CARNIVORA FROM THE EARLY PLEISTOCENE OF GRĂUNCEANU (OLTEŢ RIVER VALLEY, DACIAN BASIN, ROMANIA)

from the Early Pleistocene of Grăunceanu (Olteţ River Valley, Dacian Basin, Romania)


IntroductIon
The Olteţ River Valley in Romania occupies a strategic position for understanding mammalian dispersal patterns to, from, and within Eurasia across the middle-late Villafranchian transition (Terhune et al. 2020;Curran et al. 2021). This region is located just south of the southern Carpathians in the Dacian Basin, and multiple localities in this area have yielded fossils dated to the Pleistocene. The sites in the Olteţ River Valley (ORV) have been intermittently excavated since about 1960, producing extensive collections of large mammals (Necrasov et al. 1961;Bolomey 1965;Rădulescu & Samson 1990Rădulescu et al. 2003). Of the known sites in this area, Valea Grăunceanului (Grăunceanu herein) is the richest. The location of this site, along with the locations of other sites discussed herein are shown in Fig. 1.
The carnivoran assemblage of Grăunceanu is, as shown herein, exceptionally rich, equaling in species richness better known sites such as Saint-Vallier, Senèze, and Dmanisi. The carnivorans of Grăunceanu were first discussed by Bolomey (1965). She identified Nyctereutes megamastoides, Ursus etrus-cus Cuvier, 1823, Meles sp., Hyaena perrieri Croizet & Jobert, 1828, Felis (Lynx) issiodorensis Croizet & Jobert, 1828, Megantereon megantereon Bravard, 1828, Homotherium crenatidens Fabrini, 1890 in the material available to her. It is not clear what Bolomey (1965) meant by Felis sp., but all the other taxa she listed have been identified herein as well, albeit in some cases under slightly different names. It should be noted that of the four Grăunceanu specimens figured by Bolomey (1965) The Pliocrocuta cranial fragment of her fig. 6 has not been relocated for this study. A preliminary list of carnivoran taxa was given in Terhune et al. (2020). Some reidentification of material has taken place subsequently and the present taxon list does not precisely match any previously published.

GeoloGIcal settInG
The site of Grăunceanu is situated in the Olteţ River Valley (ORV) of Romania, approximately 175 km west-northwest of Bucharest (see Fig. 1 for geographic position). The site is located in the Dacian Basin, which is delimited to the north and west by the Carpathian Mountains and to the south and east by the Danube Valley and Dobrogea Plateau. Deposits in the ORV are attributed to the Tetoiu Formation (Andreescu et al. 2011), the sediments of which are primarily of a sandy-pebbly facies and are fluvio-lacustrine in origin. The Tetoiu Formation extends from the base of the Pleistocene (2.588 Ma) to as young as ~1 Ma (Andreescu et al., 2011). During the Villafranchian (~3.5-1.0 Ma; Rook and Martínez-Navarro, 2010) this region was dominated by the presence of the Dacian Lake, which was connected to the Pannonian and Euxinic lakes to the northwest and east, respectively. Today, the Olteţ River, a tributary of the Danube River (via the Olt River) crosscuts the valley.

MaterIal and Methods
All material described and figured herein is housed in the Institutul de Speologie "Emil Racoviţă" in Bucharest, Romania. Additional material from Grăunceanu, not described in this study but included in some calculations of faunal similarity, is housed in the Muzeul Olteniei in Craiova, Romania. These specimens were inventoried by the research team but not examined in person by the first author. The material there adds specimens to the existing species, but does not expand the list of species. All photographs were taken by the authors. Measurements were taken to the nearest 0.1 mm using digital calipers except as noted and are reported in mm. Comparative measurement data are from the files of L.W. Some additional data were kindly provided by Dr. Saverio Bartolini-Lucenti, University of Florence, Italy.
Abbreviations: VGr: prefix for Valea Grăunceanului specimens housed in the Institutul de Speologie "Emil Racoviţă"; MO: prefix for Muzeul Olteniei specimens; NRM-PZ: acronym for specimens from the paleozoological collections of the Swedish Museum of Natural History, Stockholm; c/C, canine; p/P, premolar; m/M, molar; d, deciduous tooth. Upper dentition identified by upper case letters, lower dentition by lower case letters. L, greatest mesiodistal length; Lt, greatest mesiodistal length of m1 trigonid; Lpa, length of paraconid of m1; Lpr, length of protoconid of M1; W, greatest transverse width.

systeMatIc paleontoloGy
Order Carnivora Bowdich, 1821 Suborder Caniformia Kretzoi, 1943 Family Canidae Batsch, 1788Genus Vulpes Garsault, 1764 Vulpes alopecoides (Del Campana, 1913)  Background. This small species of Vulpes from western Eurasia has been described under a variety of names, including V. alopecoides, V. praecorsac, and V. praeglacialis. Bartolini-Lucenti & Madu-rell-Malapeira (2020) have, however, convincingly argued that these are synonyms and that the valid name for all of them is V. alopecoides. The species is known from a number of sites ranging from early Villafranchian (MN 16) to late Galerian (MN 18), the time at which the extant Vulpes vulpes appears. Vulpes alopecoides is recorded (under the various names mentioned above) from numerous sites ranging from Georgia in the east to France and Spain in the west. Among Romanian sites, Terzea (1996) records its presence at Betfia-2 but it is not mentioned by Bolomey (1965), nor in other, later lists of specimens from the ORV.  Material: VGr.2334, laterally crushed and deformed cranium ( Fig. 2a, b). In dorsal view retaining parts of both nasals and left maxilla, frontal, and parietal. In ventral view retaining a very fragmented basicranium and most of the left and right maxillae including complete right M1 and partial left M1 (buccal half) and complete M2. The markings on this specimen clearly identify it as coming from Grăunceanu.
Comments. Both in measurements and morphology this specimen conforms to published material of V. alopecoides. The cranium itself is too damaged to provide any additional information on the species.

Genus Nyctereutes Temminck, 1838
Nyctereutes megamastoides (Pomel, 1842)  Background. This is the common raccoon dog species that is found at a number of sites across western Eurasia. It has a distribution similar to that of V. alopecoides, extending from Georgia in the east to France and Spain in the west (Bartolini-Lucenti 2018). Its temporal range extends in Europe from early Villafranchian or possibly late Ruscinian (see Bartolini-Lucenti et al. 2018) until at least the end of the middle Villafranchian (MN16/17 transition). It is preceded by the less derived Nyctereutes donnezani. A slightly more derived form, Nyctereutes vulpinus, may be present at Saint-Vallier (Soria & Aguirre 1976;Monguillon et al. 2004;Bartolini-Lucenti et al. 2018). Bolomey (1965) recorded Nyctereutes megamastoides as present at Grăunceanu and La Pietriş in Romania. Comments. The large sample of Nyctereutes from Grăunceanu shows great variability in both morphology and metrics (e.g., Fig. 3d-i). This raises questions regarding the taxonomic distinction of N. vulpinus from N. megamastoides (Soria & Aguirre 1976;Monguillon et al. 2004), but this issue requires a deeper analysis than is possible here. Grăunceanu may be one of the latest occurrences of N. megamastoides in Europe. A closely related, possibly conspecific taxon, Nyctereutes sinensis, survives later in China (Tedford & Qiu 1991  Background. Ursus etruscus is the common bear of the western European Plio-Pleistocene. It had its main center of distribution in western and southwestern Europe (France, Spain, Italy), but is also known from Romania, Bulgaria, Georgia, and as far east as China (Wagner 2010). Bolomey (1965) listed U. etruscus from Grăunceanu.
Comments. The identification of these specimens presents no difficulties. They are typical of U. etruscus in morphology and proportions. They are slightly smaller than average among a sample of U. etruscus from western Europe and may possibly belong to a female individual or individuals.

Comments.
Meles thorali differs in a number of respects from extant M. meles. The extant species is somewhat smaller on average. In M. thorali the buccal margin of M1 is shorter that the lingual margin and the latter bulges out beyond the buccal margin both mesially and distally. In the extant species they are more similar in length. In these characteristics the Grăunceanu specimens match M. thorali. Bolomey (1965) reported two specimens of Meles from Grăunceanu, a left mandibular corpus with very worn teeth, which is likely to be specimen VGr.2323 herein, and a second corpus that cannot be identified based on Bolomey's (1965, p. 84) description. Bolomey (1965) presented evidence for derived characteristics in the Grăunceanu Meles, including a trigonid that is tall and longer than the talonid. It is not clear how this was measured, but there is no certain evidence for this characteristic in the material available here. Bolomey also stated (p. 84): "The shape of the mandible separates it significantly from Meles meles and from M. thorali..." (our translation from the original German). Applied to VGr.2323, available to Bolomey, this is correct to some degree by comparison both with the illustration in Viret (1950) and with the other Grăunceanu specimens; VGr.2323 is more robust, with a deeper corpus and larger and deeper coronoid process. However, VGr.2323 represents an aged individual with very worn teeth and changes with usage in life may have remodeled the mandible in this specimen. The other specimens available are young, with little dental wear, show no such changes, and do not differ from published specimens of M. thorali.

Subfamily Lutrinae Bonaparte, 1838
Genus Lutraeximia Cherin et al., 2016 Lutraeximia sp. Comments. Lutrinae are generally rare in Villafranchian faunas and Grăunceanu is no exception. Only two teeth of this taxon are known from there. In both morphology and metrics the M1 shows similarities to that of Le. simplicidens from Voigtstedt illustrated by Cherin (2017, fig. 2), although the Grăunceanu tooth is more worn. The P4 is damaged but what is available is also similar to the Voigtstedt specimen in size and morphology. The main distribution of the genus is in western Europe (Germany, England, Italy) and the Grăunceanu specimens, which are among the oldest Lutraeximia are the first recorded from Romania.
Suborder Feliformia Kretzoi, 1945 Family Hyaenidae Gray, 1821 Genus Pliocrocuta Kretzoi, 1938 Pliocrocuta perrieri (Croizet & Jobert, 1828)  Background. Pliocrocuta perrieri was the first extinct hyaenid to be described, at a time when most members of the family were assigned to the genus Hyaena. It was not reassigned to Pliocrocuta until much later (Kretzoi 1938). It shares a number of features with the extant brown hyaena, but these are plesiomorphic and several apomorphic features link this species to the derived hyaenas in the Crocuta clade (Werdelin & Solounias 1991). Pliocrocuta perrieri is the common Eurasian hyaenid in the Pliocene and earliest Pleistocene and is replaced by Pachycrocuta brevirostris at the 'Pachycrocuta-event' around 2 Ma (Martínez-Navarro 2010; Rook & Martínez-Navarro 2010). Bolomey (1965) reported Hyaena perrieri from Grăunceanu.
Comments. The juvenile corpus VGr.2314 is clearly hyaenid and can be assigned to Pl. perrieri based on the size of the dp4 (Baryshnikov & Averianov 1995). The two adult corpuses match other Pl. perrieri in size and dimensions, as well as in morphology.
Genus Pachycrocuta Kretzoi, 1938 Pachycrocuta brevirostris (Gervais, 1850)  Background. Pachycrocuta brevirostris is another species with a Eurasian distribution, although unlike P. perrieri it was also distributed across Africa (Ewer 1954;Werdelin 1999). The African forms are sometimes classified as a separate species, Pachycrocuta bellax (Ewer 1954). The author of the name Pachycrocuta was previously thought to be Aymard (1846) but Alba et al. (2015) have shown that no such publication exists and that the earliest record of the name is Gervais (1848Gervais ( -1852. Pachycrocuta may have an African origin and appeared in and spread rapidly throughout Eurasia around 2 Ma (the aforementioned 'Pachycrocuta-event'). It became extinct sometime in the late Middle Pleistocene.
Comments. It is important to note, first of all, that the exact provenance of this specimen is not known. This means that it may come from a slightly younger deposit, suggesting that the 'Pachycrocuta-event' horizon is not present at Grăunceanu. The cranium is large, although not among the largest Pachycrocuta, which are found in the later part of the early Pleistocene. Nevertheless, in size and metrics, as well as in morphology, it falls well within the range of Pa. brevirostris. In Romania, the species has been reported from La Seci (Rădulescu & Samson 1990) and from Fîntîna Alortiţei (Rădulescu & Samson 1991, 2001. Family Felidae Batsch, 1788 Genus Lynx Kerr, 1792 Lynx issiodorensis (Croizet & Jobert, 1828)  Background. There is general agreement that Lynx issiodorensis is the sister taxon to the four extant species of Lynx (Werdelin 1981;Lorfèvre 2009). Its antecedents are less clear, but appear to lie in North America, possibly with 'Felis' proterolyncis Savage, 1941or 'Felis' longignathus Shotwell, 1956. The evolution of the genus Lynx subsequent to L. issiodorensis is less clear, with several fossil species named, e.g., Lynx spelaeus, Lynx shansius, and Lynx variabilis (Lorfèvre 2009) but no consensus has been reached regarding their validity or place in the phylogeny. Lynx issiodorensis appears in the Early Pliocene of Europe (Montoya et al. 2001), becomes widespread at the beginning of the Villafranchian (Late Pliocene) in Europe and survives there approximately to the end of the Villafranchian (late Early Pleistocene), when it is ostensibly replaced by Lynx spelaeus and subsequently by the extant Lynx pardinus. In Romania, L. issiodorensis is also known from La Pietriş (Bolomey 1965).
Material: VGr.2332, right mandibular corpus with c-m1 (Fig. 8a-c); (Bolomey 1965, fig. 5). The canine was separated from the corpus but has been glued back. Comments. The Grăunceanu specimen is relatively large, with wide premolars, but in morphology is typical of L. issiodorensis. The illustration in Bolomey (1965, fig. 5) shows that the canine was still attached to the corpus at that time.
Genus Puma Jardine, 1834 Puma pardoides (Owen, 1846)  Background. This is a problematic taxon that has been published under a variety of names since Owen (1846) described the first specimen, a single lower carnassial from the Red Crag Formation, southeast England. Material currently thought to pertain to this species was described by Viret (1954) as Panthera schaubi and later ascribed to Viretailurus schaubi by Hemmer (1965). In subsequent publications Hemmer (2001;Hemmer et al. 2004) has argued that the species belongs in the genus Puma. Addressing this phylogenetic issue is beyond the scope of this paper and we here accept the generic attribution to Puma. The species ranges temporally from the later part of the early Villafranchian to at least the latest Villafranchian and possibly into the earliest Galerian. Its geographic distribution extends from westernmost Europe (Britain) to Mongolia in the east. Comments. In both morphology and metrics, the Grăunceanu specimens are a good match for P. pardoides from other sites in Eurasia (Hemmer 2001;Hemmer et al. 2004;Madurell-Malapeira et al. 2010;Cherin et al. 2013). Variation within the Grăunceanu sample appears to be about as great as between the remaining known specimens, but the overall variation is quite small in most characters. This is the first record of the species from Romania.
Two specimens of Megantereon from Grăunceanu stand out as exceptional. One is the nearly complete cranium VGr.2378 and the second the mandibular corpus fragment VGr.2355 that includes a dp4 that represents the first definitely identified dp4 of Megantereon. Because of the special interest of the latter specimen, it is here be described in detail and compared with the dp4 of Smilodon sp., the sister taxon to Megantereon.
The cranium VGr.2378 adds to the small collection of crania of Megantereon from the Villafranchian of Europe. The bone is cracked into many pieces and the frontal in particular is flattened, but the overall shape of the cranium is intact. In dental metrics it is approximately average among European Megantereon, being a little larger than specimens from Les Étouaires and Perrier (France), approxi-mately equal in size to specimens from Saint-Vallier (France), and smaller than specimens from Olivola and Figline (Italy).
In dental metrics the juvenile specimen VGr.2335 is within the range of European M. cultridens (Lm1 VGr.2335 18.7; mean for European sample = 19.8, range = 18.3 -23.0, N = 14) although on the smaller side. The permanent carnassial is only partly erupted and seemingly unworn. The paraconid is shorter and lower than the protoconid (Lpa = 8.3; Lpr = 9.9). There is no metaconid or talonid. The enamel is vertically wrinkled as well as being grooved horizontally ventral to the carnassial notch. In the basal part of the leading edge of the m1 there are very minute serrations, similar to but slightly more prominent than those sometimes seen in unworn teeth of extant Felidae.
The dp4 (Ldp4 16.2M Wdp4 5.9; Ltdp4 14.4) is 87% the length of the m1. Unlike the m1, the paraconid and protoconid are nearly equal in length (Lpa = 6.2; Lpr = 6.3). The protoconid is nevertheless the taller of the two. The metaconid is set very slightly lingual to a line through the protoconid shear facet. The talonid is short, more crest-than cusp-like and comes to a point at the distal end of the tooth. The dp4 of Smilodon (Fig. 11D-E) is similar in structure to that of Megantereon. Differences include that the paraconid is shorter and lower relative to the protoconid in Smilodon than in Megantereon. Also, in Smilodon the metaconid-talonid complex is more mesiodistally compressed than in Megantereon. A comparison of the length of this dp4 (Ldp4 mean of left and right = 20.0) with that of m1 (Lm1 mean of left and right = 28.2; measured on screen, based on a CT-scan of the specimen) shows it to be 71% of the length of m1, i.e., substantially less than in Megantereon. Thus, either the dp4 of Smilodon has evolved to be relatively smaller or the m1 to be relatively larger, or both.

Background.
The genus Homotherium is present on all continents with the exception of Australia and Antarctica. It is characterized by moderately en-larged but very narrow and serrated upper canines, a very long, low cranium, and a slightly sloping back, somewhat reminiscent of hyaenas. The sister taxon to Homotherium as presently known is Lokotunjailurus from the late Miocene of Kenya (Werdelin 2003) and this, together with a very early presence of Homotherium in the Lonyumun Member of the Koobi Fora Formation, Kenya (Werdelin & Sardella 2006;Werdelin & Lewis 2013) suggests an African origin. However, an equally early record of Homotherium from the Odessa Caves, Ukraine (Sotnikova et al. 2002)

means that this issue remains open.
In Europe, Homotherium has been at times subdivided into two species, with Homotherium crenatidens Fabrini, 1890 being an earlier, slightly more robust form and H. latidens a later, more gracile form. Current consensus, however, is that only one species, H. latidens, encompasses all European material (Antón et al. 2013). Bolomey (1965) recorded H. crenatidens from Grăunceanu.

dIscussIon
In this discussion we will place Grăunceanu in the context of a number of other western Eurasian carnivoran assemblages of approximately similar age (see Fig. 13 for an overview of absolute and relative stratigraphic positions of these assemblages.). These include the somewhat older Saint-Vallier, France, the approximately coeval Senèze, also from France, the slightly younger Olivola, Italy and Dmanisi, Georgia, and Liventsovka in Rostoc Oblast, Russia. The last mentioned of these sites may, like Senèze, be approximately coeval with Grăunceanu but span slightly more time. Another site of interest in this time span is Pantalla, Italy (Cherin et al. 2017(Cherin et al. , 2022, which is the only site among the ones listed here that shares the presence of Lutraeximia with Grăunceanu. However, only five species of Carnivora are known from Pantalla and it is therefore not included in the analysis. In addition to comparing faunas and discussing the reasons for similarities and differences, we can also make a crude comparison between Grăunceanu and Saint-Vallier in the individual representation of the different taxa thanks to the detailed study of material from recent excavations at the latter locality by Argant (2004).
The list of carnivoran taxa present at each of these sites is given in Table 1. It is immediately obvious that in terms of number of species, Grăunceanu is as rich or richer than those betterknown sites. One species, H. latidens, is known from all six sites. This may be, either because it was very common and had a wide distribution, or because it is very easy to identify from even small fragments, being the largest felid species in Eurasia at the time. Two further species, Pl. perrieri and L. issiodorensis, are known from five of the six sites. All families are known from across Eurasia (not surprisingly). Mustelidae is the rarest of the families present at these sites, with only Grăunceanu and Saint-Vallier  Tab. 1 -A compilation of carnivore taxa found at some representative Early Pleistocene sites in western Eurasia. The columns are organized based on the position of the sites on a west to east axis.
Dates are very approximate. The absence of a west-east gradient in taxon representation is noteworthy. A distinct temporal difference can, however, be discerned between the two youngest sites (Olivola and Dmanisi) and the remaining sites. Non-Grăunceanu data from Viret (1954), Sotnikova et al. (2002), Argant (2004), and Bartolini-Lucenti et al. (2021). Species names in bold are those not present at Grăunceanu. The two Nyctereutes species are considered equivalent in the calculations. Although the presence of Pachycrocuta at Grăunceanu is uncertain it has been included in the calculations.
having substantial samples of a mustelid (M. thorali). This may be due to a size bias against smaller taxa at many of the sites.
Grăunceanu is most similar to the somewhat older Saint-Vallier, with nine of the 14 species at Saint-Vallier shared (nine of 11 if the three rare mustelids at Saint-Vallier are discounted). Five species at Saint-Vallier and two at Grăunceanu are not shared. Of these, three of the former and one of the latter are small mustelids that are exceedingly rare in the fossil record. In addition, one is the species of Nyctereutes, and the distinction between N. megamastoides and N. vulpinus is debatable. There remain two taxa at Saint-Vallier that are not shared with Grăunceanu, Acinonyx pardinensis and Chasmaporthetes lunensis. Both of these are reconstructed as open habitat pursuit predators and the difference between the sites may be indicative of less open habitat at Grăunceanu. The absence of Canis spp. at Grăunceanu may be another indication of more closed habitat there compared to, e.g., Senèze and Olivola, both of which include Canis etruscus in their faunal lists.
Looking geographically, it is clear that Grăunceanu is more similar to Senèze and Saint-Vallier to the west than it is to Dmanisi and Liventsovka in the east, despite the distances being approximately the same. This fits with a suggested opening up of the landscape from east to west and through time (Curran et al. 2021 and citations therein). Given the absence of Acinonyx, Chasmaporthetes, and Canis at Grăunceanu an argument can be made that there was either less open habitat or that, if present, open habitat was more distant from the core habitat represented by the non-carnivoran remains at the Romanian site. The absence of open habitat carnivorans at Grăunceanu disagrees with the reconstruction of the habitat based on evidence from a variety of sources ) and with the overwhelming dominance of open-habitat prey species (e.g., Eucladoceros, Rucervus, Equus) in the assemblage. One possible explanation may be related to the reconstruction by Curran et al. (2021) showing that there was also some locally available water source with nearby woodlands available, perhaps in the form of a gallery forest lining a river (i.e., the paleo-Olteţ). This discrepancy between the paleoecological signals from predator vs. prey species at Grăunceanu cannot be resolved based on available data; analysis of δC 13 isotope values in carnivoran teeth, and comparison to existing isotope data for Grăunceanu ) may assist in a resolution. As Hopley et al. (2022) have shown, habitat inferences for carnivorans are not always what they are traditionally made out to be.
A comparison between Grăunceanu and Saint-Vallier in terms of number of specimens and MNI (minimum number of individuals mea- sured as greatest number of duplicated elements in a species sample) shows considerable similarities but also some glaring differences (Table 2). It should be noted here that this comparison concerns the excavations at Saint-Vallier by Faure and Guérin in 1996-1999 as reported by Argant (2004) and not the earlier excavations reported on by Viret (1954). The latter author did, however, provide a semi-quantitative assessment of commonness and rarity in the older excavations at Saint-Vallier (Viret 1954, p.184), which matches the relative position of species in Argant (2004) quite well (Table  2). Note also that the data of Argant (2004) include postcranial elements whereas the Grăunceanu data concern craniodental material only. Overall, a list of number of specimens per species in descending order shows the two to be quite similar. Indeed, they would be nearly identical except for the large number of specimens of A. pardinensis and Pl. perrieri at Saint-Vallier. Thus, N. megamastoides, U. etruscus, and M. thorali are common at both sites (not taking into account the putative species difference in Nyctereutes between the sites), followed by the large cats. A similar pattern is seen when looking at MNI, although in this case Saint-Vallier has a more balanced fauna than Grăunceanu, with three species with equal MNI. Grăunceanu, on the other hand, is heavily dominated by N. megamastoides, even without including the Muzeul Olteniei specimens.
As noted above, the material from Grăunceanu ostensibly includes both Pl. perrieri and Pa. brevirostris. However, the exact provenance of the latter specimen is unclear given that it is recorded as coming from Valea Homorecia, an unknown site likely to be geographically close to Grăunceanu. Prior publications (Rădulescu & Samson 1990 noted the presence of Pachycrocuta brevirostris at two alternative sites in the ORV, La Seci and Fîntîna Alortiţei, but no other publications list this taxon as being present either at Grăunceanu or the ORV more broadly. Unfortunately, no prior publications specifically describe the partial Pa. brevirostris cranium published here. Because of the uncertainty regarding provenance we therefore regard the attribution of Pa. brevirostris to Grăunceanu as tentative. The appearance of Pa. brevirostris is an important biostratigraphic marker event in the European Pleistocene at the middle-late Villafranchian transition (cf. Iannucci et al. 2021Iannucci et al. , 2023  Tab. 2 -A comparison of relative proportions of taxa between Grăunceanu and Saint-Vallier. Data for Grăunceanu are based on craniodental material presented herein; data for Saint-Vallier are from Argant (2004), based on excavations carried out in the 1990s. These data do not include the material described by Viret (1954) but, unlike the Grăunceanu data, do include postcrania. Despite these differences, the general overall similarity in representation at the specimen level (except for the two taxa not found at Grăunceanu) suggests that the comparison has some validity. Thus, Nyctereutes (disregarding species) is the most commonly recorded taxon at both sites, with M. thorali, P. pardoides, M. cultridens, and H. latidens also well represented at both sites. The main difference lies in the relative commonness of Pl. perrieri and U. etruscus at Saint-Vallier. The first two rows are the raw numbers of specimens for each taxon and site. The following two are the MNI numbers based on max duplication of element. The following four rows are these numbers converted to percentages of the total number. Note that the two taxa furthest to the right are not represented at Grăunceanu. In addition to this, Viret (1954) describes material of three small mustelid taxa that are not represented in the 1990s excavations at the site, nor at Grăunceanu (cf. Table 1). The specimen count of Nyctereutes megamastoides does not include specimens from Museul Olteni, Craiova.
within the Grăunceanu sequence. However, given the uncertainty regarding the provenance of the Pa. brevirostris specimen, it is perhaps more likely that Grăunceanu is slightly older than the 'Pachycrocutaevent' and therefore late middle Villafranchian rather than early late Villafranchian, as suggested by Rook & Martínez-Navarro (2010).

conclusIons
In terms of both specimens and number of species of carnivorans, Grăunceanu is one of the richest sites in the Early Pleistocene of Europe. We have here identified 11 species (or 10 depending on the status of Pa. brevirostris) in the Grăunceanu assemblage: 2 Canidae, 1 Ursidae, 2 Mustelidae, 2 Hyaenidae, and 4 Felidae. Things of note include the very large sample of N. megamastoides, the presence of the rare Lutraeximia sp., and the presence of a juvenile specimen of Megantereon cultridens.
Among more or less coeval coeval sites in western Eurasia, the carnivoran assemblage from Grăunceanu shares more taxa with sites to the west than with sites to the east, and displays particular affinities with Saint-Vallier. All, or nearly all, carnivoran species found at Grăunceanu are also present at Saint-Vallier. The sequences of abundance measured as either number of specimens or MNI are also similar, although evenness is much greater at Saint-Vallier, whereas Grăunceanu is heavily dominated by the very large sample of N. megamastoides.
For habitat reconstruction the species that are not present at Grăunceanu may be more informative than those that are. Thus, three taxa that have generally been reconstructed as open habitat forms, Canis spp., C. lunensis, and A. pardinensis are all absent from Grăunceanu and a fourth, Pl. perrieri, is rare compared to its abundance at Saint-Vallier. This suggests that there may have been less open habitat at Grăunceanu than at Saint-Vallier, or that open habitat was more distant at Grăunceanu. This is not entirely compatible with multiproxy habitat reconstruction of Grăunceanu , and is especially at odds with the dominance of open-adapted prey species (e.g., Eucladoceros, Rucervus, Equus) found in the Grăunceanu assemblage. The discrepancy cannot be resolved on the basis of present evidence but requires new data sources, such as isotope analyses of carnivore teeth from Grăunceanu and other sites.