New insight into the Pleistocene deposits of Monte delle Piche, Rome, and remarks on the biochronology of Hippopotamus (Mammalia, Hippopotamidae) and Stephanorhinus etruscus (Mammalia, Rhinocerotidae) in Italy biocronología de Hippopotamus (Mammalia, Hippopotamidae) y Stephanorhinus etruscus (Mammalia, Rhinocerotidae) en Italia

Several large mammal assemblages have been collected in the Roman basin since the XIX century, but they usu- ally lack any stratigraphic datum or details about the fossiliferous localities. In this work, the stratigraphic provenance of large mammal remains discovered at Monte delle Piche (Rome) is investigated. The systematic revision of these specimens allows the recording of the presence of Hippopotamus sp., Stephanorhinus sp. and Stephanorhinus etruscus . On the basis of micropalaeontological analysis performed on sediment sampled from the studied speci- mens and considering the stratigraphy of the area, two fossiliferous levels are recognised at Monte delle Piche. The remain of the hippopotamus was collected in fluvial gravels and sand deposits, in which the presence of Cyprideis is also recorded. This deposit is chronologically related to the latest Early-early Middle Pleistocene. Hippopotamus was present in Italy and Western Europe from the latest Villafranchian to MIS 4/3. The mandible of S. etruscus was collected in marine deposits along with abundant foraminifera and ostracods, which corre-late with the late Early Pleistocene. Stephanorhinus etruscus occurred in Western Europe at the beginning of the Villafranchian, and it was documented until the end of the Villafranchian in Italy and until the Early-Middle Pleistocene transition in the Iberian Peninsula.

The aim of this paper is the systematic revision of the Pleistocene mammals collected in the fossiliferous deposit of Monte delle Piche (Rome) at the beginning of the second half of the XIX century. Ponzi (1858) described shortly the Monte delle Piche sequence, which was characterised, from the bottom to the top, by "Sub-Apennine Marne" (=?Monte Vaticano Formation), sands with some lignite beds at the base (in which mammal remains have been collected according to the author), alluvial gravels and volcanic tuff (Fig. 1). On the basis of the results of micropalaeontological analysis performed on a relatively small amount of sediment sampled from the studied specimens, the revised biochronology of large mammals and considering the stratigraphy of the area, the relative chronological position of the specimens is investigated.

Material and Methods
The considered mammal remains are housed at Museo di Paleontologia, "Sapienza" University of Rome (MPUR) and have never been studied in detail. The specimens have been collected at the foot of Monte delle Piche during the construction of the Rome-Civitavecchia railway between 1856 and 1858 (Ponzi, 1858) (Fig. 1).
The mammal remains from Monte delle Piche are represented by two fragmentary mandibles of Rhinocerotidae (MPUR 1515; MPUR 1516) and a femur of a large-sized Hippopotamidae (MPUR 999). A third reworked hemimandible of a rhinocerotid (MPUR 138), collected from the same locality, was described by Pandolfi et al. (2013b) and referred to Acerorhinus sp.
The dental nomenclature follows Guérin (1980) and Lacombat (2005); the morphometric data of the Monte delle Piche specimens were obtained following Guérin (1980) and Lacombat (2005) and are reported in Table 1. Ratio diagram was made using data on extant Diceros bicornis (data from Guérin, 1980) as standard.
Micropalaeontological content of sediment samples collected from the studied specimens was  analysed in order to add some palaeoenvironmental and biostratigraphic information or to corroborate existing data. The analyses on ostracods and foraminifera were performed on all the samples, which were disaggregated in 5% H 2 O 2 solution, washed using a 0.125 mm mesh sieve and dried. From the whole dried sieved sediments, ostracods and foraminifera was handpicked and analysed under the stereomicroscope. On the whole, ostracod valves were medium preserved, while foraminifera show a preservation spanning between bad to good.
The revised Quaternary time scale (Gibbard et al., 2010) for chronological references is used in this text; the Pleistocene spans approximately between 2.6 and 0.001 Ma. The biostratigraphic scheme adopted here is that of Iaccarino et al. (2007). Description: The femur lacks of the distal epiphysis and the proximal articular head, which was supported by a relatively elongated neck (Fig. 2). The trocanter is massive and the diaphysis has a sub-circular section.

Class
Remarks: Elongated neck is evident in H. antiquus according to Caloi et al. (1980) and can be noticed in some specimens from L'Aquila Basin (L'Aquila) (MSG 385 and 403) and S. Oreste (Rome) (MPUR/V 1950) referred to this species. Nevertheless, this character is also observable in some specimens referred to H. amphibius (Caloi et al., 1980). The breadth of the diaphysis (TDd=70.7 mm) is smaller than those measured on the femurs of H. antiquus from S. Oreste (=H. ex gr. H. antiquus according to Mazza & Bertini, 2013) given by Caloi et al. (1980) (TDd=80 and 76 mm) and from Valdarno 1=data from Guérin (1980); 2=data from Lacombat (2005Lacombat ( , 2006 Mazza & Bertini (2013). However, due to the conservation status of the remain and the wide morphological variability of Hippopotamus (Caloi et al., 1980;Petronio, 1995), a sure specific attribution of the studied specimen is excluded.
Description: The specimen MPUR 1515 consists only of a partial horizontal ramus with D/3, D/4 and M/1 (Fig. 3A). Labial cingula are absent; D/4 and M/1 display a relatively marked and deep vestibular grooves; M/1 has a long and narrow paralophid. D/3 is relatively worn and the paralophid is short and broad. The anterior lingual valley in D/3 has a narrow V-shaped morphology. The posterior valley in D/4 and M/1 has a broad V-shaped morphology. Other morphological characters cannot be observed because the specimen is covered by a thick and hard crust of sediment.
Remarks while it is closed in the specimen from Pirro Nord and some specimens from Senèze (e.g., NMB Se1682). In the D/3 from Barberino a weak mesial cingulum  is also present. V-shaped lingual valleys and marked vestibular groove are evident in D/4 and M/1 of two hemi-mandibles of S. etruscus from Navezze di Gussago (MGPP 31406 and 31406a). In the latter specimens, D/3 displays the same morphological characteristics than that from Monte delle Piche, while M/1 has a mesial cingulum. The specimen from Monte delle Piche differs from S. hundsheimensis from Mosbach (MfN 26224, 26225) in which the vestibular wall of the trigonid in D/3 is slightly concave (Fig. 3D) and D/4 has an U-shaped posterior lingual valley and a deep vestibular groove (MfN 26224, 26225, 26227). Two D/4 of S. hundsheimensis from Contrada Monticelli (MSTB 30600 and 30600a) display a more obtuse and shallow vestibular groove; this feature is also present in the specimens of S. hund sheimensis reported by Lacombat (2006). In Dihoplus megarhinus (NMB MP921), D/3 has a relatively more marked and deep vestibular groove than in the Monte delle Piche specimen (Fig. 3E); moreover D/3 and D/4 display distal and mesial cingula. In D/3 of Dihoplus megarhinus from Montepellier, a marked vestibular groove occurs between the protoconid and the paraconid (Fig. 3E). This groove appears very shallow in S. etruscus and S. hundsheimensis. Unfortunately, only three deciduos teeth of S. jeanvireti are known and they do not show distinctive morphological and morphometric features (Guérin, 1972(Guérin, , 1980. The maximal length of the D/3 from Monte delle Piche (L=41.5 mm) is close to the minimal values of D. megarhinus (L=41.5-51 mm) and S. jeanvireti (L=40-47 mm) and to the maximal value S. etruscus (L=38-42 mm) given by Guérin (1980). Moreover it falls into the dimensional range of S. hundsheimensis from Untermassfeld (L=39-44 mm) given by Kahlke (2001a) and it is longer than the specimens of S. hund sheimensis from Vallonnet (L=28.50-32.87 mm) given by Lacombat (2006).
In conclusion, the specimen MPUR 1515 from Monte delle Piche displays morphological similarities with S. etruscus but it is not compared with S. jeanvi reti because of the scarcity of data. Therefore, a sure attribution to S. etruscus cannot be confirmed.
Description: The rhino mandible MPUR 1516 (Fig. 4A) lacks of fragments of the vertical rami, part of the incisor corpus and the left P/2. The specimen is covered by a black, hard and relatively thick sediment. The lower border of the mandible is convex and the height of the horizontal ramus decreases below the premolars. The vestibular groove is marked and deep in M/3, while it is less deep in the other teeth. Labial cingula are absent. M/2 has a mesial cingulum and P/2 has a weak mesial cingulum. The lingual valleys have a V-shaped morphology in all the teeth.
Remarks: The mandible from Monte delle Piche differs from those of D. megarhinus from Montpellier (NMB MP1, MP1031, MP94, MP858, MP822) in which the lower border of the mandible below the molar portion is linear and an inflexion point is present at the level of M/1 (Fig. 4). The vestibular groove in D. megarhinus is broader and shallower than in the specimen from Monte delle Piche; vestibular, mesial and distal cingula can be also observed on the teeth of D. megarhinus. Furthermore, the lingual valleys in D. megarhinus have usually a U-shaped morphology (Guérin, 1980). In S. jeanvi reti from Monte San Pietro (MGC 9354), Capannoli (IGF 1449v) and Vialette (NMB Vt621 and Vt627), the lower border of the mandible appears less convex than in the specimen from Monte delle Piche (Fig. 4). Moreover, weak vestibular cingula can be recognised in the teeth of S. jeanvireti from Capannoli, Monte Pulgnasco (MGC 9352) and Vialette. In S. etruscus from Upper Valdarno and Olivola mesial cingula occur in M/2 and M/3, while vestibular cingula are generally absent or are represented by a more o less marked extension of the mesial cingulum. In S. etruscus from Upper Valdarno, the lower border of the mandible is regularly convex as well as in the specimen from Monte delle Piche (Fig. 4). The horizontal rami of three mandibles from Pietrafitta (MPLBP no code, probably n°583, n°584 and n°585 in Mazza et al., 1993) display a convex lower border. The mandibles of the Etruscan rhino differ from those of S. hundsheimensis from Isernia la Pineta (MPI 33083), Contrada Monticelli (MSTB no code), Torrente Stirone (MPP no code) and Hundsheim (NHMW 2013/0282/0001) which display a more slender mandible with a straight ventral border (Fig. 4). The lingual valleys in the molars of S. hundsheimensis from Contrada Monticelli (MSTB 30600, 30600a, no code) are broader than in those from Monte delle Piche, and the vestibular groove in M/1 appears more marked. The lower molars from Hundsheim displays mesio-vestibular and distal cingula; moreover, lingual cingula characterised by small cuspules are also present. Mazza & Varola (1999) didn't reported the presence of lingual cingula on the specimens from Contrada Monticelli, but a small lingual cingulum can be observed on a M/1 (MSTB 30600).
The dimensional characters of the teeth and the tooth row of the specimen from Monte delle Piche are not conclusive for an attribution to S. etruscus because they fall into the dimensional ranges of the four considered species (Table 1). The ratio diagram of the dental measurements reveals that the Monte delle Piche specimen is proportionally quite different from D. megarhinus, S. jeanvireti and S. hund sheimensis considering the mean values given by Guérin (1980) (Fig. 5). It resembles the mean of S. etruscus given by Guérin (1980) in the proportion of the tooth row and the mean of S. hundsheimensis given by Lacombat (2005Lacombat ( , 2006 in the proportion of the teeth (with the exception of P/2 and M/1). The proportion of the specimen from Monte delle Piche is partially close to that of the maximal values of S. etruscus given by Guérin (1980); a similarity with the mean of S. etruscus from Senèze can be also noticed (Fig. 5).

Micropalaeontological analyses
Micropalaeontological analyses revealed the presence of some ostracod valves and foraminiferal tests in the sediment that covers the vertebrate remains.
The MPUR 999 is barren of foraminifera, whereas in the MPUR 1515, just one test, referable to the planktonic species Globoturborotalita woodi, was found.
Recently, Bellucci et al. (2011) reported the presence of an incisor fragment of hippopotamus in the Middle Villafranchian assemblage of Costa San Giacomo (Anagni Basin). According to these authors, the remain was found "in the field where most of the Costa San Giacomo FU bones were previously collected" (Bellucci et al., 2011, p. 8) and they pre-dated the occurrence of hippopotamus at approximately 2.0 Ma.

Biochronology of Stephanorhinus etruscus in Italy
Stephanorhinus etruscus occurred in Italy during the latest Pliocene in the site of Villafranca d'Asti and during the earliest Pleistocene at Montopoli (MN16) (Guérin, 1980;Pandolfi, 2013), but it was more common during the second half of the Early Pleistocene, in particular in the Upper Valdarno, Olivola and in the Tiberino Basin (Falconer, 1868;Mazza, 1988;Petronio et al., 2003Pandolfi, 2013). Fortelius et al. (1993) also reported the first occurrence of the Etruscan rhino in the localities of Palaia (Lower Valdarno, Tuscany) and Città della Pieve (Perugia). Nevertheless, the remains from Palaia (a fragmentary nasal bone IGF 5566v and an almost complete mandible IGF 13091) display morphological and morphometric features close to D. megarhinus (Azzaroli, 1962;Guérin, 1980;Pandolfi, 2013). Unfortunately, the attribution of the specimen from Città della Pieve cannot be confirmed because it is part of a no-defined Private Collection (Fortelius et al., 1993, p. 67), it was not figured by the authors and information about the stratigraphic level was not reported.
The last occurence of S. etruscus is relatively controversial. Several small sized remains from the latest Early Pleistocene of Europe were usually ascribed to S. etruscus, S. cf. S. hundsheimensis or to S. hundsheimensis. According to  the latest Villafranchian Italian rhinoceroses from Pirro Nord (ca. 1.5 Ma), Madonna della Strada (ca. 1.3 Ma) and Imola basin (latest Villafranchian) can be ascribed to S. etruscus. The remains from Pietrafitta (latest Early Pleistocene, Farneta FU) were ascribed to S. cf. S. hundsheimensis by Mazza et al. (1993) and were related to S. etruscus by Van der Made (2010) and . However, the specimens from Pietrafitta are morphologically and morphometrically closer to S. etruscus than to S. hundsheimensis and can be ascribed to the Etruscan rhino despite the dental proportions are quite different (see Mazza et al., 1993). The hemimandibles from Pietrafitta display a more sligthly convex lower border of the horizontal rami than in S. hundsheimensis and the height below the first molar is shorter than in the Toula's species.
Therefore, the presence of S. hundsheimensis or related forms during the latest Villafranchian in Italy is not supported by clear evidence and the last occurrence of S. etruscus can be placed at the end of the Villafranchian. Nevertheless, S. etruscus was reported during the early Galerian in Spain where it seems to be a long-lived species (Cerdeño, 1993;Van der Made, 2010). The presence of the Etruscan rhino can be confirmed at Huescar (around 0.9 Ma) (Cerdeño, 1993;Van der Made, 2010) and at Atapuerca TD4, TD6 and TD8 (Brunhes-Matuyama transition) ( Van der Made, 1998, 1999. The earliest Italian record of S. hundsheimensis is coeval with those from other European localities such as Vallonnet (France) and Untermassfeld (Germany), while at Vallparadís (Spain) S. hundsheimensis was also recorded in the ancient levels of the fossiliferous section dated around 1.5 Ma (Kahlke, 2001b;Moullé et al., 2006;Madurell-Malapeira et al., 2010).

Stratigraphy of Monte delle Piche and stratigraphic position of the fossil mammal remains
Micropalaeontological analysis suggests that the specimens from Monte delle Piche were deposited in two different levels that were characterised by two different sedimentary environments: specimen MPUR 999 in the continental deposits and specimens MPUR 1516 and MPUR 1515 in marine sediments.
In agreement with the absence of foraminifera and the presence of Cyprideis sp. as well as the conglomerates and sand found on the bone, the specimen MPUR 999 (femur of hippopotamus) was deposited in fluvial sediments which underlies the volcanic tufs in the stratigraphic succession of Monte delle Piche (Fig. 1). The fluvial sediments in which the hippopotamus was collected could be correlated with the fluvial deposits of the "Paleo Tiber" described by  in the adjacent areas of Monte delle Piche and in the Ponte Galeria Formation (>0.9 Ma) (Fig. 7). Carboni & Iorio (1997) found Pleistocene continental deposits (fluvial-marsh silt) that overlap Gelasian marine sediments in the boreholes from the Portuense district near the Monte delle Piche site (Fig. 7). Ultimately, according to Fig. 1, as reported by Ponzi (1858), fluvial conglomerates are deposited below the volcanic levels in the Monte delle Piche succession, which suggests an age earlier than approximately 0.6 Ma. Indeed, volcanoclastic elements are extensively present in all of the stratigraphic successions of the area from ca. 0.6 Ma (cfr. Karner et al., 2001;Sottili et al., 2010). Based on these observations, the fossiliferous level with the hippopotamus could be chronologically related to a time span between 0.9 and 0.6 Ma. This age is in agreement with the presence of the hippopotamus in Italy represented by the species H. antiquus (Fig. 7;. Marine ostracods and foraminifera (from infralittoral to circalittoral zones) are documented in the sediments collected from specimens MPUR 1516 and MPUR 1515. They suggest that these vertebrate remains were deposited in a marine environment. Marine deposits were reported by Ponzi (1858) at the bottom of the succession of Monte delle Piche (Fig. 1). The marine sediments, in which the specimens MPUR 1515 and MPUR 1516 were collected, are represented by black and red silty deposits. A depositional level with the same chromatic and sedimentological characteristics as the sediment covering the rhino mandibles was recognised by Carboni & Iorio (1997) in the boreholes from Trastevere and Portuense districts, which was located on the Tiber River right bank near the Monte delle Piche (Fig. 7). Based on the presence of Globigerinita umbilicata and Globorotalia oscitans, these sediments were assigned to the Gelasian stage (Carboni & Iorio, 1997) (Fig. 7). On the contrary, boreholes located on the Tiber River left bank (Valco San Paolo and EUR districts) revealed marine sediments containing Globorotalia aemiliana that were ascribed to the Piacenzian stage (Carboni & Iorio, 1997). Consequently, Carboni & Iorio (1997) have identified the Piacenzian in the boreholes of the Tiber River left bank and the Gelasian in the deposits from the right bank. Both of these deposits as well as the deposits at Monte delle Piche are covered by continental sediments, indicating the presence of an erosive phase (Carboni & Iorio, 1997).
In the sediments sampled from MPUR 1516, Neo globoquadrina pachyderma sinistral coiling was found. In the Mediterranean area, the first occurrence of N. pachyderma is dated to the Upper Gelasian (Iaccarino et al., 2007). Consequently, it is possible to assign these deposits to an age close to or younger than the Gelasian, thus agreeing with the data of Carboni & Iorio (1997) for the Trastevere and Portuense adjacent areas. In addition, in the boreholes carried out in the surrounding areas, marine sediments from ages after the Gelasian are completely lacking (Carboni & Iorio, 1997); in contrast, an erosion surface is present (Carboni & Iorio, 1997). The presence of Globoturborotalita woodi (the last occurrence was during the upper Tortonian) in the sample MPUR 1515 and of Globorotalia miozea (the last occurrence was in the lower Tortonian) in the sample MPUR 1516 together with the presence of a hemimandible of Acerorhinus sp. (Pandolfi et al., 2013b), indicates a reworking of Miocene elements. On the other hand, Carboni & Iorio (1997) also found reworked Mio-Pliocene foraminifera in the boreholes drilled in the Portuense, Trastevere, Valco San Paolo and EUR districts. Based on the collected samples, the marine deposit of Monte delle Piche would correlates with the Gelasian. Nevertheless, the Monte delle Piche Member, which correlates the first half of the Calabrian, was enstablished by Conato et al. (1980) on marine sediment outcropping in this locality. A late Early Pleistocene is therefore suggested for the sediment in which the rhinoceros specimens were collected.This is in agreement with the occurrences of S. etruscus in Italy. The length of M/1 and M/2 of the specimen MPUR 1516, as well as the dimensions of P/2 and P/3, are close to the maximal values of the species given by Guérin (1980) ( Table 1) and would suggest a correlation with the first evolutionary stage of the species (Early-Middle Villafranchian, approximately correlated with the Late Piacentian-Gelasian, Fig. 7); nevertheless, the proportion of the teeth resembles that of S. hundsheimensis given by Lacombat (2006) and of S. etruscus from Senèze (Fig. 5).

Conclusions
Several mammal remains have been collected in the area of Rome since the beginning of the XIX century. They are housed in the collections of the MPUR, but they usually lack any stratigraphic data.
In the present work, the mammalian remains collected at Monte delle Piche are revisited in order to define their stratigraphic provenance and chronological position.
Micropalaeontological analysis and stratigraphy suggest that there are two fossiliferous levels at Monte delle Piche: -The younger level, in which Hippopotamus sp. and ostracods are recorded, is attributed to a time interval between approximately 0.9 and 0.6 Ma. This time span is represented by fluvial conglomerates and sands attributed to the Ponte Galeria Formation. Hippopotamus was present in Italy from the end of the Villafranchian to MIS 4/3. -The older level, in which S. etruscus is recorded, is attributed to the second half of the Early Pleistocene; this age is supported by foraminiferal data (Neogloboquadrina pachyderma sinistral coiling) and by the biochronology of the Etruscan rhino. The latter species occurred in Italy during the beginning of the Villafranchian, but it was more common during the late Villafranchian. It has never been reported in the Roman Basin where typical Villafranchian taxa are very rare (Caloi et al., 1998;Di Stefano et al., 1998). Indeed, only an isolated molar of Mammuthus meridionalis (recorded in Italy until the early Galerian; Gliozzi et al., 1997; was previously reported in the Roman Basin, in a coastal marine deposit at Monte Mario (late Early Pleistocene) (Ponzi, 1878;Di Stefano et al., 1998;Milli & Palombo, 2005;Kotsakis & Barisone, 2008;Cosentino et al., 2009). Instead, typical Villafranchian taxa were collected around the Roman Basin in the lignite beds outcropping at Castel San Pietro (Rieti, Middle Villafranchian) and in a lacustrine deposit at Capena (about 40 km north of Rome, latest Villafranchian) (Tuccimei, 1891(Tuccimei, , 1893(Tuccimei, , 1898Petronio, 1979;Kotsakis, 1988;Kotsakis & Barisone, 2008;. The finding of rhinoceroses in the marine deposits at Monte delle Piche could be explained with a transportation from nearby land. The preserved status of the mandible of MPUR 1516 suggests a rapid and complete burial of the specimen; the bathymetric data given by ostracods and foraminifera indicate that the mandible was deposited far from the coast. Deposition in a marine environment followed by a rapid burial of the mandible of MPUR 1516 can be confirmed by the characteristics of the sediment encompassing the mandible. It appears as a hard crust that was only disaggregated in acetic acid, revealing an early diagenesis. This process can be attributed to a localised alkaline condition generated by decaying organic matter (i.e., the rhinoceros), which permitted the calcite precipitation and consequent cementation of the sediment (Bathurst, 1975).