The small mammals from Gratkorn: an overview

The rich and diverse fossil mammalian assemblage from Gratkorn (Middle Miocene, Austria) is of primary importance for the understanding of the faunal evolution in Central Europe. Besides large mammals, the fauna comprises: Schizogalerix voesendorfensis, Galericinae gen. et sp. indet., Desmanodon fluegeli, Dinosorex sp., cf. Myotis sp., “Cricetodon” fandli, Megacricetodon minutus, Eumyarion sp., Spermophilinus bredai, Blackia sp., Forsythia gaudryi, Albanensia albanensis, Muscardinus aff. sansaniensis, Miodyromys sp., Keramidomys sp., Euroxenomys minutus minutus, Prolagus oeningensis, cf. Eurolagus fontannesi and Ochotonidae indet. Based on the degree of corrosion on the dental elements and the presence of pellets, most, but not all, of the material is tentatively interpreted as a result of accumulation by nocturnal raptors. In addition to the information provided by the lower vertebrates and the molluscs, which occur in abundance in the same thin fossil-enriched layer, the mammal fauna gives a mixed picture of the environment (basically forested vs. open landscape). This could indicate the presence of different microhabitats around the excavation place, but may also be a taphonomical artefact based on various different agents of accumulation contributing to the thanatocoenosis. Nevertheless, the extreme quick accumulation of the fossils provides an exceptional windows in the late Sarmatian s. str. ecosystems.


Introduction
The late Middle Miocene is fascinating for the study of terrestrial palaeoecosystems. Following the long period of the mid-Miocene Climatic Optimum, drastic climatic changes induced important faunal re-arrangement and migration within Europe (e.g. Böhme 2003;Maridet et al. 2007). Rich and diverse fossil faunas are thus of primary importance for the understanding of these processes. The late Middle Miocene is well recorded in various Iberian basins, but complete faunas of this period are rare in Central Europe. Therefore, the recent discovery of an exceptional fossil assemblage of macro-and microvertebrates in Gratkorn (early late Sarmatian, Austria, Styria; Gross et al. 2007), allows a fresh look at the palaeocommunity at higher latitudes following the Mid-Miocene cooling (e.g. Gross et al. 2011). Continued work at the site has yielded new material, allowing us to refine previous conclusions on the taxonomy of the small mammals (rodents, insectivores, bats), which are presented in the first section of this contribution. In the second section, the biostratigraphic implications will be discussed. The third part of this article concentrates on the origin of the fossil accumulation processes, allowing a better view on the palaeoenvironment of Gratkorn (section "Biostratigraphical remarks").

Materials and methods
The fossils from Gratkorn are housed in the collections of the Universalmuseum Joanneum in Graz (Geology & Palaeontology), the Bayerische Staatssammlung für Paläontologie und Geologie in Munich, and in the collections of the Eberhard Karls University in Tübingen. SEM images were captured at the Biogeology and Applied Palaeontology Laboratory of the Eberhard Karls University at Tübingen, Germany.

Insectivores, rodents, lagomorphs and chiropters from Gratkorn: taxonomic notes
The insectivores, comprising two gymnures, one talpid and one soricid, and one single chiroptera have been described by Prieto et al. (2010a). The rest of the small mammal fauna comprises four species of cricetid rodents previously described by Prieto et al. (2010b), two glirids, four sciurids, one eomyid, all described by Daxner-Höck (2010), and one beaver (Prieto et al. 2014, this issue). The lagomorphs, all pikas, are diverse, with three species (Angelone et al. 2014, this issue). Gratkorn is the type locality of two species ("Cricetodon" fandli and Desmanodon fluegeli), plus a potential new species of a largesized pika (Ochotonidae indet.) never recorded elsewhere.

Eulipotyphla
Erinaceidae Schizogalerix voesendorfensis (Rabeder, 1973) The species is the most abundant mammal in Gratkorn, and fossils are regularly found in the locality. Besides a large number of mandibles ( Fig. 1), upper jaws and isolated teeth, a newly discovered fragmentary skull completes the inventory (UMJGP 204693; Fig. 2). The specimen has been prepared in its ventral part. It preserves only the palate region; the part posterior to the crista transversa has been lost. The left P3-M1, as well as the right canine and M1-M2, are conserved in situ. The skull has been crushed as shown by the different occlusal orientations of the tooth rows.
European Miocene gymnure skulls are a rarity. The giant insular form Deinogalerix is known from several skulls (Butler 1980;Villier 2010;Villier et al. 2013). Ziegler (1983) describes in detail an exceptionally complete specimen of Galerix exilis from Steinberg (Germany, Badenian), a species of which partial skulls were previously known from the French fissure fillings of La Grive (Butler 1948). In addition, several skulls in excellent state of preservation from the sites of Cerro de Batallones are under study (Spain, late Vallesian; L.W. van den Hoek Ostende, personal data) (See also section "Semi-articulated specimens and proposed evidences of fossil pellets of birds of prey" and related figures).
Originally, the deciduous dentition was only represented by a single d4 in Gratkorn, but now two upper milk premolars (D3) are also available (Fig. 5s,t).
Taxonomical note: Schizogalerix voesendorfensis is morphologically closer to Parasorex socialis and S. pasalarensis than to any other species. Traditionally, the Iberian late Middle-Late Miocene Galericini findings are identified as Parasorex (Van den Hoek Ostende 2001). However, these species closely resemble their Central European counterparts, which are placed within the genus Schizogalerix, for instance because of the shape of the mesostyle in the upper molars . This implies that a revision of these forms is necessary to assess the geographical range of both genera, as well as of its constituting species, at the end of the Middle Miocene, and assure that the biogeographical patterns found are not a taxonomic artefact.

Galericinae gen. et sp. indet.
This species is only represented by a single M1 which is somewhat larger than Schizogalerix voesendorfensis, and has a different configuration of the mesostyle. Prieto et al. (2012a: 230) noticed that this molar closely resembles the one classified as Parasorex sp. from the Middle Badenian of Sámsonháza 3 (Hungary, Nógrad County). However, as we only have one element and, given the taxonomic confusion in the Galericini mentioned above, we hesitate to assign the Gratkorn specimen to a genus. More material is needed to clarify its taxonomic position.
Gratkorn is the type locality of the species. Desmanodon fluegeli shares characteristics found in the Anatolian lineage D. minor-D. major but is not directly linked to it. Nevertheless, it may represent an eastern immigrant, which is interesting, as Schizogalerix also has its origin in Anatolia (Engesser -1980;de Bruijn et al. 2006). The most interesting new specimen is a complete humerus (Fig. 3).
Two bat molars have so far been discovered. Only one broken tooth (UMJGP 211015) is added to the known sample.

Rodentia
Cricetidae "Cricetodon" fandli Prieto, Böhme and Gross, 2010 Gratkorn is the type locality of the species, and the holotype is a very well-preserved skull with the mandibles in anatomic connection. A second skull has been found ( Fig. 4b; currently under study) and, although being more fragmentary, provides new morphological information because it presents parts which were covered by sediment in the holotype (see also section "Semiarticulated specimens and proposed evidences of fossil pellets of birds of prey" and related figures). Taxonomical note: The generic status of the early species of Byzantinia and Hispanomys, as well as that of some species of Cricetodon is under discussion (e.g. Prieto et al. 2010b;López-Guerrero et al. 2013). As a consequence, related species may have been assigned to different genera, and thus have not been compared to one another. In the following, we discuss the relationships of the Cricetodontini listed in Table 1 (for the discussion of other taxa, see Prieto et al. 2010b).
Hispanomys cf. bijugatus is described from Gaweinstal based on one M1. Harzhauser et al. (2011) recognised some similarities with C. klarankiae, but refrained to link the molar to this species without a clear explanation. A similar observation was previously made by Prieto et al. (2010b: 427), who hypothesised a relationship between C. klarankiae and H. bijugatus, without entering into details. Unfortunately, López-Antoñansas and Mein (2009) did not take the Hungarian species into account in their detailed review of H. bijugatus. The two species are of comparable size, but C. klarankiae has longer m2/ M2. A detailed morphological comparison is indeed needed. For this reason, the Hungarian species is classified as "C." klarankiae, and the species from Gaiweinstal as ? "C." klarankiae in Table 1. It is hypothesised here that these species belong to a single lineage, but the inclusion of ? "C." klarankiae could be questioned.
In addition, the newly discovered and unpublished Cricetodon sp. sample from Vărciorog (Romania, Bihor District; Hír 2010) might be related to these species.   The Cricetodon sp. from Tăşad (Hír et al. 2001) and Mátraszőlős 1-3 Kókay 2004, 2011) are also characterised by long ectolophs, but their relationship with "C." klarankiae cannot be satisfactorily established (Prieto et al. 2010b). They also share some characteristics with Cricetodon hungaricus from Hasznos (Badenian, Kordos 1986). Whether these forms belong to another lineage, or are related to "C". klariankae, remains an open question. Depending on its outcome, the evolution of the lineage "C." fandli-"C."klarankiae as proposed by Prieto et al. (2010b) may have to be re-considered.
On the other hand, the following taxa cannot be related to the Gratkorn form: -Hispanomys sp. from Nebelberg TGL III was originally assigned to Cricetodon by Rummel and Kälin (2003:140), who underline the similarity of the species with C. jumaensis from the German fissure fillings of Petersbuch. This form has been excluded from the "Cricetodon" lineage by Prieto et al. (2010b) based on the absence of complete ectolophs on the M1. In addition, the m1 lacks the posterior metalophulid. -Cricetodon sp. from Bullendorf (Daxner-Höck 1996, here as cf. Cricetodon sp.) needs to be described. cf. Byzantinia sp./div. sp. from Comăneşti 1 ) is/are too poorly documented.
In conclusion, it has to be stressed that the lack of sufficient Cricetodontini material in the Central/East European Sarmatian s.str. and early Pannonian does not allow to precisely decide on the relationship of most of the finds. For the moment, the taxonomic position of "C." fandli from Gratkorn as stated by Prieto et al. (2010b) is considered the most viable working hypothesis. Daxner, 1967 Small-sized Megacricetodon species have been reported from a wide range of localities at the end of the Middle Miocene and the beginning of the Upper Miocene. Their taxonomy is often controversial (see . Here, the proposal of Joniak (2005) is followed, who confirms the validity of M. minutus, a species defined on a few specimens from Brunn-Vösendorf (Daxner 1967). However, detailed study is needed to clarify the phylogeny of these forms. The main characteristics of the species from Gratkorn ( Fig. 5e-j), apart from its size, are the anterolophulid being frequently not directly connected to the anteroconid in m1, as shown in Fig. 5g, and the forked anterolophule which connects to both cusps of the divided anterocone in M1. Several mandibles are known from Gratkorn: two mandibles with m1-m3 (UMJGP 204742;Fig. 5e;UMJGP 204194;Fig. 5f) and one mandible with m2-m3 (UMJGP 210799).

Megacricetodon minutus
Democricetodon sp. nov. (sensu Kälin and Engesser 2001) This species is related to the Democricetodon species found in Nebelbergweg (Switzerland, Middle/Late Miocene transition). Kälin and Engesser (2001) considered this to be a new form, but refrained from naming it, awaiting further investigation by G. Daxner-Höck. The entire dentition has been previously described in Prieto et al. (2010b). The additional material, partially presented in Fig. 5a-d, confirms the taxonomic homogeneity of the Democricetodon sample (see also section "Semi-articulated specimens and proposed evidences of fossil pellets of birds of prey"). The main characteristics of the species are: its relatively small size, the M1 having an anterocone subdivided into two adjoining cusps by a superficial groove on its anterior wall and the presence of a low but long labial spur of the anterolophule on some specimens (present in the two figured M1); and a long mesolophid on the lower molars.
Additional material: UMJGP 204683: several associated remains of this species are preserved on a slab (Fig. 6c). These include a fragmentary maxillary with M1-M3 (M1: 1.98×1.36; M2: 1.45×1.41; M3: 1.11×1.22) which belongs to a relatively old specimen. It lies close to a tibia. The dental morphology is difficult to access, because the specimen has been covered by glue that obscures some details. Its study would require removing the glue without breaking the fragile specimen. An upper right incisor with part of the premaxilla (Fig. 6b)  Taxonomical note: Because of taxonomic ambiguities and the limited number of specimens, the species from Gratkorn could not be identified with confidence in previous works. The specimens added here, although providing new morphological information, do not help any further. Interestingly, both lower and upper incisors are now assignable to Eumyarion sp.. The lower incisor is slightly damaged (Fig. 6a2), but two striae are observable on the anterior part of the enamel. Similarly, on the upper incisor (Fig. 6b2), two striae run along the enamel, but they are placed closer to the lateral margin of the incisor. Similar striae were also observed in E. leemanni from the Vallès-Penedès Basin (Casanovas-Vilar 2007).
The previously described Spermophilinus tooth sample consisted of two d4 and one M3; its specific assignment was made on a dimensional basis. Only one m1,2 is added here ( Fig. 5m; UMJGP 211006). The tooth belongs to a juvenile specimen (not completely erupted at the time of the death of the animal).

Blackia sp.
No additional material is reported herein. The single tooth found from Gratkorn, a d4, shows the typical morphology of Blackia, but is very large in comparison to B. miocaenica.
The rare monospecific genus Forsythia is fortunately recorded by one mandible in Gratkorn. Pliopetaurista and Forsythia have very similar tooth patterns and differences are best shown in the upper dentition. However, the assignment of the Austrian specimen to this genus is based on characteristics of the lower tooth row provided by Daxner-Höck (2010: 515).

Albanensia albanensis (Major, 1893)
No additional dental material is reported here. The fossils from Gratkorn, including mandibles, fragmentary skulls, and post-cranial material, are among the most complete published record available for the genus. Also, A. sansaniensis (Lartet, 1851) from the French locality Sansan is well documented, and a mandible, a maxillary and numerous isolated teeth have been described (Ginsburg and Mein 2012). Albanensia fossils are recognised in the literature as subspecies of A. albanensis (A. a. quirensis) are only known by scarce material from the Vallès-Penedès Basin (Catalonia, e.g. Mein 1970), and thus of questionable validity. Albensia grimmi, originally only known from its holotype (found at the Middle Miocene German locality Marktl), is now recorded by dozens of teeth from various Pannonian localities, also including sites from Austria (Daxner-Höck 2004a) and Hungary (Kretzoi and Fejfar 2004). The descriptions of the cranial and post-cranial material, together with comparisons to fossil and extant flying squirrels, will be published separately.

Gliridae
Muscardinus aff. sansaniensis (Lartet, 1851) No additional material is reported here. The small tooth sample from Gratkorn (three specimens) is considered to belong to a relative advanced (large) species in the lineage Muscardinus sansaniensis-M. vallesiensis (Daxner-Höck 2010) Miodyromys sp. The scare material did not allow Daxner-Höck (2010) to provide any assignment at the species level. Although new dental elements are now available (p4 and m1), a definite identification of the species is still hazardous. Compared to the previously published m2, the two lower molars presented here are a bit more complicated because they both have a very short extra ridge between metalophid and centrolophid. The connection mesolophid/entoconid is interrupted in one m1. In the same specimen, the posterotropid connects to the posterolophid. The two p4 have five transversal crests.
Previously, a d4 and a m3 have been described from Gratkorn. The m1,2 (UMJGP 211011; Fig. 5n; 0.84× 0.81) presented here has five long transversal lophids and a continuous longitudinal crest. The crest-like entoconid is directed forwards, and the posterolophid ends shortly before it. Daxner-Höck (2010) places the species near to K. mohleri or K. ermanorum. Unfortunately, the new molar cannot lead to a more precise taxonomic assignment.

Euroxenomys minutus minutus (von Meyer, 1848)
This small-sized beaver is known from a few specimens, including a fragmentary skull with the two associated mandibles in almost anatomic connection that have recently been excavated. This exceptional find is described in Prieto et al. (2014, this issue). Euroxenomys minutus is a common element of the Badenian-Pannonian fauna in Austria.

Prolagus oeningensis (König, 1825)
This species is the most abundant pika in Gratkorn, and upper and lower jaws have been excavated in the locality. It shows more primitive characters than western and central-western European populations of P. oeningensis.

cf. Eurolagus fontannesi (Depéret, 1887)
The remains consist of isolated teeth and an upper jaw. This rooted ochotonid is very similar but does not completely fit the morphometric characteristics of E. fontannesi (Angelone et al. 2014, this issue). Indeed, the taxonomical status and the phylogeny of the Middle and Late Miocene rooted ochotonids are in need of revision. Pending such a revision, our identification is somewhat tentative.
Only two isolated upper molariforms are available from Gratkorn. This ochotonid is characterised by its large size, and ever-growing robust teeth. To our knowledge, nothing equivalent has ever been previously recorded in Europe.

Biostratigraphical remarks
Arrangement of the localities provided in Tables 1, 2, 3 and 4 The tables summarise the small mammal fossil record of various Central-Eastern European localities which are chronologically close to Gratkorn. The relative order of the localities, between and even within basins, is partly unresolved. Therefore, the ordering proposed here should be regarded as a working hypothesis; future work in those sites may change the relative age of the sites.

Pannonian localities of the Vienna and Styrian basins
The Pannonian localities of the Vienna and Styrian basins are ordered following the division of Papp (1951). The relative positions of, respectively, Vösendorf/Borský Sv. Jur, Atzelsdorf/Gaweinstal, and Bullendorf/Mataschen are arbitrary. The stratigraphic position of Magersdorf is uncertain.

North Alpine Foreland Basin (NAFB)
The mammal localities of the NAFB are not directly correlated to the Central/Eastern Parathethys subdivision. Hillenloh, Nebelbergweg, Nebelberg TGL III and Hammerschmiede 1-3 are traditionally correlated to MN 9, but this is not satisfactory for our purposes, because it cannot be clearly demonstrated at present if these localities are equivalent to Sarmatian s.str. or to Pannonian (see below). Moreover, in the MN system (e.g. Mein 1999), MN 9 is characterised by the entrance of the hipparionine horses, which provides a problem for localities of small mammals in which no large mammal remains have been found ). Therefore, we assigned these localities in our tables to the 'Middle to Late Miocene transition'.
Giggenhausen and Anwil belong to the Deperetomys hagni taxon range zone of the Swiss local biozonation (Kälin and Kempf 2009), which ranges from 13.8 to 13.2 My (equivalent Badenian).

Late Sarmatian s.str. localities
The ordering of the succession of the Felsőtárkány localities (Hungary) follows the proposal of Hír and Kókay (2010). Some mammal taxa are considered important stratigraphic markers.
First of all, Microtocricetus molassicus was often used as typical species of the MN 9. This would imply that the localities containing this species correlate with Pannonian deposits. This was, however, challenged by Prieto and Rummel (2009) and Prieto et al. (2011) based on a cricetid rodent lineage and the migration of a gymnure (see below). Arranging the localities from the NAFB and Hungary using Microtocricetus leads to placing Nebelbergweg and Nebelberg TGL III below the first occurrence of the species (in Hungary, Felsőtárkány 3/8), near to Felsőtárkány 3/2. This proposal is in apparent conflict with the occurrences of the eomyid rodents Eomyops catalaunicus (Late Miocene species; Daxner-Höck and Höck 2009) in Nebelbergweg and E. oppligeri (Middle Miocene species) in Felsőtárkány 3/8, for instance. Originally, both species of Eomyops were recognised in the Swiss locality (Kälin and Engesser 2001). However, Prieto (2012) considered the sample as monospecific, with a relatively large intra-specific variability. Actually, it cannot be ascertained that the material from Nebelbergweg and Felsőtárkány 3/2 would represent different species. Because the intraspecific variability of Eomyops is accessible only in few localities, and with the lack of general recognition of the taxonomic homogeneity of the genus in Nebelbergweg, it is better not to use the genus here for fine biostratigraphic purposes.
Finally, Prieto and Rummel (2009) proposed a biostratigraphical sequence based on the size increase of the cricetid rodent Collimys. These authors recognise a lineage C. hiri (Hammerschmiede)-C. longidens (Nebelbergweg)-C. dobosi (e.g. Felsőtárkány 3/2). While clear differences are observed between C. hiri and C. dobosi from Hillenloh, C. longidens and C. dobosi (Hungary) show a larger intraspecific variability, and both species differ only slightly in size and morphology. Arranging Hammerschmiede at the basis of this lineage implies that this locality is older than Felsőtárkány 1, 2, 2/3 and 2/7, which do not contain Microtocricetus. This arrangement could evidently be discussed but the following points restrain us: (1) the value of the lineage C. hiri-C. dobosi is in need of confirmation, because regional variation cannot be ruled out, and (2) the appearance of Microtocricetus in a restricted area is documented in the Felsőtárkány Basin.
In short, based on small mammals only, different approaches result in different arrangements of the late Sarmatian localities, and in our current state of knowledge, the stratigraphical sequence is unresolved. Moreover, it cannot be excluded that a part of the localities from the NAFB belong to the base of the Pannonian. Also, Felsőtárkány 3/8 and 3/10 could belong to the base of the Pannonian (Pannon A or B) (Hír and Kókay 2010). It has been chosen here: (1) to correlate these localities to the Sarmatian s.str., mainly because of the presence of Sarmatian molluscs in Felsőtárkány 3/8 and 3/10; and (2) to arrange the localities with Microtocricetus together. In order to provide a better arrangement in the tables, Bełchatów A is included in this group (association Microtocricetus + Collimys), as well as Opole 1 and 2, based on the presence of Collimys as proposed by Prieto and Rummel (2009).
The late Volhynian/Bessarabian localities Tauţ and Comăneşti 1 correlate with the central Parathethys stages to the late Sarmatian s.str. , and are thus slightly younger than Gratkorn.

Early Sarmatian s.str.
Tăşad belongs to the Mohrensternia Zone of the early Sarmatian s.str. The exact correlation of the faunas from Subpiatră is not clear in this scheme, and they are here placed at the Badenian/Sarmatian s.str. transition

Biostratigraphic position of Gratkorn based on small mammals
For the integrative studies dealing with the dating of the locality, we refer to Harzhauser et al. (2008) and Gross et al. (2011).
The genus Eurolagus was thought to be a marker of MN7+ 8, but recent studies bring the temporal range of Eurolagus into question, and thus its value as a marker (Angelone et al. 2014, this issue and references therein). For the moment, we prefer not to use cf. Eurolagus fontannesi for fine biostratigraphical purposes.
Similarly, Desmanodon fluegeli cannot be integrated in the Anatolian lineage D. minor-D. major without reservations, and the use of the taxon for correlation purposes is limited. According to Prieto et al. (2010a), Gratkorn represents the first occurrence of Schizogalerix voesendorfensis. New investigations (Table 3) indicate that a related form might be present in the late Badenian of Hungary. Recent advances in the study of the late Middle Miocene fissure fillings from Petersbuch (J. Prieto, personal data) suggest the presence of the genus in Germany. These new discoveries show that the galericine record is even more complicated than indicated by Prieto et al. (2011) in their discussion of this record in the NAFB.
Democricetodon sp. nov. (sensu Kälin and Engesser 2001): Democricetodon brevis and Megacricetodon germanicus from Felsőtárkány 2/3 and 2/7 (Hír 2006: plate II, figs. 5, ?6, 10 and 21) are most probably related to this form and are listed as Democricetodon cf. sp. nov. in Table 1. The same observation can be made for Tăşad. In Subpiatră, the specimens assigned to Democricetodon freisingensis and D. brevis by Hír and Venczel (2005) share some characteristics with this form, but any conclusion is hazardous at the present state of knowledge. Democricetodon zarandicus resembles the Nebelbergweg Democricetodon in many points , and might possibly be closely related to this form, as well as Democricetodon sp. from Felsőtárkány 1 and 2. Even though we cannot name the species, it is clear that the morphospecies from Gratkorn is a common faunal element of the Sarmatian s.str.
Among the Gliridae and the Sciuridae, Daxner-Höck (2010) considers Muscardinus aff. sansaniensis, Albanensia albanensis and Forsythia gaudryi to be good biostratigraphic markers. Based on the evolutionary stage of Muscardinus, she concludes that Gratkorn is somewhat younger than La Grive and Anwil, and about the same age as Giggenhausen and Kleineisenbach (NAFB, Germany). Here, part of this proposal cannot be followed because, as stated above, the Swiss and German localities are close in age (e.g. Prieto 2007), and most probably even Badenian. A revision of the Muscardinus material listed in Table 2 is of interest to better understand the evolution of the taxon. Gratkorn most probably shows one of the last records of Albanensia albanensis and Forsythia gaudryi in the studied area.
On the other hand, Gratkorn records one of the earlier Megacricetodon minutus for Central/Central-East Europe, although we must acknowledge that the taxonomic identity is quite ambiguous. As underlined by Prieto et al. (2010b: 428) Considering these localities, the figured Cricetodontini, especially Cricetodon cf. pasalarensis, are of interest for our purpose. As stated above, the lineage "C." fandli-"C." klarankiae as proposed by Prieto et al. (2010b) is a working hypothesis which permits to relatively date Gratkorn, but has to be confirmed by further finds, and by the study of its relationship with, for instance, Anatolian forms.
In short, the small mammal fauna does not contradict the conclusions provided by the integrative studies available for Gratkorn. While a correlation to the early late Sarmatian s.str. can be justified, it has to be stressed that the lack of sufficient knowledge regarding the evolution of the fauna in Sarmatian s. str. in Central Europe makes it difficult to definitively demonstrate this correlation based on rodents and insectivores.
Origins of the small mammal accumulation and indirect evidence of the rodent activity Prieto et al. (2010b) and Gross et al. (2011) discussed the origin of a few dozen of small mammals and ectothermic vertebrate remains (skulls, jaws, extremities) discovered in the upper part of the palaeosol from Gratkorn, and concluded that it represents an accumulation of fossil owl pellets, at least in part. This type of accumulation is well known for fissure and cave infillings, where thousands of remains can be accumulated in a very short period (e.g. Mellet 1974;Andrews 1990;Kowalski 1990). For instance, the potential of bone accumulation by raptors is illustrated by Guérin (1928) who showed that Tyto alba can accumulate around 1,200 small vertebrates in 6 weeks. In the case of flood plain sediments, the picture might be more complicated because of time averaging and the multiple sources of the fossil remains that may ultimately define the accumulation.
A detailed taphonomic study of Gratkorn would require a separate paper, but as an understanding of the origin of the thanatocoenosis is vital for any palaeoenvironmental reconstruction, some indicative data follow.

Corrosion on small mammals teeth
The presence/absence and amount of corrosion are used, among other evidence, as proxies to determine potential predators (e.g. Andrews 1990;Fernández-Jalvo and Andrew 1992;Fernández-Jalvo et al. 1998;Laudet 2000). The intensity of the corrosion on bone and especially teeth, the frequency of the damage and the relative abundance of skeletal elements allow the recognition of damage categories, ranging from predators that induce little damage (e.g. Tyto alba) to predators in which the digestion violently disturbs the remains (e.g. Carnivora).
Corrosion on molars and premolars has been scored here in four categories (absent, light, moderate and strong) in order to get a general idea of the trend of corrosion in the sample. The moderate and strong grades are presented in Figs. 7a-c, and d, respectively. Light grade indicates that marks are observed that only superficially attack the enamel.
Chiropteran teeth do not show any corrosion traces, and the same goes for the teeth of the Gliridae, Sciuridae and Eomyidae. The single M1 of the Galericinae gen. et sp. indet. shows relatively strong spurs (Prieto et al. 2010a: fig. 2k). Among the Dinosorex sp. teeth, only the isolated P4 is damaged, but clearly less than the former erinaceid molar (Prieto et al. 2010a: fig. 2l). The skull of the small-sized beaver Euroxenomys minutus minutus does not show corrosion that can be assigned to the effect of predation. In the second individual recorded in Gratkorn (Prieto et al. 2014, this issue), the damage deserves a more in-depth analysis.
The molars conserved in situ in the partial skull of Eumyarion sp. are slightly corroded (Prieto et al. 2010b: fig. 4). Minor damage is also observed on the enamel of the incisor of the mandible in Fig. 6a1, a2.
Much more material is available for the each of the other taxa. The corrosion grade can be different in each teeth of a single jaw, and therefore we consider each teeth independently.
The rare species cannot provide a clear idea on the potential predator(s).
Most commonly, the fossils are either not or little corroded. It would indicate that the accumulation is mostly due to the action of night-active birds of prey (which have less aggressive gastral fluids than their diurnal counterparts; e.g. Andrews 1990), or/and natural death.
In addition, it is interesting to notice that the gastric juice damages the specimens in different ways. While most of the teeth are corroded on their occlusal surface (e.g. Figs. 7c, d), a representative part of the sample shows marks preferentially on the labial part (e.g. Figs. 7a, b) indicating that the lingual side was protected from the juice. This effect would result from the ingurgitation of complete skull with mandibles by the predators, and thus a reinforced protection of the lingual area (Andrews 1990;Casanovas-Vilar et al. 2008a).
In conclusion, a large part of the sample seems to have been accumulated by nocturnal raptors which produce little corrosion on the cheek teeth, as is the case for Tyto alba or Asio otus in the recent fauna. The most corroded specimens might have been provided by predators which induce moderate to strong dissolution of the dental parts, as occurs with Bubo bubo or Athene noctua, for instance. The rare extreme strong corrosion may also be the result of digestion by mammal predators, diurnal raptors or others.
This proposal excludes, because of their size, the beaver specimens. The remains of the large-sized Albanensia albanensis-skulls, mandibles, and post-cranial of several individuals-taxa were excavated from a very restricted area, and no additional finds have been recovered elsewhere in the fossil-enriched layer(s) since then. This might indicate a different cause for the fossil accumulation. Flying squirrels might be victims of raptors (e.g. Airapetyants and Fokin 2003), but Albanensia is a large species, which would require a different agent of accumulation than the small rodents and insectivores. Potential predators of large size have been reported in recent fauna, as for instance by Fan and Jiang (2009), who report black-crested gibbons attacking large Petaurista in China or in Taiwan, where the most common prey of the mountain hawkeagle might be the giant flying squirrel Petaurista philipensis (Sun et al. 2009).

Semi-articulated specimens and proposed evidences of fossil pellets of birds of prey
Evidence of intact fossil pellets is, because of their fragility, extremely rare to our knowledge. For instance, Meng and Wyss (1997) tentatively recognise such fossils from the Late Paleogene beds from Inner Mongolia. Transport, even across small distances disintegrates the fragile rejecta and leads to the dispersal of the bony remains (e.g. Terry 2004). However, bone concentrations of only a few individuals have been recovered and interpreted as such (e.g. Czaplewski 2011). Semi-articulated and exceptionally wellpreserved specimens are also supposed to derive from such regurgitates (e.g. Tobien 1977;Kälin 1993).
In the following, we list bone accumulations from a very restricted area (few cm 2 ), which are prepared in such a way that at least part of the original bone arrangement is conserved.
The remains belong to one "Cricetodon" and at least three "Schizogalerix" individuals. Only one long bone is present but not identifiable. In the sediment plate, the labial part of the M2 of Schizogalerix, as well as the P3, show light corrosion spurs. In contrast, the trigonid of the m1 in the mandible is well corroded. Lingual parts of the molars are corroded, while the labial sides are intact. The isolated specimens show all light to more rarely moderate damages, most often on the occlusal part.
Almost only cranial parts are found in this specimen. This reflects, at least in two cases, the situation in recent pellets: (1) during the nidifications, adults might consume only the skulls, while the rest of the animal is destined to the young (Vein and Thévenot 1978); and (2) when prey is in abundance, the skull might be preferentially eaten (Raczynski and Ruprecht 1974).

UMJGP 204728
Böhme and Vasilyan (2014, this issue: fig. 3i, j) report a partial skeleton of a small Anguidae Ophisaurus spinari. Since this taxon is fossorial, they tentatively propose that the specimen is preserved in its own burrow. Actually, a right upper jaw with P3-M2 of Schizogalerix voesendorfensis comes from the three sediment plates holding the specimen and has been isolated. The teeth show consequent corrosion spurs on their whole surface, and thus indicate that at least this fossil derives from remains of a prey.

UMJGP 204730 and 204692
The amphibians are also documented by semi-articulated pieces. Partial skeletons of a crocodile newt Chelotriton aff. paradoxus (UMJGP204730) (frontal, trunk vertebra, ribs) and Pelobates sanchizi (UMJGP204692) (maxilla, premaxilla, sphenethmoid, squamosal, scapula, trunk and sacral vertebrae), is presented by Böhme and Vasilyan (2014, this issue). This specimen underlines the fact that part of the small vertebrate accumulation might derive from "natural" death. The spatefoot toads are fossorial amphibians, and the presence of an articulated skeleton of Pelobates sanchizi indicates an apparently dead animal in its own burrow. Such burrows are documented in the fossil record (e.g. Martin and Bennett 1977;), but no direct evidence has so far been found in at Gratkornt.

UMJGP 204240
The holotype of "Cricetodon" fandli, a skull with the two mandibles in anatomical position and forelimb, wrist and hand bones, is figured in Prieto et al. (2010b). In addition, postcranial material was found in the vicinity of the specimen and most probably belongs to it. These are in need of detailed description. The skull does not show damage produced by gastric juices, but some original breakages are observed on the post-cranial. Whether the animal was a victim of predation or died naturally needs a more in-depth analysis.

UMJGP 204683
The remains belonging to Eumyarion sp. have been described in the "Insectivores, rodents, lagomorphs and chiropters from Gratkorn: taxonomic notes" of this paper (Fig. 6). The corrosion evidence demonstrates that the specimen was a prey.

UMJGP 211012
Several bone-enriched blocks originating from one accumulation place have been reserved for micro-CT analysis, and are not yet prepared. The most important material, as far as can be observed are the blocks rich in molluscs and the Celtis endocarps that are present. "Cricetodon" fandli is present in several plates with at least a fragmentary right mandible with m1-m2, a fragmentary left mandible with i + m1 and an in situ broken m2 (light corrosion spurs), a left fragmentary maxillary with M3, a left fragmentary maxillary with M1-M2, and a right upper jaw with M1-M3 (light corrosion spurs).
Democricetodon is also recorded by a right mandible with i-m1.
Much bony material is present, often fragmentary. In addition, osteoderms of Ophisaurus sp. have been recognised.
Synopsis concerning the small vertebrate fossil accumulation 1) Following the observations made on the corrosion traces on teeth and the evidence of pellets, it can be concluded that most of the small mammal remains derive from regurgitations of nocturnal raptors, but the influence of other types of predators is noticed. A similar origin is also proposed for part of the lower vertebrates. 2) This proposal cannot be demonstrated for the following taxa because of the lack of sufficient material and spurs: Muscardinus, Miodyromys, Spermophilinus, Forsythia, Keramidomys, Dinosorex, cf. Myotis.
3) While the pika might represent a representative part of the diet of the birds of prey, as shown, for instance, in the German fissure fillings from Petersbuch (Prieto 2007), the origin of the accumulation of Prolagus, cf. Eurolagus and Ochotonidae indet. cannot be confidentially enlightened here. 4) The "natural" death of some articulated specimens cannot be excluded, especially for lower vertebrates (amphibians and reptiles, indicated by their fossorial aptitude). 5) The accumulation of the beaver Euroxenomys-skull remains only-follows another pathway, probably closer to the accumulation factor linked to larger vertebrates. 6) Similarly, other factors may have been responsible for the presence of the large-sized squirrel Albanensia albanensis.
Indirect evidence of rodent activity Gross et al. (2011) reported gnawing marks on many tortoise plates and mammalian bones. This kind of bone altering allows diverse conclusions. Due to the usually facet-like surface of gnawing marks produced by rodents, these traces are readily distinguished from bite marks of other animals. Based on the shape and the dimensions of the grooves and facets, the originators can sometimes be identified even to the specific level (e.g. Maul 2001). In addition, conclusions can be drawn on the behaviour of the gnawing rodents, because among extant rodents, not all gnaw bones or at least not with the same intensity. Nowadays, there are some 'specialists' (porcupines, squirrels, rats) who gnaw bones more frequently than others (e.g. Klippel et al. 2007;Kibii 2009). In addition, the ways and aims of gnawing might be taxa-related. Klippel and Synstelien (2007) demonstrated that the brown rat gnawed on fat-laden cancellous bone, while the grey squirrel took more advantage in attacking the thicker bone cortices. Consequently, gnawing marks might contribute to taphonomic inferences, since the gnawed bones must have been accessible during the gnawing process, so they were neither covered by water nor by sediment, and in most cases were in a very advanced grade of putrefaction (see section "Time averaging"). A detailed study of these marks is ongoing, but, at present, it has not been possible to clearly observe brown rat-type modifications.
Time averaging Most small vertebrate finds, as well as invertebrates and Celtis endocarps, derive from the uppermost part of the fossilenriched palaeosoil (Layer 11b in Gross et al. 2011), a layer which indicates a rising of the water table, switching from alluvial to lacustrine conditions. The thickness of this layeraround 15 cm-indicates a relatively short time of accumulation on the geologic scale. The presence of pellet remains and semi-articulated vertebrates, because of their susceptibility to diagenetic processes, also indicates rapid burial of the fossils. The switch from alluvial to lacustrine conditions provided an ideal mixture for the accumulation and fossilisation of the specimens, the latter requiring a low energy environment with little disturbance (as in a lake), and providing a protecting layer above the fossil-enriched sediments.
Estimating the time of accumulation of the uppermost part of the palaeosol is much more problematic. Brand et al. (2003), in an actualistic experiment, come to the conclusion that the burial of small vertebrate remains is achieved from a maximum three months in aquatic conditions to over one year in soils. This gives an indication of burial times for single specimens, but not for the whole sample.
The gnawing marks observed on part of the larger bones occurring in the lower 40 cm of the fossil-enriched layer (Layer 11a in Gross et al. 2011) indicate that the remains were putrefied while the rodents damaged them. Klippel and Synstelien (2007) estimate that the post mortem interval for initial gnawing on human remains by the grey squirrel is over 30 months.
These observations underline the fact that a detailed taphonomic study of the small vertebrates from Gratkorn is needed, in order to better approach the accumulation process(es), and thus the deposit time of this exceptional locality. Anyway, the preliminary observations on small mammals certainly do not contradict the proposal of Gross et al. (2011), who suggest only a few decades for the accumulation of the layer 11, of which only a minor fraction would thus involve the small vertebrate-enriched part.
Impact of the sampling technics and accumulation agents on the relative abundance of the species Two techniques were used for sampling the micro-vertebrate remains from Gratkorn. First, surface sampling allowed the discovery of the pellets, semi-articulated specimens and larger fossils, as, for instance, skull fragments and mandibles. In contrast, screen-washing techniques resulted in the winning of most of the isolated teeth. This induces an underrepresentation of the small-sized fossils, and restrains us from giving an important weight to the relative abundance of the species.
In addition, while this abundance might be useful for palaeoecological reconstructions (e.g. Daams et al. 1988;López-Martínez 2001), the impact of the accumulation agent (s), the diversity and ecology of the predator(s) (nocturnal vs. diurnal, for instance), as well as, consequently, the size of the prey plays a role in the repartition of the taxa in the fossil assemblage, and the abundance does not reflect the original abundance of the species in the fauna (e.g. Andrews 1990).
As a guide, the representativeness of the taxa in the screenwashing is indicated in Table 5. Only Schizogalerix dominates the samples from both surface and wet collecting methods.

Insectivores
The assemblage consists of only Schizogalerix voesendorfensis, Desmanodon fluegeli, Dinosorex sp. and Galericinae gen. et sp. indet. The latter two are represented by just one individual. Indeed, the diversity is considerably lower than, for instance, at Anwil (13, Engesser 1972), Kleineisenbach (13, Prieto 2007), La Grive M (17, Mein and Ginsburg 2002), or Steinheim (9, Ziegler et al. 2005. Even in the late Sarmatian s. str., preliminary results show that the diversity is clearly higher in the rich Hungarian faunas (Tables 3, 4, 5). Prieto et al. (2010a) has already commented on the low α-diversity of the Gratkorn eulipothyphlans, concluding that there are taphonomical biasses. Although sample size will certainly play a role, the Gratkorn insectivore assemblage is by no means small, and Schizogalerix is even the most common micro-mammal of the fauna. Gymnures do not have spikes, and have played in some places an important role in the diet of birds of prey, as indicated by their overrepresentation in some fissure fillings from Germany (e.g. Ziegler 2005). Prieto (2007) observed that Parasorex socialis remains from Petersbuch 48 belong in large majority to young adults, and hypothesises that the animals were caught during their dispersion phase. Preliminary observations show that this picture cannot be applied at Gratkorn, where the proportion of adult Schizogalerix specimens is clearly higher, and might be closer to the results obtained in the floodplain locality of Kleineisensenbach.
The composition of the assemblage differs considerably from the late Middle Miocene insectivore assemblages of southern Germany (Ziegler 2006a). This fact probably results partially from the lack of sufficient Sarmatian s.str. fossil localities in the NAFB (see Table 1) and surrounding areas. Schizogalerix only enters that area near to the Middle-Late Miocene transition (Nebelbergweg, ?Aumeister; Prieto et al. 2011), the dominant Galericini in the late Middle Miocene still being Parasorex socialis. As stated in "Biostratigraphical remarks", this viewpoint might be nuanced in the future because new fissure fillings from Petersbuch probably contain Schizogalerix, but a detailed study is needed to confirm this observation, and a relative dating of the fauna cannot be provided at present. Galericini form the most common element in almost all insectivore assemblages of the Miocene (Furió et al. 2011), but the late Middle Miocene is a notable period of change, with the entrance of Schizogalerix in Europe. In this respect, the re-occurrence of Galerix in the NAFB ) is also notable, showing that environmental changes led to the flux of this tribe of gymnures.
Interestingly, Gratkorn is dominated by elements of Anatolian origin, viz. Schizogalerix and Desmanodon, which is also true for part of the rodents. This indicates environmental conditions favourable for Anatolian-related forms to extend their distribution further eastwards during the latest part of the Middle Miocene. Desmanodon has some incursions into the NAFB in the Latest Badenian/earliest Sarmatian s.str. (Prieto 2010; and see Table 4) but the species found in Kleineisenbach and Giggenhausen most probably had its origin in Hungary (J. Prieto, L.v.d. Hoek Ostende and J. Hír, personal data) and is most probably not related to D. fluegeli. Whereas Talpidae are usually considered good indicators of humidity, Desmanodon seems to take an exceptional position in the family. Van den Hoek Ostende (1997 noted that the genus was the only talpid to survive after the Early Miocene in the relatively open environment of the Daroca-Calatayud area of the Teruel Basin, and only went extinct when conditions became too dry. In the case of Gratkorn, the presence of Desmanodon as the only talpid is even more remarkable, as moles were quite common and diverse at the end of the Middle Miocene in the nearby NAFB (Ziegler 2006a) and surrounding areas (Ziegler 2003;Tables 4, 5).
Dinosorex has a wide geographical and stratigraphical range, and seems of limited palaeoecological use.
Although it is hazardous to draw conclusions on the basis of the absentees, the key to understanding the eulipotyphlan assemblage lies primarily in the species that were not found, either because of true absence or rarity. No true litter feeders (Uropsiline and Urotrichine moles and Soricidae) have been found. This would tie in nicely with the results from the sedimentology, suggesting moist soils in a rather open landscape. Also, burrowing talpids (Talpa, Proscapanus) have not been found. Although moist soils would allow these animals to make burrows, in the absence of sufficient litter production, the food supply in these burrows would be limited. Overall, the eulipotyphlans of Gratkorn suggest a relatively open and dry landscape.

Chiroptera
Regarding the few specimens recorded at Gratkorn, as well as their poor taxonomy, any conclusion is hazardous. In general, bats are much more abundant in fossil assemblages from cave environments (e.g. Rosina and Rummel 2012), and large taxonomic differences are observed when dealing with other types of localities (Sigé and Legendre 1983). Bats are usually rare in fluvio-lacustrine sediments, and in this respect Gratkorn is no exception. Only exceptional events lead to the conservation of representative samples (e.g. Messel: Storch et al. 2002;Anwil: Engesser 1972;Green River Formation: Simmons et al. 2008).

Rodents
Albanensia, Forsythia and Blackia are usually regarded as flying squirrels, although no postcranial material of any of these genera has been described to date. Thorington et al. (2005) have correctly pointed out that the dental features used to recognise fossil flying squirrels (such as the rugose enamel of their cheek teeth) also occur in tree squirrels, which, on the other hand, are very rare in the Miocene fossil record. It cannot be discarded that many purported Miocene 'flying squirrels' are in fact tree squirrels. Nevertheless, this would still indicate the presence of relatively dense forest environments when those taxa are found. Interestingly, the three species present at Gratkorn range in size from large (Albanensia albanensis) to very small (Blackia, which is the size of the smallest extant species of the genus Petinomys). Such a broad range of sizes is also observed in the extant diverse squirrel faunas of the tropical forests of SE Asia and Africa, where various species that differ in their diet and behaviour (being either nocturnal or diurnal) exploit the abundant plant resources of those environments (see MacKinnon 1978;Emmons 1980).
Spermophilinus bredai, on the other hand, is a ground squirrel the size of a chipmunk. The skull of Spermophilinus shows two small depressions posterior to the upper incisors for the attachment of the cheek pouch musculature (De Bruijn and Mein 1968). The presence of cheek pouches is a synapomorphy of the Marmotini, and these structures are not found in other ground squirrels such as the Xerini. Cheek pouches allow for the rapid collecting of food and serve as a temporary storage for food (mainly seeds and grains) until the squirrels bring it to their lairs or burrows. In the light of this evidence, a similar lifestyle to that of extant holartic ground squirrels can be inferred for S. bredai.
Eomyids, which only comprise Keramidomys sp., are supposed to have favoured humid forest environments. Such an interpretation is based on their co-occurrence with certain 'flying' squirrels and dormice (Daams and Van der Meulen 1984;Casanovas-Vilar and Agustí 2007). Most eomyids are known only from isolated cheek teeth, although several skull and mandible fragments as well as an exceptionally preserved skeleton are known (Storch et al. 1996). The latest belongs to Eomys quercyi from the Late Oligocene of Enspel (Germany) with preserved soft tissues, clearly showing the existence of a patagium similar to that of flying squirrels (Storch et al. 1996). Nothing is known of other European eomyid taxa, but it is unlikely that all the members of this diverse group were gliders (Engesser 1999).
Regarding the Gliridae, the extant representatives of this group are mostly associated to forest environments, although some of them, such as Myomimus, inhabit drier open landscapes. Van der Meulen and De Bruijn (1982) grouped living and fossil Gliridae species on the basis of characteristic features of their M1 and M2 and extrapolated the ecology of the living representatives to fossil species. Virtually all glirids have been interpreted as scansorial to arboreal frugivores/ omnivores with the exception of most of the Myomiminae. Miodyromys, which shows a dental pattern similar to that of Myomimus, could have preferred somewhat open landscapes. The genus Muscardinus is still extant, and it inhabits the temperate forests of Europe and is arboreal and mainly frugivorous. It hibernates in winter and is capable of going into torpor at other times of the year when the weather is unfavourable and food is scarce. It is unknown whether the Miocene species of this genus were capable of hibernating or entering torpor, but this ability seems to be a synapomorphy of all extant species, so it is likely that it was also present in their fossil relatives.
The cricetids include four species that differ widely in size and morphology. Several authors (Van de Weerd and Daams 1978; Casanovas-Vilar and Agustí 2007) have suggested that Eumyarion would have preferred humid and forested environments, which would explain why it is only recorded when this kind of environment is sampled. Such an interpretation is based on its molar morphology (Casanovas-Vilar and Agustí 2007) as well as on the co-occurrence of this genus with other taxa that are assumed to have favoured forested environments (Van de Weerd and Daams 1978). Regarding Democricetodon, while the medium-sized species (former Fahlbuschia) are considered to have inhabited arid environments, the small-sized ones would have preferred moister biotopes. Such interpretations are based on the distribution of the different species in the Calatayud-Teruel Basin, where the small-sized Democricetodon are more abundant in sites that include a greater number of inferred forest dwellers (Van de Weerd and Daams 1978). The same pattern is observed in the Vallès-Penedès, with D. brevis (a species that could be closely related to Democricetodon n. sp. from Gratkorn) and D. nemoralis being more common in sites that include arboreal dormice, flying squirrels and/or beavers, such as Castell de Barberà and Barranc de Can Vila 1 (Aguilar et al. 1979;Casanovas-Vilar et al. 2010). Similarly, the small-sized Megacricetodon species, such as M. minutus, are believed to have inhabited more humid and forested environments than their larger-sized congeners (Daams et al. 1988), but, given the broad geographic range of these species, they may better be considered as ubiquists. Finally, 'Cricetodon' fandli is a large-sized cricetid that shows relatively higher crowned molars and complete ectolophs in the upper molar series. The presence of complete ectolophs provides evidence for an important propalinal component in the mastication besides the usual lateral one. Therefore, this cricetid appears to have been able of coping with tougher or more abrasive plant material than the other cricetids present at Gratkorn.
The single beaver taxon, Euroxenomys minutus minutus, is interpreted based on the study of postcranial material as having been more agile when moving on the ground than other beavers (Daxner-Höck 2004b;Casanovas-Vilar et al. 2008;see Prieto et al. 2014, this issue for details).
Overall, the rodent fauna indicates the presence of forest environments, although some of its components such as "Cricetodon" fandli and Spermophilinus bredai may have inhabited more open landscapes.

Lagomorpha
The co-occurrence of three ochotonids (a primitive, rooted one plus two forms with ever-growing teeth, usually Amphilagus/Eurolagus + Prolagus + Lagopsis) has been documented in central and western Europe from the early to the late Middle Miocene (e.g. Wallenried, Sandelzhausen, Anwil; Engesser 1972; Angelone 2009; C. Angelone, personal data). This evidence probably reflects ecological/environmental conditions in which the food resources were abundant and diverse for the ochotonids.
The Gratkorn lagomorph assemblage differs from more western ones as it is composed of a rooted primitive lagomorph, Prolagus, and a large-sized pika, undetermined and probably recorded for the first time. The absence of Lagopsis is notable in Austria, but this may result from a lack of a sufficient amount of pre-Pannonian small mammal-bearing localities in the country. The genus is also missing at present from the rich Sarmatian s.str. record from Hungary (see Table 3).
Apart from the rooted ochotonid from Bełchatów A, assigned to Eurolagus fontannesi, detailed studies about the palaeodiet of European ochotonids are not available. Fostowicz-Frelik et al. (2012) analysed the dental microwear of these Polish remains and indicated that the species was a typical browser (fruit-leaf/fruit seed groups) or a seasonalregional mixed−feeder in the sense of Solounias and Semprebon (2002), preferring most probably soft grass-like plants and fruits, and fine seeds. They deduce from this feeding pattern and from the study of the Bełchatów A fossil assemblage that the species was a forest dweller.
As for Prolagus oeningensis, considering the wide geographic and stratigraphic range, a certain degree of plasticity regarding the environmental condition can be suspected. At the genus level, and based on the variations Prolagus versus Lagopsis ratios in the Iberian fossil record, López-Martínez (2001 and references therein) and Hordijk (2010) propose that Prolagus preferred relative wet environments.

Summary
The small mammals provide a mixed picture for the palaeoevironment of Gratkorn: while part of the fauna tends to indicate the presence of forested covering, more open landscapes would have been favoured by some members of the assemblage. This can be partially explained by taphonomic bias, most of the fossil remains deriving from the regurgitation of nocturnal raptors. This implies restrictions in the faunal spectrum, although the predators are considered as generalists. These restrictions are influenced by the size of the taxa, their ecology (under-representation of diurnal and forager species, for instance), and their abundance. Importantly, the hunting preferences of the predator influence the prey spectrum, and could explain in some part the mixed picture obtained for Gratkorn. Indeed, Petty (1999) observes that Strix aluco, although nesting preferentially in forest, where the pellets can thus accumulate, hunts in open landscapes. Similarly, Michelat and Giraudot (1991) show that Tyto alba in a 21 %forested surface, only consume 2 % of its active time in this habitat.
Interestingly, the reptiles and amphibians, which partly have a similar origin of accumulation, also speak for a mosaic of habitats (Böhme and Vasilyan 2014, this issue), but record, in addition to open dry landscapes, short-lived ponds, streams or rivulets in the close vicinity. The presence of such water-points is only indicated by the few beaver specimens. The savannah-like open landscapes environment under a semi-arid climate which these authors propose for the environment of Gratkorn seems in contradiction with the information provided by the small mammals, although a minority of the mammal species certainly preferred open habitats.
The environmental information provided by the molluscs occurring in abundance in the same layer corroborates the mixed vision of the palaeo-situation at Gratkorn, but emphasise woodland with moist soil (Harzhauser et al. 2008). This forested influence is underlined in the fauna studied here by the large-sized sciurid Albanensia.
A possible, but not restrictive, explanation is that the fossil accumulation was achieved in the margins of a forest area, providing nesting and roosting places for the birds (pellet accumulation), as well as a diverse mosaic of micro-habitats for their prey.

General conclusions
With regards to the previous works dealing with the small mammals from Gratkorn, the beaver taxon has been emended into Euroxenomys minutus minutus, while one large-sizedbut indeterminated-pika is newly reported. Some rare species (Spermophilinus, Miodyromys, Keramidomys, cf. Myotis) have yielded new material, but this did not allow the improvement of their taxonomy.
Based on the common occurrence of digestion traces on some of the teeth and the presence of pellets, most of the material is interpreted as being accumulated by nocturnal raptors. This could explain the low α-diversity of the fauna. But a plurality of accumulation factors, at least for part of the fauna, is certainly indicated.
Similarly, and based on the information provided by the lower vertebrates and the molluscs, which occur in abundance in the same thin fossil-enriched layer, the mixed picture of the environment (basically forested vs. open landscape) deriving from the study of the small mammal fauna might find its explanation in the plurality of accumulation factors and/or of microhabitats around the excavation site.
The extremely rapid accumulation of the fossils allows us to open an exceptional window onto the late Sarmatian s.str. life. The quality and the quantity of material promises that, in future ,Gratkorn will continue to provide further insight into the palaeontology of Miocene rodents and insectivores.