   Shahack-Gross, R., Bar-Yosef, O. & Weiner, S. 1997. Black-coloured bones in 1113   Hayonim Cave, Israel: differentiating between burning and oxide staining. Journal of 1114   Archaeological Science 24, 439-446. 1115   1116   Siliceo, G., Salesa, M.J., Antón, M., Monescillo, M.F.G. & Morales, J. 2014: 1117   Promegantereon ogygia (Felidae, Machairodontinae, Smilodontini) from the Vallesian 1118   (late Miocene, MN10) of Spain: morphological and functional differences in two 1119   noncontemporary populations. Journal of Vertebrate Paleontology 34, 407-418. 1120   1121   Siliceo, G., Salesa, M.J., Antón, M., Pastor, J.F. & Morales, J. 2015: Comparative 1122   Anatomy of the Shoulder Region in the late Miocene Amphicyonid Magericyon anceps 1123   (Carnivora): Functional and Paleoecological Inferences. Journal of Mammalian 1124   Evolution 22, 243-258. 1125   1126   Siliceo, G., Salesa, M.J., Antón, M., Peigné, S. & Morales, J. 2017: Functional anatomy 1127   of the cervical región in the late Miocene amphicyonid Magericyon anceps (Carnivora, 1128   Amphicyonidae): implications for its feeding behavior. Palaeontology 60, 329-347. 1129   1130   Siliceo, G., Antón, M., Morales, J. & Salesa, M.J. 2020: Built for Strength: Functional 1131   Insights from the Thoracolumbar and Sacrocaudal Regions of the Late Miocene 1132   Amphicyonid Magericyon anceps (Carnivora, Amphicyonidae) from Batallones-1 1133   (Madrid, Spain). Journal of Mammalian Evolution 27, 497-518. 1134   1135   Stiner, M.C. 1990: The use of mortality patterns in archaeological studies of hominin 1136   predatory adaptations. Journal of Anthropological Archaeology 9, 305-351. 1137   1138   Stiner, M.C., Arsebük, G. & Howell, F.C. 1996: Cave bears and Paleolithic artifacts in 1139   Yarimburgaz Cave, Turkey: dissecting a palimpsest. Geoarchaeology 11, 279-327. 1140   1141   Stiner, M.C. 1998: Mortality analysis of Pleistocene bears and its paleoanthropological 1142   relevance. Journal of Human Evolution 34, 303-326. 1143   1144   Stock, C. 1972: Rancho La Brea: a record of Pleistocene life in California. Natural 1145   History Museum of Los Angeles County Science Series 20, 1-81. 1146   1147   Stock, C. & Harris, J.M. 1992: Rancho la Brea: A record of Pleistocene life in California. 1148   Natural History Museum of Los Angeles County Science Series 37, 1-113. 1149   1150   Tjelldén, A.K.E., Kristiansen, S.M., Matthiesen, H. & Pedersen, O. 2016: Impact of 1151   roots and rhizomes on wetland archaeology: a review. Conservation and Management 1152   of Archaeological Sites 17, 370-391. 1153   1154   Valenciano, A., Abella, J., Sanisidro, O., Hartstone-Rose, A., Alvarez-Sierra, M.A. & 1155   Morales, J. 2015: Complete description of the skull and mandible of the giant mustelid 1156   Eomellivora piveteaui Ozansoy, 1965 (Mammalia, Carnivora, Mustelidae) from 1157   Batallones (MN10), late Miocene (Madrid, Spain). Journal of Vertebrate Paleontology 1158   35, e934570. 1159   1160   Valenciano, A. 2017: Mofetas, martas, tejones y rateles gigantes del Cerro de los 1161   Batallones. In Morales, J. (ed.), La Colina de los Tigres Dientes de Sable. Los 1162   yacimientos miocenos del Cerro de los Batallones (Torrejón de Velasco, Comunidad 1163   de Madrid), 322-336. Museo Arqueológico Regional, Madrid, Spain. 1164   1165   Valenciano, A. & Govender, R. 2020: New fossils of Mellivora benfieldi (Mammalia, 1166   Carnivora, Mustelidae) from Langebaanweg ‘E’ Quarry (South Africa, early Pliocene): 1167   re-evaluation of the African Neogene mellivorines. Journal of Vertebrate Paleontology, 1168   e1817754. 1169   1170   Valenciano, A., Pérez-Ramos, A., Abella, J. & Morales, J. 2020: A new 1171   hypercarnivorous mustelid (Mammalia, Carnivora, Mustelidae) from Batallones, late 1172   Miocene (MN10), Torrejón de Velasco, Madrid, Spain). Geodiversitas 42, 103-121. 1173   1174   Villa, P. & Mahieu, E. 1991: Breakage patterns of human long bones. Journal of 1175   Human Evolution 21, 27-48. 1176   1177   Walton, R., Baxter, R., Bunbury, N., Hansen, D., Fleischer-Dogley, F., Greenwood, S. 1178   & Shaepman-Strub, G. 2019: In the land of giants: habitat use and selection of the 1179   Aldabra giant tortoise on Aldabra Atoll. Biodiversity and Conservation 28, 3183-3198. 1180   1181   Wang, X. & Martin, L.D. 1993: Natural trap cave. National Geography Research 1182   Explorer 9, 422-435. 1183   1184   White, J.A., McDonald, H.G., Anderson, E., Soiset, J.M. 1984: Lava blisters as 1185   carnivore traps. Special Publication Carnegie Museum of Natural History, Carnegie 1186   Museum of Natural History 8, 241-256. 1187   1188   Wolverton, S. 2001: Caves, ursids and artifacts: a natural-trap hypothesis. Journal of 1189   Ethnobiology 21, 55-72. 1190   1191   Wolverton, S. 2006: Natural-trap ursid mortality and the Kurtén Response. Journal of 1192   Human Evolution 50, 540-551. 1193   Table 1. Number of Identified Specimens (NISP), Minimum Number of Elements (MNE) and Minimum Number of Individuals (MNI) of the large 1194   taxa in Batallones-3 levels. Mortality patterns of individuals is given in brackets (young/prime adult/old). 1195   Level  I   Level  II   Level  III           Family/Order   NISP   %NISP   MNE   MNI   %MNI       NISP   %NISP   MNE   MNI   %MNI       NISP   %NISP   MNE   MNI   %MNI   Machairodus  aphanistus   Felidae/Carnivora   752   15.83   730   7  (0/7/0)   15.56   844   20.91   824   8  (1/7/0)   20.51   537   31.44   530   7  (1/6/0)   41.18       Promegantereon  ogygia   Felidae/Carnivora   956   20.13   913   10  (1/9/0)   22.22   1117   27.67   1042   12  (1/11/0)   30.77   110   6.44   101   3  (0/3/0)   17.65       Leptofelis  vallesiensis   Felidae/Carnivora   4   0.08   4   1  (0/1/0)   2.22   1   0.02   1   1  (0/1/0)   2.56   0   0.00   0   0  (0/0/0)   0.00       Felidae  indet.   Felidae/Carnivora   29   0.40   -­‐   -­‐   -­‐   27   0.59   -­‐   -­‐   -­‐   5   0.23   -­‐   -­‐   -­‐       Indarctos  arctoides   Ursidae/Carnivora   918   19.33   782   9  (3/6/0)   20.00   517   12.81   434   6  (2/4/0)   15.38   149   8.72   136   3  (0/3/0)   17.65       Magericyon  anceps   Amphicyonidae/Carnivora   12   0.25   12   1  (0/1/0)   2.22   32   0.79   30   1  (0/1/0)   2.56   50   2.93   50   1  (0/1/0)   5.88       Ammitocyon  kainos   Amphicyonidae/Carnivora   14   0.29   14   1  (0/1/0)   2.22   14   0.35   14   1  (0/1/0)   2.56   56   3.28   56   1  (0/1/0)   5.88       Amphicyonidae  indet.   Amphicyonidae/Carnivora   2   0.04   -­‐   -­‐   -­‐   7   0.17   -­‐   -­‐   -­‐   90   5.27   -­‐   -­‐   -­‐       Protictitherium  crassum   Hyaenidae/Carnivora   110   2.32   110   6  (0/6/0)   13.33   105   2.60   102   3  (0/3/0)   7.69   13   0.76   12   1  (0/1/0)   5.88       Eomellivora  piveteaui   Mustelidae/Carnivora   85   1.79   85   4  (0/4/0)   8.89   109   2.70   109   3  (0/3/0)   7.69   2   0.12   2   1  (0/1/0)   5.88       aff.  Adroverictis  sp.   Mustelidae/Carnivora   8   0.17   8   1  (0/1/0)   2.22   1   0.02   1   1  (0/1/0)   2.56   0   0.00   0   0  (0/0/0)   0.00       Circamustela  peignei   Mustelidae/Carnivora   5   0.11   5   1  (0/1/0)   2.22   1   0.02   1   1  (0/1/0)   2.56   0   0.00   0   0  (0/0/0)   0.00       Mustelidae  indet.   Mustelidae/Carnivora   65   1.37   -­‐   -­‐   -­‐   65   1.61   -­‐   -­‐   -­‐   9   0.53   -­‐   -­‐   -­‐       Mephitidae  nov.  gen.  et  sp.   Mephitidae/Carnivora   3   0.06   2   1  (0/1/0)   2.22   0   0.00   0   0  (0/0/0)   0.00   0   0.00   0   0  (0/0/0)   0.00       Promephitis  nov.  sp.   Mephitidae/Carnivora   19   0.40   -­‐   -­‐   -­‐   4   0.10   -­‐   -­‐   -­‐   2   0.12   -­‐   -­‐   -­‐       Carnivora  indet.   Carnivora   85   1.79   -­‐   -­‐   -­‐   18   0.45   -­‐   -­‐   -­‐   3   0.18   -­‐   -­‐   -­‐       Total  Carnivora   3067   64.36   2665   42  (4/38/0)   93.33   2862   70.82   2558   37  (4/33/0)   94.87   1026   60.01   887   17  (1/16/0)   100           Hipparion  cf.  matthewi   Equidae/Perissodactyla   3   0.06   3   1  (0/1/0)   2.22   4   0.10   4   1  (0/1/0)   2.56   0   0.00   0   0   0.00       Rhinocerotidae  indet.   Rhinocerotidae/Perissodactyla   0   0.00   -­‐   -­‐   -­‐   1   0.02   -­‐   -­‐   -­‐   0   0.00   -­‐   -­‐   -­‐       Total  Perissodactyla   3   0.063   3   1  (0/1/0)   2.22   5   0.12   4   1  (0/1/0)   2.56   0   0.00   0   0   0.00         Hippopotamodon  major   Suidae/Artiodactyla   2   0.04   2   1  (0/1/0)   2.22   1   0.02   1   1  (0/1/0)   2.56   0   0.00   0   0   0.00       Austroportax  sp.   Bovidae/Artiodactyla   3   0.06   3   1  (0/1/0)   2.22   0   0.00   0   0   0.00   0   0.00   0   0   0.00       Bovidae  indet.   Bovidae/Artiodactyla   3   0.06   -­‐   -­‐   -­‐   0   0.00   -­‐   -­‐   -­‐   1   0.06   -­‐   -­‐   -­‐       Total  Artiodactyla   8   0.168   5   2  (0/2/0)   4.44   1   0.02   1   1  (0/1/0)   2.56   1   0.06   0   0   0.00         Herbivore  indet.   10   0.21   -­‐   -­‐   -­‐   0   0.00   -­‐   -­‐   -­‐   0   0.00   -­‐   -­‐   -­‐           Aves  indet.   178   3.75   -­‐   -­‐   -­‐   185   4.58   -­‐   -­‐   -­‐   44   2.58   -­‐   -­‐   -­‐           Testudines  indet.   Testudines   103   2.17   -­‐   -­‐   -­‐   44   1.09   -­‐   -­‐   -­‐   67   3.92   -­‐   -­‐   -­‐         Varanus  marathoniensis   Varanidae/Squamata   40   0.84   -­‐   -­‐   -­‐   3   0.07   -­‐   -­‐   -­‐   0   0.00   -­‐   -­‐   -­‐       Reptilia  indet.   4   0.08   -­‐   -­‐   -­‐   3   0.07   -­‐   -­‐   -­‐   0   0.00   -­‐   -­‐   -­‐           Indetermined       1337   28.15   -­‐   -­‐   -­‐       934   23.14   -­‐   -­‐   -­‐       570   33.37   -­‐   -­‐   -­‐   Total   4750   100   2683   45   100   4037   100   2566   39   100   1708   100   888   17   100         1196   Table 2. Chi-square tests to compare the three fossiliferous levels of Batallones-3 and 1197   Batallones-1. Above the diagonal are the p-values. Below diagonal is the chi-square 1198   statistic. Significant differences when p<0.01. Batallones-­‐3           Level  I   Level  II   Level  III     Batallones-­‐1   %NISP   Batallones-­‐3  Level  I   -­‐   1.00   1.00   1.00   Carnivora  -­‐   Batallones-­‐3  Level  II   0.00   -­‐   1.00   1.00   Artiodactyla  -­‐   Batallones-­‐3  Level  III   0.00   0.00   -­‐   1.00   Perissodactyla   Batallones-­‐1   0.00   0.00   0.00   -­‐   Batallones-­‐3  Level  I   -­‐   0.52   <  0.01   <  0.  01   %MNI   Batallones-­‐3  Level  II   5.19   -­‐   0.  02   0.  06   Carnivoran   Batallones-­‐3  Level  III   20.98   15.25   -­‐   <  0.  01   species   Batallones-­‐1   53.31   45.27   49.67   -­‐   Batallones-­‐3  Level  I   -­‐   0.63   0.16   <  0.  01   %NISP   Batallones-­‐3  Level  II   1.86   -­‐   0.80   <  0.  01   Number  of   Batallones-­‐3  Level  III   5.19   0.93   -­‐   <  0.  01   fractures   Batallones-­‐1   22.11   30.86   36.81   -­‐   Batallones-­‐3  Level  I   -­‐   0.013   0.12   <  0.  01   %NISP   Batallones-­‐3  Level  II   16.10   -­‐   <  0.  01   0.03   Fracture  angle   Batallones-­‐3  Level  III   7.35   22.47   -­‐   <  0.  01   Batallones-­‐1   33.24   13.77   44.59   -­‐   Batallones-­‐3  Level  I   -­‐   0.60   0.10   <  0.  01   %NISP   Batallones-­‐3  Level  II   1.03   -­‐   0.09   <  0.  01   Fracture   Batallones-­‐3  Level  III   4.64   4.75   -­‐   <  0.  01   surface   Batallones-­‐1   29.73   32.30   45.70   -­‐   Batallones-­‐3  Level  I   -­‐   0.15   <  0.  01   <  0.  01   %NISP  Bone   Batallones-­‐3  Level  II   5.40   -­‐   <  0.  01   <  0.  01   completeness   Batallones-­‐3  Level  III   25.50   20.40   -­‐   <  0.  01   Batallones-­‐1   26.88   12.52   38.23   -­‐   Batallones-­‐3  Level  I   -­‐   1.00   1.00   1.00   %NISP   Batallones-­‐3  Level  II   0.00   -­‐   1.00   1.00   Weathering   Batallones-­‐3  Level  III   0.00   0.00   -­‐   1.00   Batallones-­‐1   0.00   0.00   0.00   -­‐   Batallones-­‐3  Level  I   -­‐   0.31   0.02   1.00   %NISP   Batallones-­‐3  Level  II   1.01   -­‐   0.09   0.31   Abrasion   Batallones-­‐3  Level  III   5.18   2.75   -­‐   0.02   Batallones-­‐1   0.00   1.01   5.13   -­‐   Batallones-­‐3  Level  I   -­‐   1.00   1.00   1.00   %NISP   Batallones-­‐3  Level  II   0.00   -­‐   1.00   1.00   Trampling   Batallones-­‐3  Level  III   0.00   0.00   -­‐   1.00   Batallones-­‐1   0.00   0.00   0.00   -­‐   Batallones-­‐3  Level  I   -­‐   0.02   <  0.  01   <  0.  01   %NISP  Root   Batallones-­‐3  Level  II   5.18   -­‐   0.46   <  0.  01   marks   Batallones-­‐3  Level  III   8.97   0.54   -­‐   <  0.  01   Batallones-­‐1   54.93   84.04   94.57   -­‐   Batallones-­‐3  Level  I   -­‐   0.05   <  0.  01   <  0.  01   %NISP   Batallones-­‐3  Level  II   3.72   -­‐   0.46   <  0.  01   Manganese   oxides   Batallones-­‐3  Level  III   11.51   2.41   -­‐   <  0.  01   Batallones-­‐1   44.63   68.81   88.27   -­‐   Batallones-­‐3  Level  I   -­‐   1.00   1.00   1.00   %NISP   Batallones-­‐3  Level  II   0.00   -­‐   1.00   1.00   Carnivore   Batallones-­‐3  Level  III   0.00   0.00   -­‐   1.00   marks   Batallones-­‐1   0.00   0.00   0.00   -­‐   1199   1200   1201   1202   1203   1204   1205   1206   1207   1208   1209   1210   1211   1212   1213   1214   1215   1216   1217   1218   1219   1220   Table 3. Skeletal element frequencies for Batallones-3 levels expressed as Number of 1221   Identified Specimens (NISP) and percentages. Level  I   Level  II   Level  III       NISP   %         NISP   %         NISP   %   Horn   2   0.04   0   0.00   0   0.00   Skull   52   1.08     35   0.86     22   1.29   Mandible   55   1.15     43   1.06     11   0.64   Molar   23   0.48     16   0.39     3   0.18   Premolar   46   0.96     45   1.11     8   0.47   Canine   34   0.71     57   1.40     15   0.88   Incisor   118   2.46     116   2.86     30   1.76   Indet.  Tooth   15   0.31     6   0.15     2   0.12   Hyoid   10   0.21     3   0.07     8   0.47   Atlas   19   0.40     16   0.39     3   0.18   Axis   22   0.46     14   0.34     9   0.53   Vertebra   721   15.04     630   15.52     309   18.09   Sternebra   39   0.81     34   0.84     23   1.35   Rib   367   7.66     262   6.45     134   7.85   Clavicle   9   0.19     14   0.34     2   0.12   Scapula   41   0.86     35   0.86     21   1.23   Pelvis   69   1.44     43   1.06     11   0.64   Baculus   5   0.10     1   0.02     5   0.29   Humerus   21   0.44     57   1.40     30   1.76   Radius   72   1.50     55   1.36     19   1.11   Ulna   78   1.63     54   1.33     28   1.64   Femur   78   1.63     84   2.07     30   1.76   Tibia   84   1.75     70   1.72     18   1.05   Fibula   23   0.48     44   1.08     10   0.59   Patella   47   0.98     48   1.18     10   0.59   Carpal/Tarsal   415   8.66     356   8.77     148   8.67   Astragalus   53   1.11     51   1.26     18   1.05   Calcaneus   60   1.25     60   1.48     17   1.00   Indet.  Metapodial   0   0.00     85   2.09     36   2.11   Metacarpal   167   3.48     150   3.70     59   3.45   Metatarsal   184   3.84     141   3.47     47   2.75   1st  Phalanx   374   7.80     287   7.07     126   7.38   2nd  Phalanx   296   6.17     255   6.28     98   5.74   3rd  Phalanx   247   5.15     230   5.67     94   5.50   Indet.  Phalanx   107   2.23     111   2.73     52   3.04   Sesamoid   340   7.09     321   7.91     145   8.49   Turtle  shell   40   0.83     20   0.49     16   0.94   Indetermined   461   9.62         210   5.17         91   5.33   TOTAL   4794   4048   1711   1222         1223   1224   Table 4. Other surface bone modifications per level. Indet. = Unidentifiable remains 1225   with no assigned fossiliferous level. Level  I   Level  II   Level  III   Indet.       NISP   %         NISP   %         NISP   %         NISP   %   Weathering   Stage  0    479   100.0   405   100.0   170   99.4   976   97.2   Stage  1   0   0.0     0   0.0     1   0.6     25   2.5   Stage  2   0   0.0     0   0.0     0   0.0     3   0.3   Abrasion         Stage  1    479   100.0   399   98.5   162   94.7   923   91.9   Stage  2   0   0.0     6   1.5     9   5.3     77   7.7   Stage  3   0   0.0     0   0.0     0   0.0     4   0.4   Root  marks   210   43.8     242   59.7     110   64.4     618   61.6   Trampling   0   0.0     0   0.0     0   0.0     16   1.6   Manganese  oxides   352   73.5     343   84.7     157   91.8     598   59.6   1226         1227   1228   1229   1230   1231   1232   1233   1234   1235   1236   1237   1238   1239   1240   1241   1242   1243   1244   Table 5. Skeletal completeness for Batallones-3 Levels I, II and III most abundant 1245   carnivorans. Teeth, vertebrae, ribs, and sesamoids were removed for the estimation of 1246   the percentage of skeletal completeness. Kolmogorov-Smirnov test compares the MNE 1247   and the expected number of elements for the MNI per taxa, per level. p > 0.01 = not 1248   significantly different. p < 0.01 = significantly different. 1249   Kolmogorov-­‐Smirnov  Test       %  Skeletal  completeness   D   p   Level  I   Machairodus  aphanistus     71.29   0.50   0.039   Promegantereon  ogygia   71.17   0.43   0.111   Indarctos  arctoides   70.15   0.36   0.267   Level  II   Machairodus  aphanistus     75.20   0.57   0.012   Promegantereon  ogygia   72.69   0.43   0.111   Indarctos  arctoides   62.07   0.50   0.039   Level  III   Machairodus  aphanistus     55.09   0.57   0.012   Promegantereon  ogygia   45.62   0.50   0.039   Indarctos  arctoides   33.67   0.64   0.029   1250   1251   1252   1253   1254   1255   1256   1257   1258   1259   1260   1261   1262   1263   1264   Figure legends 1265   1266   Figure 1. A. Map of the Iberian Peninsula showing the location of Madrid and 1267   Valdemoro. B. Location of Batallones-1 and Batallones-3 sites, west of Valdemoro. C. 1268   Theoretical hourglass shaped cavity, with an upper herbivore-dominated assemblage 1269   and a lower, carnivoran-dominated assemblage. Modified from Pozo et al. (2004) and 1270   Domingo et al. (2013a). 1271   1272   Figure 2. Spatial distribution of Batallones-3 fossil remains. A. Plan view (X-Y 1273   coordinate plane). B. X-Z cross section (left), showing the fine-tuned discrete 1274   fossiliferous level models. C. X-Z cross section (right), showing the fine-tuned discrete 1275   fossiliferous level models. 1276   1277   Figure 3. Skeletal element proportions expressed as Relative Abundance (Ri) in 1278   Batallones-3. A, D and G. Machairodus aphanistus skeletal element proportions in 1279   Batallones-3 Levels I, II and III, respectively. B, E and H. Promegatereon ogygia 1280   skeletal element proportions in Batallones-3 Levels I, II and III, respectively. C, F and I. 1281   Indarctos arctoides skeletal element proportions in Batallones-3 Levels I, II and III, 1282   respectively. Mand = mandible, Vert = vertebra, Scap = scapula, Pel = pelvis, Hum = 1283   humerus, Rad = radius, Uln = ulna, Fem = femur, Tib = tibia, Pat = patella, Carp/Tars = 1284   carpal/tarsal, Mtc/Mtt = metacarpal/metatarsal, Phal = phalanx, Sesa = sesamoid. 1285   1286   Figure 4. Stereographic projections of the elongated axis of remains. Mean trend and 1287   plunge represented by plus sign. A. Level I stereographic projections. B. Level II 1288   stereographic projections. C. Level III stereographic projections. Cold colours: lower 1289   density of remains. Warm colours: higher density of remains. 1290   1291   Figure 5. Indarctos arctoides fossil densities for Batallones-3 levels. Density values 1292   represent the number of points per square unit (cm), per quadrat. A. Level I Indarctos 1293   arctoides fossil densities. B. Level II Indarctos arctoides fossil densities. C. Level III 1294   Indarctos arctoides fossil densities. The MNI of Indarctos arctoides per level is provided 1295   in the lower right corner. 1296   1297   Figure 6. Bone breakage at Batallones-3. A, B and C. Number of fractures. D, E and F. 1298   Degree of bone completeness. G, H, and I. Fracture type/angle. J, K and L. Fracture 1299   surface. A, D, G and J. Level I remains. B, E, H and K. Level II remains. C, F, I and L. 1300   Level III remains. 1301   1302   Figure 7. A. Common Batallones-3 taphonomic modifications. Ailuropodine bear 1303   Indarctos arctoides skull BAT-3’15-1000 (Level I), displaying typical Batallones-3 1304   taphonomical modifications on autochthonous remains. Fractures are due to 1305   fossildiagenetic breakage. Note that, despite the high frequency of fractures, the skull 1306   is complete. B. Detail of the root-marked upper carnassial (P4). C. Detail of root marks 1307   on the mandible. D and E. Sabretooth felid Promegantereon ogygia right femur BAT- 1308   3’09-1236 (Level III) with perpendicular fossildiagenetic fractures with smooth fracture 1309   surfaces. F and G. Allochthonous herbivore remains in the Batallones-3 assemblage. 1310   F. Hipparion inferior molar BAT-3’12-1695 (Undetermined level). G. Hipparion incisors 1311   with mandible fragments BAT-3’11-1716 and BAT-3’11-1717 (Level I). 1312   1313   Figure 8. Batallones-3 associated and articulated remains. A. Machairodus aphanistus 1314   articulated carpals, metacarpals, and phalanges. B. Associated Magericyon anceps 1315   radius, ulna, tibia, femur, and vertebrae. Note diagenetic fractures affecting both bones 1316   and surrounding matrix. 1317   1318   Figure 9. Mortality patterns for Indarctos arctoides individuals for each level, compared 1319   to other ursid-abundant assemblages. YC = Yarimburgaz Cave, a hibernation death 1320   assemblage (Stiner et al. 1996; Stiner 1998). LC = Lawson Cave and JC = Jerry Long 1321   Cave (Wolverton 2006), two natural trap bear assemblages. 1322   1323   Figure 10. Taphonomic history of the Batallones-3 assemblage. A. Permanent water 1324   body before cavity roof collapse. B. Descent of phreatic level, roof collapse and debris 1325   cone formation. C, E and G. Natural trapping of carnivores (autochthonous 1326   assemblages). D, F and H. Alluvial sediment and allochthonous remains input. C. 1327   Batallones-3 Level I. E. Batallones-3 Level II. G. Batallones-3 Level III. I. Present day 1328   slope erosion, pine replantation. Due to slope erosion, it is uncertain if an upper 1329   herbivore-dominated assemblage was ever formed after step H.