The Danube Corridor Hypothesis and the Carpathian Basin: Geological, Environmental and Archaeological Approaches to Characterizing Aurignacian Dynamics

Current archaeological evidence suggests that modern humans entered Europe from East Africa either along the Mediterranean coast (Bar-Yosef 1998; Mellars 2006), through the East European Plain (Anikovich et al. 2007; Mellars 2004), and/or through a general axial trajectory along the Danube catchment from Southeastern into Central Europe (Conard and Bolus 2003, 2008). The last theory is supported by numerous well-studied and well-dated early Upper Paleolithic sites in the western Danube region, for example, Hohle Fels, Willendorf II, Geißenklösterle, and Keilberg-Kirche (Conard and Bolus 2008; Higham et al. 2012; Nigst 2006; Nigst et al. 2014; Teyssandier 2004; Uthmeier 1996). Slightly older early Upper Paleolithic assemblages (Bachokirian and Kozarnikian) are also found in the lower reaches of the Danube catchment, for example, at Kozarnika, Temnata and Bacho Kiro, pointing to a hypothesized hominin starting point in Southeastern Europe (Kozłowski 2004; Kozłowski and Otte 2000). Hominin discoveries and dating projects have additionally identified an early presence of anatomically modern humans in the Middle Danube catchment, exemplified by the skeletal remains at the Peștera cu Oase (Trinkaus et al. 2003, 2012).

Still, the role of the Danube catchment in early modern human migrations is poorly understood, as the catchment’s early Upper Paleolithic sites have not been verified and tested alongside the more extensive surrounding archaeological record. Current archaeological research along the Danube catchment remains limited to specific regions such as the surrounding highlands, the Inner Carpathians, Lower Austria, southern Moravia, and the Dobruja province of the lower catchment; little is known from the Basin itself (Anghelinu et al. 2012a; Ioviță et al. 2014; Steguweit et al. 2009). Additionally, although a greater emphasis on collection re-examination, site formation processes and re-dating efforts has helped to clarify erroneous sites, many findspots remain poorly understood, while others with single and multiple layers are only just being identified/re-excavated—for instance, Beregovo I and Crvenka–At (Chu et al. 2014; Usik 2008). There is also surprisingly little discussion among archaeologists about Middle Danube topographic and paleoclimatic variability that could have influenced modern human migration.

To evaluate this, data from the Middle Danube catchment, here defined as the Carpathian Basin, are compiled to explore the possible expansion of the early Upper Paleolithic from Southeastern Europe into Central Europe. Known sites are positioned within a broad synchronic perspective of modern human subsistence in the Carpathian Basin across a varied spatial, climatic and environmental context. This paper draws on archaeology, paleontology, geomorphology and climatology to explore the archaeological signal of the early Upper Paleolithic in the Carpathian Basin and to answer the questions of how the early Upper Paleolithic record of the Carpathian Basin fits into the Danube corridor hypothesis, and what avenues of research are the most promising.

The Carpathian Basin is a back-arc basin around 200 m above mean sea level (mamsl), located at the intersection of the Balkan Peninsula and Central and Eastern Europe (Fig. 1). It measures roughly 400 km from north to south and 800 km from east to west and is longitudinally bisected by the Danube and Tisza rivers. Geographically, the Carpathian Basin is clearly demarcated—bounded to the north and east by the Carpathians, to the west by the Alps/Dinarides, and to the south by the Balkans. Politically, the Carpathian Basin is highly fragmented, though it is primarily located within the borders of Hungary, eastern Croatia, western Romania, southern Slovakia, northern Serbia, southwestern Ukraine and northern Bosnia and Herzegovina.

The Carpathian Basin began forming during the Late Cretaceous and Cenozoic, as a result of a collision between the Eurasian and Nubian plates (Horváth et al. 2015). Outward thrusting from the present Carpathian Basin was directed toward the European platform and the Adriatic Sea. The interior regions of the thrust belts were formed during the Mesozoic, while the exterior regions were formed later in the Early Cenozoic (Royden et al. 1982). Gradual uplift of the Carpathian Mountains resulted in subsidence associated with the Late Miocene retreat of the Paratethys Sea to form the Pannonian Lake within the rough current boundary of the Carpathian Basin (Horváth 1993).

By the beginning of the Late Miocene, as the Pannonian Lake retracted, it gave rise to the Danube and Tisza river systems and an associated interconnected paleolake system (Gábris and Nádor 2007). From the Late Pliocene until the beginning of the last glacial, the Danube, coming from the Visegrad Gorge, the Tisza River from the Eastern part of the alluvial plain, and the paleo-Bodrog River all converged on the south Tisza Graben (Kiss et al. 2015; Starkel et al. 2015).

Tectonic activity and global climatic cooling during the last glacial generally resulted in a large braided river system across the basin (Mezősi 2017), but complex regional tectonics, riparian vegetation and climate shaped local hydrology differently. Rivers in uplifted/tectonically stable areas (broadly the western Carpathian Basin) had narrow (a few kilometers wide), incised (hundreds of meters) terraced valleys, whereas rivers in areas of net subsidence (broadly the eastern Carpathian Basin) have buried terraces with heavy sedimentation and floodplains that transformed into large, unstable alluvial plains (Perșoiu et al. 2017).

The Danube had a multi-channel, anastomosing pattern moving southeast across the current Danube–Tisza Interfluve, where it built a broad alluvial fan (Gábris 1994; Mezősi 2017, p. 43). However, broadly 50 ka ago, tectonic activity forced the river west into its current direction, where it began to incise, leaving behind a series of (misfit) channels and dry surfaces across the abandoned eastern course (Kiss et al. 2015; Starkel et al. 2015). Thus, the former alluvial fans and floodplain levels became dry surfaces, while the low-lying areas became subject to frequent inundation.

By contrast, rivers in the subsiding Eastern Carpathian Basin were primarily meandering during the last glacial (Cserkész-Nagy 2012; Gábris et al. 2012). Optically-stimulated luminescence (OSL) dates and sediment mineralogy suggest that the Tisza ran 80 km east of its current trajectory c. 46–39 ka ago (Nádor et al. 2007; Timár et al. 2005) and was prone to frequent tectonically influenced local and regional channel avulsions (Gábris and Nádor 2007; Perșoiu et al. 2017). However, in the mid-altitude feeder rivers, there is evidence that between 45 and 37 ka ago, medium-sized rivers had shallow, gravel-bedded, braided channels (Perșoiu et al. 2017; Starkel et al. 2015).

The current course and morphology of the Danube and its main tributaries give a false impression of the size and magnitude of the Pleistocene river systems, as their regimes and planforms have changed markedly since the Holocene, due to increased seasonal flow and a switch to small-scale meanders (Perșoiu and Rădoane 2017). The Danube channels were also heavily controlled in the nineteenth century through extensive straightening and the construction of hundreds of levees and thousands of km of drainage canals. The most extensive of these works were along the Tisza, which was shortened by one-third (c. 500 km), while its 10–100 km wide floodplain was reduced to 1–5 km (Csekö and Hayde 2004).

Nonetheless, the complex hydrological changes throughout the Quaternary in the Carpathian Basin have left an abundant sedimentary archive throughout the basin. Obtaining a cohesive understanding of the distribution of these sediments has been problematic due to differences in cross-border source data (Lindner et al. 2017). Still, a generalized picture for the area can be painted for Hungary and this can be extrapolated to the wider region. In the interior mountain areas (e.g. the North Hungarian Range), 1–25 m thick Quaternary rock fields, debris and lag gravels appear infrequently; however, in the hilly regions, 0–140 m thick sediments are comprised of nearly ubiquitous loess-paleosol complexes whose loose sediments have been prone to mass wasting. In the basin, the surface of the higher alluvial fans and low-lying floodplains are made up of fine and coarse Quaternary fluvial sediments, whose thickness varies between 20 and 750 m (Jámbor 2012).

This large amount of sedimentation in the Carpathian Basin, along with a lack of appropriate geoarchaeological methods, has partly impeded the discovery of open-air sites. While geochemical and microscopic methods have gained traction in recent years, geoarchaeological research in the Carpathian Basin at a landscape scale has mostly focused on tracing raw materials sources. However, many parts of the Carpathian Basin have not been systematically surveyed, and advanced geophysical techniques focused on locating new sites are still relatively unused (e.g. remote sensing, magnetometry, electrical resistivity and ground-penetrating radar). Additionally, the application of other field-based and laboratory methods to elucidate site formation processes is also uncommon, which has been problematic for the comprehensive understanding of site integrities and chronometric datings.

Paleolithic research in the Carpathian Basin has a long history that in its early phase (c. 1900) was part of a larger pan-European interest in the antiquity of humans (Romanowska 2016). Early scholars focused their efforts primarily on cave sites (e.g. Doboș et al. 2010, p. 21; Lengyel et al. 2008), though, as elsewhere, their excavations frequently resulted in poorly-documented, biased archaeological collections lacking in contextual information.

As archaeological research grew, the increasing geopolitical fractionation of the Carpathian Basin following World War I inhibited cross-border academic exchange, with a greater propensity to publish in local languages and national publications leading to divergent national archaeological methods and theory. Isolation meant that some countries in the Carpathian Basin never fully implemented advancing theoretical frameworks, continuing to use type fossils and adapting them to create a menagerie of local lithic industries that did not traverse national borders, complicating intraregional comparisons. Additionally, Paleolithic research was increasingly used (to different extents) to advance Marxist perspectives and to uphold ethnogeneses according to a culture-historical paradigm (Anghelinu 1998; Bartosiewicz 2016; Dragoman and Oanţă-Marghitu 2006).

Nevertheless, numerous early Upper Paleolithic sites in the Carpathian Basin are known to contain important geochronological, paleontological and archaeological evidence. The following section presents short descriptions of these sites and positions them within a regional context. These data are drawn from sites located across the Carpathian Basin within the Danube catchment (Hungary, Croatia, Slovakia, Serbia, Romania, Ukraine, and Bosnia and Herzegovina) and broadly situated across the same environmental conditions during MIS 3. This review underscores the variable quality of archaeological data reporting, with the number of lithics, assemblage compositions and stratigraphic context often being unclear or unreported. Thus, this study focuses on published sites defined as early Upper Paleolithic or Aurignacoid (based on retouched blades/bladelet production, carinated pieces and bone industry) and, ideally, sites for which cosmogenic dates were provided. Where available, data extracted include site location, age, type (open/cave), techno-typological attribution and radiometric dating (Table 3). Additionally, sites are ascribed a quality value based on the presence of diagnostic artifacts and independent chronological markers. The collation of this database of early Upper Paleolithic archaeological sites provides a broader structure with which to compare and contextualize the Carpathian Basin’s archaeological record. This methodology not only allows for a wider reassessment of early modern human subsistence but also provides the first systematic and synchronic comparison of the role and importance of topography and climate in the modern human occupation of the Carpathian Basin.

Important locations in the Carpathian Basin are found in the regions of Romania (Banat, the Ceahlău Massif/Bistrița Valley and Transylvania), northern Hungary and southern Slovakia, Ukraine (Upper Tisza Valley), Bosnia and Herzegovina (the southwest), and the exceptional karstic cave localities of the Croatian Zagorje and Hungarian Bükk Mountains. Unfortunately, the dissociation of archaeological assemblages and the poor quality of old excavations makes it difficult to assign many of the key sites to specific time frames. Nevertheless, if correct, the radiocarbon ages of the Peștera cu Oase indicate an early modern human presence in the Carpathian Basin reaching back to at least 40 ka ago; it is therefore important for assessing the nature and timing of the earliest modern human occupation of Europe.

Carpathian orogeny has provided the Carpathian Basin with abundant, easily accessible igneous, metamorphic and sedimentary rocks and minerals that potentially enable the spatial and landscape behaviors of the early Upper Paleolithic archaeological record to be explored in considerable depth. However, while there is a long history of research in the region, there are still significant gaps in identifying and sourcing raw materials, even in the best-understood areas (Biró 2009). Still, among the open-air sites, the raw material evidence is strongly suggestive of locally-focused activities with sites located close to or at raw material sources (Table 4).

At the Banat sites (Românești, Coșava, Tincova and Crvenka–At), lithics are primarily manufactured from a variable quality ‘Banat flint.’ Petrochemical analyses indicate a local/meso-local source, possibly in the surrounding highlands (Leonard in prep). In spite of the apparent raw material abundance and quality, it is notable that a small number (< 5%) of the Banat tools and other artifacts were made from other semi-local raw materials. However, the small number and their unknown provenience suggests that they may have been transported from unknown sources upstream.

In the northern Carpathian Basin, the situation is markedly different. Here, the availability of raw materials shifts in favor of varieties of local hydroquartzites and other semi-local stones such as radiolarite, obsidian and erratic flints. Although recent work has focused on identifying potential outcrops in the Northern Carpathian region, there are still questions surrounding the local provenience and classification of raw materials (see articles in Mester 2013). The near exclusive use of poor quality local (< 7 km) hydroquartzites at Nagyréde and silicified sandstones at Egerszalók-Kővágó is in contrast to the other Eger sites (Andornaktálya-Zúgó, Eger-Köporos), where assemblages appear to contain higher quality semi-local and exotic constituents. While it has been suggested that some of this material was transported from across the Carpathians from sources in present-day Poland, the Hornad Valley sites across the border feature nearly exclusively limnosilicites likely acquired from local fluvial sources (Kaminská 2014). The situation is similar farther east at Nižný Hrabovec and Tibava, where lithic raw materials comprise local cherts and claystones, with semi-exotic raw materials such as obsidian also present in small quantities (Kaminská 2014; Williams-Thorpe et al. 1984).

The Upper Tisza Valley has provided further evidence for limited transport and modification of artifacts across the paleolandscape, highlighting a dynamic, mobile and structured technological system. The Transcarpathian Ukraine has abundant high-quality materials, and from the long stratigraphy at Korolevo it may be concluded that the site was used as a raw material source for high-quality andesite throughout the Paleolithic. The situation is identical at Beregovo and Sokirnitsa, where artifacts were manufactured on Beregovo flint or quartzite and slate, respectively. Still, at Beregovo, some flint, chalcedony and andesite are believed to be exotic, possibly imported from Korolevo, some 30 km downstream.

The early Upper Paleolithic archaeology of the Carpathian Basin exhibits a range of technologies that together form a mosaic of regionally distinct, exceptional and familiar assemblages. However, they are still poorly understood as a result of unclear stratigraphies, poor assemblage contexts, lack of chronological resolution and disparate raw material qualities. While blades dominate the lithic artifact record for the early Upper Paleolithic, they are not always the predominant blank type (such as at Nagyréde) and they often vary considerably in forms and production methods, underscoring the roles of raw materials, site function and local traditions.

Several early Upper Paleolithic technocomplexes from the Carpathian Basin are reported aside from the Aurignacian, including the Initial Upper Paleolithic, Bohunician and Olschewian. The Initial Upper Paleolithic is attested to by some of the Eger sites (Eger-Köporos, Egerszalók-Kővágó), but these sites are unstratified and mixed with numerous Late Pleistocene assemblages, and it is unclear how these industries relate either temporally or technologically to other Initial Upper Paleolithic assemblages across Eurasia (Kuhn and Zwyns 2014). Fifteen Bohunician artifacts are also known from Nižný Hrabovec, where they are found within a palimpsest containing assemblages from the entire Late Pleistocene (Kaminská et al. 2000; Kaminská et al. 2009). The assemblages at Vindija and Velika Pecina have sometimes been ascribed to the Olschewian (a regional Aurignacian variant), because of the presence of both bifaces and bone points (Karavanić 2000), but the integrity of the assemblages is still a matter of debate (Zilhão 2009). Additionally, we lack the formal tools with which to accurately designate the unidentified early Upper Paleolithic assemblages from the Upper Tisza Valley; however, the new excavations at Beregovo I may attest to the presence of proto-Aurignacian features.

Still, early Upper Paleolithic assemblages in the Carpathian Basin have been ascribed to the Aurignacian, none more confidently than the Banat sites which contain Dufour bladelets (subtype Roc de Combe), carinated pieces and prismatic cores. However, to what extent these assemblages are directly comparable to those found in Western Europe, and whether they are relevant to discussions concerning the multi-phasic chronology of the Aurignacian, is still unclear. The highly-fragmented nature of the collections and the scarcity of chronometric data renders the chronocultural attribution of the Banat Aurignacian difficult at present.

Because of their close temporal and spatial proximity, some researchers have emphasized the homologies of Tincova, Coșava (level I), Românești I (level III) and Crvenka–At to Western European Aurignacian collections at Krems-Hundssteig in Lower Austria in terms of their techno-typological attributes (Hahn 1977; Mihailović 1992; Mogoșanu 1978). According to a recently resurrected Krems-Dufour interpretation, the Banat sites belong to a specific Aurignacian facies, itself a ‘genuine’ part of a techno-typologically uniform pan-European technocomplex (Demidenko and Noiret 2012).

On the other hand, the Tincova assemblage has been used in discussions of the technology of the earliest modern humans reaching the Carpathian Basin during MIS 3, and Teyssandier and Zilhão have suggested that the collection assemblage is ‘strongly suggestive of the proto-Aurignacian because of the targeted production of elongated rectilinear blade forms’ (Teyssandier 2006, 2008; Zilhão 2006). They have also encouraged additional comparisons with the Kozarnikian further east in Bulgaria, implying that its position might have served as a waypoint between Southeastern and Central Europe. Still, no direct comparative study of any of these sites has been made, and the Tincova site remains undated. If correct, however, these comparisons raise questions as to Tincova’s association with the other Banat sites and may be critical to unraveling the validity of the proto-Aurignacian and various other Aurignacian subtypes.

In caves, the Aurignacian is almost exclusively known from bone artifacts, though some researchers have cautioned against taking osseous points as indicators of Aurignacian assemblages when the complementary lithics are not present. However, if accepted, the bone artifacts form a critical part of our understanding of the Carpathian Basin Aurignacian material culture, as they can be directly dated and are sometimes the predominant or only diagnostic Aurignacian artifact present. Most of the bone artifacts are split-based points, though other forms and tools are present (Dobosi 2002). Bone artifacts are exclusively found in the cave sedimentary sequences, undoubtedly a function of better preservation than in the open-air sites but also possibly due to a system of caching for later use (Verpoorte 2012).

In spite of the long history of research in the Carpathian Basin, the region has little unambiguous evidence for substantial early modern human occupations. This is in contrast to the presence of dozens of early Upper Paleolithic sites and find spots surrounding the Carpathian Basin and across Western Europe. Although the open-air sites of the Banat, southern Slovakia and the Upper Tisza Valley form regional bright spots with large, stratified, sometimes dated early Upper Paleolithic collections in close proximity, there are far more regions where the early Upper Paleolithic record is dim or void of archaeological residues.

This absence is particularly notable in the lowland plains in the heart of the Middle Danube watershed (Fig. 1). This is perplexing, as the regional paleoclimatic studies derived from the loess in exactly these archaeologically-poor areas have borne out the idea that the Carpathian Basin would have been conducive to human occupation. If the Danube is an important trajectory into Europe, and the Carpathian Basin is an important part of this route, why are there so few unambiguous sites? At present, the reason for this may be in part methodological (Anghelinu and Niță 2014):

Notwithstanding these methodological shortcomings, which can be problematic even in the best-studied areas, the conspicuous lack of sites in the Carpathian Basin, particularly in the lowland plain, seems to be a genuine phenomenon for which researchers have suggested three functional hypotheses. They argue that this phenomenon may be the result of:

Understanding the relative validity of these three competing hypotheses is a challenging prerequisite for confronting the role of the Carpathian Basin in early modern human migration patterns.

As for the first hypothesis, the early Upper Paleolithic in the Carpathian Basin truly appears to have particularly low site density, especially when compared to the nearby dense site-concentrations of the Danube in Austria, Moravia and Western Slovakia (Nigst 2006). Low hominin demographics may indeed be a factor in the low findspot density and scanty artifact counts. While it is notoriously difficult to extrapolate paleodemographic estimates, archaeological, climatic and genetic studies agree that during the early Upper Paleolithic they were low indeed (Davies et al. 2015; Fu et al. 2014, 2015, 2016; Seguin-Orlando et al. 2014). However, while low population density may help explain the paucity of archaeological sites, it does not explain the preference for open-air sites in the inner foothills.

As for the second hypothesis, there are few convincing explanations for land-use patterns favoring specific topographic zones, that is, the assumption that the foothills were preferred landscapes. Human foragers subsist through diverse land-use strategies, making use of diverse ecotones as evidenced by their use of both the foothills and high-altitude caves. This shifting between the plane and other altitudinal belts would have allowed, even encouraged, modern humans to make use of the diverse landscapes of the MIS 3 plain to increase their access to fresh water, diverse fauna (including aquatic resources) and raw materials. Using southwestern Romania as an example, Riel-Salvatore et al. (2008) have suggested that higher altitudes positively correlate with artifact curation. Following this logic, we would expect to find higher density sites within the plain itself, a scenario that is not played out in the extant archaeological record.

The validity of hypothesis 3 hinges upon an assumption that there is a systematic taphonomic skew towards the lowlands of the Carpathian Basin, due to either reworking or high Late Pleistocene sedimentation. It is clear from many of the loess profiles in the Danube Valley that several meters of sediments have been deposited since early Upper Paleolithic hominins were present. Additionally, many of the moderately sloped mid-altitude zones where most open-air early Upper Paleolithic sites are located may have experienced significant erosion at the end of the Pleistocene and in the Holocene.

Finally, it is worth pointing out that well-preserved sites and findspots do exist in the plain, for instance at Crvenka–At and potentially Zarilac, which were both identified in alluvial sediments at profound sedimentary depths and only as a result of commercial activity. However, these examples are still by far the exception in the Carpathian Basin, though the many finds and good faunal preservation suggest that there are more buried sites near fluvial terraces and paleolake shores that can yield promising results.

In conclusion, scholars have argued that population size and land-use patterns have influenced the composition of the Carpathian Basin’s early Upper Paleolithic record. However, given the flexibility of early Upper Paleolithic hunters, the availability of resources in the plain and the virtual absence of Pleistocene archaeological material, it is likely that modern humans used a spectrum of the Carpathian Basin’s ecotones and that the absence of archaeological sites in the basin is a taphonomic bias that may be rectified with future research.

Conard and Bolus have asserted that the initial settlement of Europe, especially Central Europe, by modern humans after approximately 40 ka ago was in part achieved by humans ‘rapidly enter[ing] the interior of Europe via the Danube Valley’ into the Swabian Jura. From there, they would have dispersed further west via various forms of socio-cultural dynamics and/or environmental causality, the so-called Kulturpumpe hypothesis (Conard and Bolus 2003). After reviewing current knowledge of the Carpathian Basin’s climatic and archaeological record and the various problems facing both, we return to the question, does the Carpathian Basin early Upper Paleolithic record provide clear evidence for a rapid entry along the Danube corridor during MIS 3?

In short, the Carpathian Basin early Upper Paleolithic record offers limited direct support for this hypothesis beyond the early modern human fossils from the Peștera cu Oase (c. 42.5–40.5 ka cal BP), the lithic assemblages from Românești (c. 42.1–39.1 ka), and to a lesser extent the single bone point from Peskő (c. 42–40.5 ka cal BP), all of which are broadly coeval with the earliest evidence from the Swabian Jura at Geißenklösterle (c. 43.1–41.5 ka cal BP) but post-date the early Aurignacian remains at Willendorf II (c. 43.5 ka cal BP). This suggests that, at present, only a tenuous direct link (spatially or temporally) can be made between these sites. Furthermore, the other modern human remains at Muierii and Cioclovina and the few securely attributed and radiocarbon-dated Aurignacian sites in the Carpathian Basin all indicate dates that are later than those found farther upstream, pointing to a more sustained occupation of the region.

Still, these limited data points do not refute the Danube corridor hypothesis and there is little to suggest that the Danube could not have been a viable corridor into Europe. By the Late Pleistocene, the incised Danube and Tisza river systems would have created high (dry) alluvial banks that would have formed stable conduits for both humans and fauna. Such rivers may have (in parts) occluded modern human movement into areas of the Carpathian Basin; however, given the early Upper Paleolithic sites west of the Tisza, it is clear that rivers did not pose definitive obstacles to modern humans. Based on their known presence in the highlands (e.g. Peștera cu Oase) and the residues of their movement throughout the foothills, it is possible that they circumvented these barriers by fording the shallow, braided feeder rivers along the Carpathian rim. Additionally, if the early lithic remains from Tabula Traiana Cave on the Western bank of the Danube Gorges (at its narrowest point) are telling, we might also envisage scenarios where they traversed large rivers at strategic points where channels narrowed or where river archipelagos were present (e.g. Ada Kaleh Island). Such corridors may have only been periodically available during droughts, or when lower seasonal discharge decreased stream flow and channel breadth/depth exposing point/mid-channel bars, or when sustained cold formed temporary ice bridges. Our current chronological/temporal resolution limits our ability to understand the relationship between early Upper Paleolithic cultural units and Pleistocene rivers, but such a use of the landscape may imply a familiarity inconsistent with a speedy unidirectional dispersal (cf. Hussain and Floss 2016).

The modeled climate for the Carpathian Basin also suggests that there would have been no critical barrier to modern humans entering the region, as is known from periods in the Late Pleistocene (Maier et al. 2016). In fact, it appears that the southern parts of the Carpathian Basin would have been amenable to human occupation even during the colder climatic downswings of MIS 3. This may not have been the case in the Northern Carpathian Basin, where the modeled climatic temperatures indicate a region less temperate than in the south. Nevertheless, climate was not a main challenge for modern humans entering the Carpathian Basin. It is known from the repeated occupations at Willendorf that modern humans were capable of surviving in the temperatures of GIS 11, colder and drier than present-day and harsher than the climates their antecedents endured in the eastern Mediterranean Levant during the Early Pleistocene or even earlier in East Africa. It is also understood that modern humans had mastery of composite extrasomatic technologies such as shelters, clothing and fire, as testified to by the purported habitation structures and hearths at the Slovakian sites (e.g. Aiello and Wheeler 2003; Chu 2009). These would have provided both thermoregulatory benefits against the prevailing cold and wind, and access to locations where fresh water and the associated lithic and faunal resources could be found.

Returning to the most durable part of the archaeological record, the lithics, what can they tell us about the use of the Danube as a corridor? Among the sites located near rivers, for instance in the Banat and southern Slovakia, raw materials were partly tethered to the local fluvial system. However, in other sites, the raw material evidence demonstrates an overwhelming use of locally available lithic sources (despite their quality) for both artifact blanks and tools. There are instances of allochthonous artifacts from southern Poland found in northern Hungary/southern Slovakia, but a guarded interpretation would regard them as stemming from unrecognized local sources or geogenic transport, rather than as anthropogenic traces of long-distance trade systems and wide hunting ranges. If these connections are real, it is interesting to note that they would suggest a path through the Carpathians rather than through the Danube corridor.

While the early Upper Paleolithic lithic assemblages suggest that there was some local flexibility in morphotype, tool ratios, raw materials and flaking capacities, the overall impression is of a blade-dominated industry. Yet where raw materials were of poor quality or small size, hominins manufactured Aurignacian tools on flakes, suggesting that broadly similar hominin behaviors, at least as expressed in lithic technology, were implemented across a range of habitats and environments. While often the attribution of these assemblages to the Aurignacian is clear, ascribing them to specific sub-industries (i.e. proto-Aurignacian, late Aurignacian) is difficult. Furthermore, the association between the Carpathian Basin Aurignacian and the early Upper Paleolithic of the Lower Danube sites in Bulgaria (e.g. Temnata, Bacho Kiro) remains unclear.

The landscape settlement patterns of the Aurignacian in the Carpathian Basin also do not suggest a migrating population forging upstream at a rapid pace. It is true that little is known of the Aurignacian sites in the Carpathian Basin; however, it is known that they were not expediently used. Instead, sites in the Carpathian Basin mostly occupy large tracts of land, sometimes over kilometers long (e.g. Crvenka–At), with considerable artifact density. In certain areas, such as southern Slovakia, the discovery of habitation structures, storage pits and other ‘site furniture’, implies a more sustained occupation at least on a semi-regular basis (Binford 1979).

Conard and Bolus (2003) have inferred that the animal, plant and lithic resources afforded by the Danube River catchment, combined with a continuous source of human populations from the East, would have set the stage for a precocious entry into Western Europe of modern humans and their associated new technological industries (i.e. the Aurignacian and its various relatives). However, drawing migratory arrows linking similar archaeological findspots to track Pleistocene hominin movements has repeatedly led to erroneous conclusions in the past. Though the Carpathian Basin record currently supports the idea of an exogenous, early entrance of the early Upper Paleolithic into the Carpathian Basin unrelated to any of the preceding MP or transitional industries, the dispersal across the Carpathian Basin is not suggestive of rapid demic expansion, as is evidenced by the relatively late hybridization of the Peștera cu Oase fossil and implied by persistent Mousterian technological elements (Fu et al. 2015; Horvath 2009; Noiret 2005).

This begs the question of where the makers of the early Upper Paleolithic in the Carpathian Basin came from. Aside from a handful of Aurignacian sites (e.g. Bacho Kiro, Temnata) and Kozarnika, whose link to the Aurignacian remains tenuous, no other sites directly connect the Carpathian Basin Aurignacian to the south in the Balkans. Additionally, Anatolia has also to provide empirical evidence of a connection between Southwestern Europe and the early hominin technocomplexes of the Levant. Therefore, a western source for the Carpathian Basin early Upper Paleolithic is conceivable, especially considering the early Willendorf dates which, if correct, pre-date any of the evidence in the Carpathian Basin. If the Danube was as easy a conduit as has been suggested, it is equally likely that it may have seen hominin movement in the opposite direction (Sitlivy et al. 2014). Indeed, increasing genetic and archaeological evidence (Adler et al. 2008; Anikovich et al. 2007; López et al. 2016) supports the idea that the earliest modern humans coming from the Middle East and into Europe may have bypassed Southeastern Europe (at least overland), opting for a route running through the Caucasus, dispersing east through the East European Plain and then north of the Carpathians.

The recent reanalysis of Central European early Upper Paleolithic assemblages and possibly Initial Upper Paleolithic sensu lato finds farther east in the geographically connected Moravian Plains (Bohunician) suggests that early modern humans were present in Central Europe far sooner than previously recognized (Müller et al. 2011; Nigst et al. 2014; Richter et al. 2008, 2009). This notion could lead to a major modification in our understanding of the origin and cultural ontogeny of the Aurignacian technocomplex (Sitlivy et al. 2014).

This suggests that while hominins were undoubtedly present within the Middle Danube catchment in the late Upper Pleistocene, it is currently difficult to tell from the archaeological record whether they entered the Carpathian Basin on direct ‘highways’, in waves (Hublin 2015), or more piecemeal; furthermore, the evidence is too sparse to suggest a directional trajectory. Indeed, the gap in the Danube record suggests that the situation may be more complicated than has previously been thought. Furthermore, climatic reconstructions, illustrated by advances in loess stratigraphy, faunal/floral records and geochemistry, suggest a necessary diversion from the rugged karstic regions of the basin that may have been more familiar hunting areas for previous (Neanderthal) populations. This may have resulted in more frequent or seasonal use of the lowlands within the earlier parts of MIS 3 that may have prompted subsequent modifications in hominin subsistence behavior. A prolonged/intense modern human presence in the Carpathian Basin throughout the late Upper Pleistocene is testified to by a higher frequency of lithic sites with increased artifact density. Increased sedimentation rates in the later part of the Pleistocene may have also helped to offset the palimpsest effect that might have skewed the record.

Archaeological research has confirmed the presence of modern humans and early Upper Paleolithic sites in scattered parts of the Carpathian Basin. Despite the importance of the Carpathian Basin early Upper Paleolithic to paleoanthropological models, there has been no systematic attempt to examine how Carpathian Basin industries figure in linking the Aurignacian of Western Europe and the Bachokirian and Kozarnikian of Southeastern Europe. In situations such as that of the earliest occupation of Europe by modern humans, where the chronological resolution is insufficient to transform the archaeological record into a coherent cultural narrative, it is appealing to use large geographic features such as the Danube to explain cultural migration routes. While rivers may have served as valuable migration axes in the early Upper Paleolithic and even earlier (Ashton et al. 2006), they do not preclude other productive biomes, which may have been welcome alternatives to climatic variations in both annual and longer timescales. Furthermore, if the Danube did play a role in the migration route of modern humans in the Carpathian Basin, we currently lack the archaeological resolution to tell us in what direction it may have been used.

This paper has reviewed the geological, climatological and archaeological record of the Carpathian Basin, summed up in the following points:

Given the sparse archaeological record, it is clear that robust testable hypotheses concerning the early Upper Paleolithic record of the Carpathian Basin can only be formulated with the location of new sites as opposed to the continued analysis of old excavations, for which most stratigraphic/contextual information has now been lost (e.g. the Bükk and Zagorje cave sites). Such an approach includes broadening our search to areas where little research has been conducted, paying particular attention to the basin margins, intermountain basins (e.g. Požega Valley, Croatia), paleolake shores and fluvial terraces where loess and other fine-grained mantles are thin or accessible (Tourloukis 2016). Such efforts may be amplified by predictive modeling (e.g. using slope, aspect, distance to water and lithic resources), systematic amateur stray-find recording, and shallow geophysical techniques. Efforts may also productively be directed towards the re-excavation of ‘flagship sites’, focused on obtaining radiometric dates and understanding their site formation processes. Both approaches are essential to building reliable archaeological distributions and chronology for the Carpathian Basin early Upper Paleolithic record. Integrating these methodological tools would enhance the reliability of model-bound approaches, giving us a better understanding of how and when the earliest modern humans entered Europe.