The Late Pliocene–Middle Pleistocene Large Mammal Faunal Units of Greece

: Located at the eastern corner of Mediterranean Europe, Greece occupies a critical position for mammal dispersals to/from Europe, Asia, and Africa and constitutes a potential passageway towards Western Europe. During recent decades, numerous fieldwork campaigns in several Pliocene– Pleistocene sites have greatly enriched the fossil record and provided valuable taxonomic and biostratigraphic data. However, a fully developed reference biochronological unit scheme for the Greek record that could contribute to correlations at a continental scale is still pending. In this article, we provide the updated Late Pliocene to Middle Pleistocene large mammal succession, and we introduce the Faunal Units (FUs) of Greece. We define eight FUs, the Milia, Dafnero, Gerakarou, Tsiotra Vryssi, Krimni, Apollonia, Marathousa, and Apidima FUs (from the oldest to the youngest), which are determined by a set of first and last local occurrences. The results form the basis for discussion of already set turnovers, dispersals, and extinction/immigration events and showcase the importance of the local record for the investigation of the European terrestrial ecosystems. By developing the first detailed biochronological scheme for the Pliocene–Pleistocene of Southeastern Europe, this study comprises the basis for an expanded Balkan faunal unit scale and a reference framework for future investigations.


Introduction
As one of the most likely passageways from Africa/Levant and as the main gate from Asia to Europe, the Balkans have long been considered crucial in understanding and developing the Neogene/Quaternary European continental biostratigraphy and biochronology (e.g., [1][2][3][4][5][6][7]).Greece occupies a critical position at the southernmost tip of the Balkans and at the eastern corner of the Mediterranean Europe, and the region's fossil record is of fundamental importance and interest in this discussion.
Research of fossil vertebrate in Greece has a long tradition for almost two centuries, with numerous sites and finds (e.g., [8,9]) that systematically add to the knowledge of the evolution of the Quaternary continental and insular faunas of Europe.During the last two decades, numerous fieldwork campaigns at multiple sites (e.g., Dafnero, Sesklo, Vatera, Volakas, Libakos, Mygdonia Basin, and Megalopolis Basin) by various paleontological teams have greatly enriched the Late Pliocene to Middle Pleistocene fossil record in Greece and provided valuable taxonomic, biostratigraphic, paleoenvironmental, and taphonomic data.Although basic, local biochronological-biostratigraphic schemes have already been proposed and revised on many occasions (e.g., [6,[10][11][12][13]), they remain atypical, and no systematic effort has been made so far to manage the available data towards the creation of a formal and broader local/regional biochronological unit scheme that could contribute to correlations at a continental scale.
Based on previous studies and new data, we provide the updated Late Pliocene to Middle Pleistocene mammal succession in Greece by means of faunal units and discuss key taxa in both the regional and pan-European frame.Using available local chronological markers and biochronological evidence, we time-calibrate the local faunal scale and correlate it with well-dated faunas of Western Europe.The results form the basis for a discussion of already set mammal turnovers, dispersals, and extinction/immigration events and showcase the importance of the local record for the investigation of the European terrestrial ecosystems.We expect this effort to provide a basis for an expanded Balkan Pliocene-Pleistocene faunal unit scale and a reference framework for future investigations.
Our dataset includes Greek mammal assemblages spanning from ~3.5 to ~0.12 Ma (Late Pliocene to Middle/Late Pleistocene) that include more than three large mammal taxa identified at least at the genus level.Faunal lists of local faunal assemblages (LFAs) derive from a wide range of literature references (e.g., [6,8,9,13,26,27] and references cited).Published and unpublished data recovered from recently described sites and faunas (such as GAS, SES-L, DFN3, KZ, AK, TSR, PLN, KMN, KYP-3, 4, and MAR-1; Table 1), updated lists of already known LFAs (such as DFN, APL, APD, and LIB), and new chronological calibrations (such as for DFN/DFN3, TSR, and MAR-1) are all involved in the analysis in order to achieve the maximum possible information and time control.In some cases, published taxonomic identifications have been updated and/or reviewed according to the current knowledge to reach a basic taxonomic consensus.
A hierarchical cluster analysis based on presence/absence data at the genus level was applied to a subset of the 23 most complete Greek LFAs (number of recorded genera ≥5) and 43 genera of large mammals (Supplementary Material) in order to test their best possible grouping(s) by means of faunal synthesis consistency and correlate clusters with particular time slices.The analysis was performed using PAST [28].The Q-mode dendrogram [following UPGMA: unweighted pair group method with arithmetic mean algorithm] was obtained by applying several similarity indices, which provided consistent final groupings; the results of the Raup-Crick similarity index, which is used to statistically test presence/absence data through a randomization method [29], were selected for presentation.We excluded, however, a posteriori Aghia Kyriaki (AK; N genera = 6) as it stands as an outlier in most of the runs.
Distinct Greek faunal units (GFUs) have been defined within each cluster based on a species-level analysis on the whole set of data.Local first and last historical appearances (hereinafter referred to as first and last local occurrences, FLOs and LLOs, respectively) of large mammal taxa have been used for the definition of each GFU.Fossiliferous sites included in this study along with their location and selected references are provided in Table 1.

Results
The cluster analysis provided three main groups of LFAs (similarity index ≥ 0.5) that roughly correspond to recognized ELMAs.Galerian and Aurelian LFAs (KYP-3, 4, MAR-1, APD-A, B; cluster "A" in Figure 1) are clearly separated from those of the Villafranchian-Epivillafranchian (cluster "B" in Figure 1) based on the Homo + Palaeoloxodon + Cervus s. str.+ Dama + Hippopotamus large mammal association.Subcluster "a1" can be dated between 780 and ca.450 ka as it includes the sites KYP-4 (lower part of the Middle Pleistocene <780 ka [70,71], MAR-1 (ca.450 ka, correlated to Marine Isotope Stage 12 [72]), and KYP-3 (intermediate stratigraphic position between KYP-4 and MAR-1 [70]).Subcluster "a2" is represented in the analysis only by APD-A and B LFAs and corresponds broadly to the late Middle Pleistocene; the crania of Homo from Apidima Cave A provided ages of ca.210 and 170 ka [5], and a comparable or slightly younger age can be assumed for Apidima Cave B.
Cluster "B" incorporates Villafranchian and Epivillafranchian LFAs (Figure 1).Within subcluster "b1", the LFA grouping largely follows the informal scheme of the early ("b1.a"),middle ("b1.b"), and early late ("b1.c")Villafranchian subdivision, whereas late late Villafranchian ("b2.a") and latest Villafranchian-Epivillafranchian ("b2.b")assemblages are incorporated in subcluster "b2".The "b1.a" subcluster includes the LFAs of MIL, SES-L, and APO, characterized by the co-occurrence of Anancus + Hipparion in contrast to any other group of LFAs included in the analysis.Vlachos et al. [57] correlated the best-documented Milia (MIL) faunal assemblage with MN16a, suggesting an approximate age of 3.0-2.8Ma [73].We find the link of this cluster with "b1.c" artificial and therefore of very low credibility; it is mainly based on the common presence of Sus and Gazella, though species-level taxonomy is sharply different in these two sets of faunas.
Cluster "B" incorporates Villafranchian and Epivillafranchian LFAs (Figure 1).Within subcluster "b1", the LFA grouping largely follows the informal scheme of the early ("b1.a"),middle ("b1.b"), and early late ("b1.c")Villafranchian subdivision, whereas late late Villafranchian ("b2.a") and latest Villafranchian-Epivillafranchian ("b2.b")assemblages are incorporated in subcluster "b2".The "b1.a" subcluster includes the LFAs of MIL, SES-L, and APO, characterized by the co-occurrence of Anancus + Hipparion in contrast to any other group of LFAs included in the analysis.Vlachos et al. [57] correlated the best-documented Milia (MIL) faunal assemblage with MN16a, suggesting an approximate age of 3.0-2.8Ma [73].We find the link of this cluster with "b1.c" artificial and therefore of very low credibility; it is mainly based on the common presence of Sus and Gazella, though species-level taxonomy is sharply different in these two sets of faunas.Subcluster "b1.b" includes the LFAs of DFN/DFN3, SES-U, VAT, VOL, KZ, and PYR (Figure 1).With the exception of Pyrgos (PYR), this faunal assemblage is characterized by the association of Nyctereutes + Equus + Palaeotragus + Gazella + Gazellospira + Gallogoral, while Paradolichopithecus is also present in three out of the remaining five LFAs.Based on magnetochronology of the DFN/DFN3 sites [39], this group of LFAs is dated around 2.4-2.3Ma.The PYR LFA, rather, has to be excluded from this subcluster and is likely associated with "b1.c", as suggested by the lack of basic taxa of this association (e.g., Gallogoral, Nyctereutes) and evidenced by the species-level taxonomy of its equid content [74].
The subcluster "b1.c" includes the LFAs of GER and VSL (Figure 1) and is characterized by the association of Equus + Sus + Gazella + Leptobos.Although Canis is recorded only in GER LFA, we suggest it is part of this association, and its absence from VSL is due to the limited material from this LFA.Both GER and VSL belong to the same stratigraphic levels of Mygdonia Basin, below the TSR LFA which provided a maximum age of 1.78 Ma [66].Hence, we estimate an age of about 2.0 Ma for this faunal group, confirmed by the common presence of Pliocrocuta and Pachycrocuta in GER.
Within "b2", the subcluster "b2.a" includes the LFAs of TSR, LIB, KLT, KRI, KAL, and KMN (Figure 1).The faunal assemblage is characterized by the co-occurrence of Mammuthus + Equus + Stephanorhinus + Hippopotamus + Dama-like deer and, although not Subcluster "b1.b" includes the LFAs of DFN/DFN3, SES-U, VAT, VOL, KZ, and PYR (Figure 1).With the exception of Pyrgos (PYR), this faunal assemblage is characterized by the association of Nyctereutes + Equus + Palaeotragus + Gazella + Gazellospira + Gallogoral, while Paradolichopithecus is also present in three out of the remaining five LFAs.Based on magnetochronology of the DFN/DFN3 sites [39], this group of LFAs is dated around 2.4-2.3Ma.The PYR LFA, rather, has to be excluded from this subcluster and is likely associated with "b1.c", as suggested by the lack of basic taxa of this association (e.g., Gallogoral, Nyctereutes) and evidenced by the species-level taxonomy of its equid content [74].
The subcluster "b1.c" includes the LFAs of GER and VSL (Figure 1) and is characterized by the association of Equus + Sus + Gazella + Leptobos.Although Canis is recorded only in GER LFA, we suggest it is part of this association, and its absence from VSL is due to the limited material from this LFA.Both GER and VSL belong to the same stratigraphic levels of Mygdonia Basin, below the TSR LFA which provided a maximum age of 1.78 Ma [66].Hence, we estimate an age of about 2.0 Ma for this faunal group, confirmed by the common presence of Pliocrocuta and Pachycrocuta in GER.
The subcluster b2.b includes the LFA of APL and most likely ALK-VOL (Figure 1).At the genus level, it is defined by the association of Equus + Canis + Hemitragus and likely Pachycrocuta, Mammuthus, and Pontoceros.The large mammal association at the genus level appears quite similar to that of the previous assemblage, but key taxa such as Palaeotragus and Leptobos are missing.Although no geochronological evidence is available for this assemblage, it is safely placed between 1.5 and 0.8 Ma and likely around 1.2 Ma [6].

Establishment of the Greek Faunal Units
Based on the cluster analysis results and the local stratigraphic, biostratigraphic, biochronologic, and geochronologic evidence, the entire set of Greek Late Pliocene to Middle Pleistocene LFAs were reclassified chronologically based on the degree of faunal consistency (Figure 2, fourth column).A species-level analysis within and between successive LFA clusters allow eight GFUs to be recognized, which are (from the base to the top): the Milia FU, Dafnero FU, Gerakarou FU, Tsiotra Vryssi FU, Krimni FU, Apollonia FU, Marathousa FU, and Apidima FU (Figure 2, fifth column).Each GFU is defined by a set of FLOs and LLOs (Figure 3).The succeeding Gerakarou FU at around 2.0 Ma includes GER, VSL, AK, and possibly PYR and MKN LFAs (e.g., [6,26,31,47] and references cited).The FLOs of Canis s. str.along with Panthera gombaszoegensis, Pachycrocuta brevirostris, and Equus altidens are found here, whereas Pliocrocuta perrieri, Equus stenonis s. str., Croizetoceros ramosus, Gazella bouvrainae, and most likely Gazellospira appear for the last time in the Greek record (Figure 2).
The Tsiotra Vryssi FU follows at about 1.8-1.6Ma, represented only by the homonymous LFA from Mygdonia Basin (e.g., [63,66]).The faunal assemblage incorporates most taxa that first occurred in the previous GFU but also includes the FLOs of Stephanorhinus hundsheimensis, Equus ex.gr.apolloniensis, Praemegaceros, Eobison, and Pontoceros (and perhaps a smaller-sized Megantereon), indicating the beginning of an important renewal of the mammal community.At the same time, canids related to the Canis mosbachensis lineage, alongside Canis (Xenocyon) [79], also appear in this GFU.
The Apollonia FU includes APL, RVL, PNT, KAL, MAR, and RIZ LFAs from Mygdonia Basin (e.g., [6,33,43,44,63]) and most likely ALK-VOL from Thessaly [26,30,59].No new FLOs occur in this assemblage, yet the absence of middle ("archaic") Villafranchian relics (i.e., Palaeotragus, Leptobos) and the presence of mostly late Villafranchian newcomers indicate the more "modern" characteristics of the fauna.On the other hand, this GFU documents the LLOs of several large mammal taxa, such as Mammuthus meridionalis, Pach- The lack of solid, local evidence concerning early to early/middle Villafranchian LFAs from Greece and the rather sparse record allow only the Milia FU to be recognized within this interval at about 3.0 Ma, including MIL and possibly GAS LFAs (e.g., [42,56,75]).The Milia FU is marked by the LLOs of "Mammut", Alephis, Parabos, and Procapreolus and the FLOs of Ursus, Homotherium latidens, Stephanorhinus jeanvireti, Gazellospira torticornis, Croizetoceros ramosus, and cervids of the "Pseudodama" lineage (Figure 2).Mammuthus might also firstly occur in this FU [76], though data are rather sporadic and not well time-controlled.The older Apolakkia (APO) LFA from Rhodes Island bears very few large mammals and mostly micromammals, the latter indicating a latest Ruscinian (MN15) age [26,37,38].It is therefore quite possible that at least one more FU can be established below the Milia FU, but data are currently insufficient.Due to the absence of Equus, the Sesklo lower level (SES-L) LFA is certainly dated before the Pliocene/Pleistocene boundary [58].It shares with Milia several common taxa at the genus level (Figure 2), but the large and advanced Plesiohipparion and the possible presence of Gazella bouvrainae [58] indicate a younger (MN16b) age, and the assemblage might represent a distinct FU between those of Milia and Dafnero, occupying the uppermost part of the early Villafranchian.The overlain Marathousa FU, dated between <0.80 and ca.0.45 Ma, includes LFAs from the Megalopolis Basin in Peloponnese (e.g., KYP-3, 4, MAR-1, 2; [48][49][50]54,55,80]) and most probably the older (attributed to the early Middle Pleistocene) faunal assemblage from Petralona cave in Chalkidiki Peninsula (e.g., [60,61]).This GFU encompasses the Galerian faunas and is marked by an important reorganization of the fauna including the FLOs of Homo, Palaeoloxodon antiquus, Lutra simplicidens, Stephanorhinus hemitoechus, Bison, Sus scrofa, the Cervus elaphus lineage, and possibly Bos, while, considering the Middle Pleistocene fauna of Petralona this FU may also include the FLOs of Ursus deningeri, Panthera spelaea, Crocuta crocuta, and "Hyena" prisca (the latter as Pliocrocuta perrieri in [62]).In addition, the presence of a Panthera sp., possibly Panthera pardus, at KYP-3 [48] could also potentially represent the FLO of this species, as could also be the case for Mammuthus trogontherii from Loussika and other localities (e.g., [81]).On the other hand, the Marathousa FU documents the LLOs of Macaca sylvanus, Equus altidens, Hippopotamus antiquus, and Praemegaceros as well as of Lutra simplicidens.
The uppermost, Apidima FU, represents the establishment of modern faunal elements at the end of the Middle Pleistocene but can practically be extended to the Late Pleistocene as the large mammal fauna does not evidence any renewal during the latter epoch, besides several disappearances towards its end.In addition to the LFA of the Apidima cave complex (APD-A, B) several other LFAs can be included in this FU, for instance, the caves of Petralona (younger fauna), Vraona, Diros, Agios Georgios, Almopia, Kalamakia, Lakonis, Klissoura, Franchthi, Mavri Spilia, and Melitzia as well as Penios valley (e.g., [82][83][84][85][86][87][88][89][90][91][92][93]).The possibly oldest LFAs of this FU, APD-A and B [5]  Apart from the well-known LFAs of DFN/DFN3, SES-U, VAT, VOL, and KZ (e.g., [6,40,51,58,67,69]), the Dafnero FU may also include TB 3-5 and possibly KSR, both known for limited large mammal remains [11,47,65].The Dafnero FU is characterized by the presence of Paradolichopithecus arvernensis and Nyctereutes megamastoides; the LLOs of Anancus, Nycteureutes tingi, Chasmaporthetes, Stephanorhinus jeanvireti, and Gazella borbonica; and the FLOs of Lynx issiodorensis, Megantereon cultridens lineage, Ursus etruscus, Mammuthus meridionalis, Stephanorhinus etruscus, Sus strozzii, Palaeotragus inexpectatus, Eucladoceros s. l., and Leptobos etrusus.The Damatria (DAM) LFA from the upper part of the homonymous Damatria Formation in Rhodes Island is more difficult to assess within this scheme, due to the puzzling evidence arising from the combination of small and large mammals, which are broadly assigned to MN16 [11,26,77].Local biostratigraphic evidence places the Plio-Pleistocene boundary in the lower part of the overlying Kritika Formation ( [78] and references therein).Hence, the presence of Equus and Sus strozzii (if this latter is confirmed) suggest that the DAM LFA slightly predates or roughly corresponds to the early/middle Villafranchian transition and could potentially be included in the Dafnero FU, which is suggested to last from ca. 2.3 to ca. 2.5 Ma.
The Tsiotra Vryssi FU follows at about 1.8-1.6Ma, represented only by the homonymous LFA from Mygdonia Basin (e.g., [63,66]).The faunal assemblage incorporates most taxa that first occurred in the previous GFU but also includes the FLOs of Stephanorhinus hundsheimensis, Equus ex.gr.apolloniensis, Praemegaceros, Eobison, and Pontoceros (and perhaps a smaller-sized Megantereon), indicating the beginning of an important renewal of the mammal community.At the same time, canids related to the Canis mosbachensis lineage, alongside Canis (Xenocyon) [79], also appear in this GFU.
The Apollonia FU includes APL, RVL, PNT, KAL, MAR, and RIZ LFAs from Mygdonia Basin (e.g., [6,33,43,44,63]) and most likely ALK-VOL from Thessaly [26,30,59].No new FLOs occur in this assemblage, yet the absence of middle ("archaic") Villafranchian relics (i.e., Palaeotragus, Leptobos) and the presence of mostly late Villafranchian newcomers indicate the more "modern" characteristics of the fauna.On the other hand, this GFU documents the LLOs of several large mammal taxa, such as Mammuthus meridionalis, Pachycrocuta, Lynx issiodorensis, Homotherium, Megantereon, Panthera gombaszoegensis, Ursus etruscus, Stephanorhinus etruscus, Equus ex gr.apolloniensis, Eobison, Soergelia, and Pontoceros.The age of the Apollonia FU is currently estimated at about 1.2 Ma, but it is also possible that the herein included LFAs may cover a wider time span from ca. 1.5 up to 0.9 Ma; alternatively, the scarcity/absence of Greek LFAs from this time interval may obscure the potential presence of an additional "ghost" GFU.

The Greek Faunal Units in the European context
Based on available data from Italy, France, and Spain [14,21,73,94,95], the Milia GFU roughly corresponds to the Triversa FU of Italy and Viallete LFA level in France (Figure 4).The Dafnero GFU is correlated to the St. Vallier FU of Italy and France and the Huelago/Fuenta Nueva-1 LFA level in Spain (Figure 4).The Gerakarou GFU is placed between the Costa St. Giacomo and Olivola FU of Italy, Senèze LFA level in France, and La Puebla de Valverde LFA level in Spain (Figure 4).The Tsiotra Vryssi GFU corresponds to the Tasso FU or to the late Tasso/early Farneta FUs of Italy and the Villanueva di Pitamo LFA level in Spain (Figure 4).The Krimni GFU corresponds to the Farneta FU in Italy and the Venta Micena/Fuente Nueva-2 LFA level in Spain (Figure 4).The Apollonia GFU is placed between the Pirro Nord and Colle Curti FUs of Italy and correlates with the Fuente Nueva-3/Barranco León 5/Sima del Elefante LFA level in Spain (Figure 4).The Marathousa GFU represents a unit within the Isernia and Fontana Ranuccio FUs in Italy and also corresponds to the Gran Dolina TD8, Sima de los Huesos, L'Escale and Arago CM LFAs in Spain and France, respectively (Figure 4).Finally, the Apidima FU can be correlated to several European localities collectively dated to the end of the Middle-Late Pleistocene.[14,16,95,96]).
Several authors discuss the problem of diachrony/asynchrony in transcontinental correlations based on mammals and the establishment of mammal dispersal bioevents (e.g., [23,97,98]).The progressive and often longitudinally or latitudinally directed (N↔S, E↔W) environmental changes, the geographical proximity/distance to neighboring regions that are potential reservoirs of new taxa (such as Africa and Asia for Europe), and the changing geographical barriers/routes to and from target regions (such as the three peninsulas of Southern Europe) accentuate local diversification and create basic conditions for diachronicity.On the other hand, discontinuities and gaps in the local faunal record (sampling effect), spatiotemporal imbalances in the distribution of LFAs at local, regional and interregional scales, asymmetries in the density of available geochronologic age estimates, and long-lasting taxonomic inconsistencies (e.g., discussion on species-level systematics of European Canis, Eucladoceros, Megantereon, "Pseudodama", and Equus species-level systematics) inhibit or even prohibit the recognition of mammal dispersal events and the ascertainment/control of their possible diachronicity.At present it seems rather unreasonable to try to interpret the S. European Pliocene to Pleistocene data at a finer scale than the temporal resolution corresponding to a full FU (100-200 kyr).
Even under these unfavorable conditions, the Greek record provides supporting evidence for already well-known and long-established mammal dispersal events such as those of Equus at 2.6-2.5 Ma (as redefined by [98]) and Pachycrocuta at ~2.0 Ma [99].The former is recorded between SES-L and DAM LFAs (or between Milia and Dafnero GFUs).From the Balkan to the Iberian Peninsulas at least, the Equus event coincides with the decline of hipparionine horses [58,100], the replacement of Sus arvernensis by Sus strozii (e.g., [101]), and the appearance of Mammuthus meridionalis (e.g., [102]).
The arrival and rapid expansion of Pachycrocuta in Europe is associated with the last occurrences of Pliocrocuta, both events recorded in the Gerakarou GFU, which also documents the occurrence of Panthera gombaszoegensis [103] that entered Europe roughly at the same time as the giant hyena (e.g., [104]).Although the simultaneous and widely discussed Canis event ("wolf event" of [105]) is largely abolished due to the diachronous early evidence of wolf-like taxa across Eurasia (e.g., [14,[106][107][108][109]), the south European record rather agrees with the firm establishment of cursorial pack-hunting canids (Canis ex gr.etruscus and Canis arnensis) between 2.1 and 2.0 Ma (e.g., [110,111]).It is immediately followed (2.0-1.8Ma) by the last European occurrences of Gazella and Gazellospira among herbivores.At the same time Equus altidens makes its first appearance in Eastern Europe/the Balkans (e.g., [112,113]).
The next complex bioevent of similar duration starts at 1.7-1.6Ma.All across Southern Europe, Leptobos was replaced by Eobison, cervids of the Praemegaceros-lineage expanded, Stephanorhinus hundsheimensis first occurred in the East [14,46,114], and Pontoceros expanded from the north peri-Pontic to the south Balkan territories.At the same time or slightly after (1.6-1.5 Ma), a smaller Megantereon replaced the previous larger, one and Hippopotamus was firmly established in the northern Mediterranean area (although the genus had sporadically occurred earlier; see [115] and references therein).
A major large mammal turnover is evident in the beginning of the Middle Pleistocene as part of the faunal reorganization close the Early/Middle Pleistocene transition that is associated with the period generally called the "Mid-Pleistocene Revolution" (e.g., [96] and references therein).This event encompasses at the (South) European scale the disappearance of several Villafranchian-Epivillafranchian taxa (e.g., Mammuthus meridionalis, Pachycrocuta, Ursus etruscus, Megantereon, Panthera gombaszoegensis, and Soergelia) and the arrival of several Galerian newcomers (e.g., Palaeoloxodon, Mammuthus trogontherii, Crocuta, Panthera spelaea, Panthera pardus, Sus scrofa, and Stephanorhinus hemitoechus).At the Greek scale, this turnover is noticeable in the transition from the Apollonia FU to the Marathousa FU.

Conclusions
Low-rank biochronological scales such as FUs have a restricted significance, representing a summary/synthesis of local to regional (bio-)chronological ordering and evolution of the mammalian network under the pressure of climatic/environmental oscillations.In light of the common and multifactorial diachrony/asynchrony phenomena across wide physical and ecological boundaries, the role of FUs appears, however, to be crucial for understanding the dynamics of mammalian dispersal in larger geographical scales and for monitoring bioevents of broader significance (e.g., [14,16,98,116,117]).As local biochronological resolution is based on continuous or adjacent and therefore relatively well-correlated sedimentary sequences, FU scales may provide complementary data to interregional/continental scale correlations, especially in the discontinuous and highly fragmentary terrestrial environments.As such, they facilitate the detection, definition, and control of important bioevents, which in turn lead to the establishment of higher-rank biochronological units/scales.
Previous and ongoing investigations in the Greek Pliocene and Pleistocene allow us to generate a first attempt for a local faunal unit scale (GFU) representing the evolution of large mammal communities from Mid-Pliocene to the Late Pleistocene.Each one of the eight recognized Greek faunal units is defined by a set of first and last local occurrence data, though the latest Pliocene and latest Early Pleistocene are still poorly resolved.Additionally, both the local record as well as those from the south of Western Europe (Italy, France, Spain) suggest that the temporal resolution cannot be improved by current knowledge below a critical interval of 100-200 kyrs, which equals the duration of a full FU.Comparison of the GFU scheme with Western European ones indicates a good correspondence in several dispersal events of particular large mammal taxa.Nevertheless, both the early to middle Villafranchian transition at ~2.6 Ma and the Epivillafranchian faunal reorganization at ca. 1 Ma, thoroughly discussed in the literature (e.g., [14,95,118]), are not expressed as clearly in the large mammal record of Greece as in other Southern European countries due to inadequate data.
Author Contributions: Conceptualization, methodology, investigation, resources, D.S.K. and G.E.K.; validation, formal analysis, data curation, visualization, G.E.K.; writing-original draft preparation, D.S.K.; writing-review and editing, G.E.K.All authors have read and agreed to the published version of the manuscript.Funding: G.K. is supported by the Deutsche Forschungsgemeinschaft (DFG Project no.463225251, "MEGALOPOLIS").

Figure 1 .
Figure 1.Hierarchical cluster analysis at the genus level showing the faunal similarities among selected Villafranchian to Aurelian large mammal assemblages of Greece.

Figure 1 .
Figure 1.Hierarchical cluster analysis at the genus level showing the faunal similarities among selected Villafranchian to Aurelian large mammal assemblages of Greece.

Figure 2 .
Figure 2. The Late Pliocene-Middle Pleistocene (Villafranchian to Aurelian) large mammal faunal units of Greece, the examined fossiliferous localities, and biochronological range of selected large mammals (silhouette images from PhyloPic, phylopic.org).

Figure 2 .
Figure 2. The Late Pliocene-Middle Pleistocene (Villafranchian to Aurelian) large mammal faunal units of Greece, the examined fossiliferous localities, and biochronological range of selected large mammals (silhouette images from PhyloPic, phylopic.org).

Figure 3 .
Figure 3. Biostratigraphic table of the large mammal localities and faunal units of Greece with the first and last local occurrences of the various taxa.
, mark the beginning of the faunal modernization documenting Capra ibex, Dama dama, Lynx lynx, Vulpes vulpes, and Felis sylvestris ([32]; the red fox and the wildcat could have already been present in the Marathousa FU) while other modern species such as Ursus arctos and Canis lupus are present in the other LFAs.The Apidima FU includes the LLOs of Palaeoloxodon antiquus, Panthera pardus, Panthera spelaea, Crocuta crocuta, Ursus spelaeus, Stephanorhinus hemitoechus, Bison priscus, Megaloceros, and Hippopotamus amphibius.

Figure 3 .
Figure 3. Biostratigraphic table of the large mammal localities and faunal units of Greece with the first and last local occurrences of the various taxa.

Figure 4 .
Figure 4. Correlation of the large mammal localities and faunal units of Greece with those of Italy, Spain, and France (data from[14,16,95,96]).

Table 1 .
Fossiliferous sites included in the study including their abbreviations, locations, and selected references.