The Chronostratigraphy of the Aurignacian in the Northern Carpathian Basin Based on New Chronometric/Archeological Data from Seňa I (Eastern Slovakia)

The northern Carpathian Basin has important geological, paleoenvironmental, and archeological records that are key to our understanding of the first modern human occupation of Europe. However, the nature and timing of hominin settlements in the region during the Late Pleistocene remain poorly understood. New fieldwork at Seňa I, Slovakia, has identified the only known stratified open-air Aurignacian site in the region and assigned the assemblage to at least 33.5 ± 2.4 ka ago through infrared-stimulated luminescence dating. Additionally, new archeological and geoscientific data are presented. This paper discusses the Seňa I findings in the context of the regional archeological record and shows how they contribute to the establishment of clear time constraints for the Aurignacian in eastern central Europe.


Introduction
Research over the last decade has improved our understanding of the European geological, environmental, and archeological records for the Late Pleistocene. The northern Carpathian Basin is an important region for this period as it forms part of a potential crossroad where Middle Paleolithic (e.g., Mousterian, Jankovician) and early Upper Paleolithic (e.g., Bohunician, Szeletian, Aurignacian, and Gravettian) technocomplexes are described (Chu 2018 and references therein).
Despite increased research, the chronology of the Late Pleistocene archeological record of the northern Carpathian Basin remains poorly understood. In the Bükk Mountains, recent attempts to clarify the age estimates and cultural successions of the famous Szeleta, Istállós-kő, and Pes-kő caves have highlighted a scarcity of diagnostic artifacts, a poor understanding of stratigraphies, and difficulties in producing meaningful numerical chronologies (Bánesz 1989;Davies and Hedges 2008;Hauck et al. 2016;Lengyel and Mester 2008;Markó 2015;Patou-Mathis et al. 2016). In the surrounding landscape, open-air sites are known in and around the North Hungarian Range but are mostly unstratified, and secure radiometric ages are still unknown (Dobosi 1995(Dobosi , 2013Kozłowski et al. 2009;Kozłowski et al. 2012;Lengyel et al. 2006;Markó et al. 2002). In nearby northwestern Romania (c. 175 km southeast), reexaminations of the sites of Boinești, Bușag, Călinești, and Remetea Șomoș I & II suggest that, while there may be Aurignacian artifacts in the assemblages (Anghelinu and Niţă 2014;Dobrescu 2008), they have been reworked with Mousterian and Gravettian elements and therefore cannot be confidently separated and dated (Tuffreau et al. 2013).
Thus, the main goal of our work was to reexamine the Paleolithic record of the northern Carpathian Basin. Following test pitting at multiple sites , we present the results of a systematic excavation at the site of Seňa I that allowed us to obtain new lithic material, re-examine their stratigraphic context and acquire radiometric ages. Our results indicate the existence of the only known open-air stratified Aurignacian assemblage in the northern Carpathian Basin and establish an age consistent with chronological ages from the surrounding regions. This timing has implications for the regional modern humans' occupation of eastern central Europe.

Background
The site of Seňa I (48.556307, 21.254560; 196 m asl) is located in eastern Slovakia 3 km north from the Hungarian border at the boundary of the villages of Seňa and Kechnec (Košice Region) on a terrace of the Hornád River. The site was discovered in 1951 and excavated in 1955 by L. Bánesz (1958, p. 195;1989). The main excavation area (45 m 3 ) was located in the northwestern corner of the Seňa cemetery where six clusters of Paleolithic artifacts were found (Fig. 1). Fifty meters further north on a gravel outcrop, an artifact bearing pit was also excavated. Bánesz noted a simple stratigraphy starting with a topsoil (40-50 cm), followed by a 20-30 cm layer of yellowish loess containing scattered cold-adapted malacofauna and charcoals. Archeological finds were located at the base of the yellowish loess at the interface with the underlying terrace gravels.
Lithic artifacts (N = 1211) were made from heavily patinated local limnosilicites (i.e., limnoquartzite), though radiolarite tools and worked silicified sandstones were also present (Kaminská 2013). Limnosilicites, radiolarite, and sandstone are mesolocal in origin (Slanské Hills; c. 25 km, Pieniny Klippen Belt; c. 65 km; Vihor Hills; c 80 km, respectively) but are present in the riverbed near the site. Both cores and their associated knapping residues were recovered. Flake production predominated the assemblage (> 85%; Bánesz 1956 p. 626), and the 51 tools were mostly endscrapers including three carinated forms. Bánesz ascribed the assemblage to the Middle Aurignacian. Hahn (1977) later suggested that it was Aurignacian, but only tentatively, due to the absence of clear fossiles directeurs due to the poor raw material quality and lack of faunal preservation (e.g., Dufour bladelets, strangled blades, bone points).

Excavation
In July of 2016, four test trenches were installed surrounding the original Seňa I excavation site with the aims to examine the stratigraphy, obtain new archeological material, and produce radiometric ages (Fig. 1). Fifteen square meters were excavated in four trenches at 2 cm spits (that did not traverse geological horizons) to a depth of8 5 cm (the top of the gravels). All objects above 5 mm were recorded with a total station and the collected sediments from quarter square meters of which 25% were wetsieved (1.2 mm mesh) for smaller artifacts. The subsequent lithic analyses focused on measuring artifacts (i.e., length, width, thickness, and weight) and describing features specific to known archeological cultures including technology and typology (sensu Demars 1992;Inizan and Féblot-Augustins 1999).

Stratigraphy and Sedimentology
The stratigraphy, texture, and sediment composition were recorded in the field. Grain size measurement was conducted using a laser diffraction particle size analyzer (Beckman Coulter LS 13320 PIDS) and by calculating the mean diameters of the grains within a range of 0.04-2000 μm. Each sample was measured twice with two different concentrations of sediment in the solution to increase accuracy. Measurements with reliable obscuration (8-12%) were averaged and are previously reported and discussed in Chu et al. (2018).
Here, end member modeling analysis (EMMA) of the entire set of the previously reported grain size data was performed using the EMMAgeo R package (Dietze et al. 2013; R core team 2018) to ascertain the sedimentary processes and post-depositional processes of the sediments. EMMA is a powerful and flexible multivariate statistical method used to identify and quantify sediment transport processes from multimodal grain size distributions. The grain size data used are available in the Pangaea database (Zeeden et al. 2016b).

Age and Chronostratigraphy
Three sediment samples, from which polymineral fine grains (4-11 μm) were extracted for luminescence measurements, were collected at and above the main archeological finds between 60 and 85 cm depth below surface (Fig. 2) and were prepared for luminescence dating using standard techniques (Frechen et al. 1996). Measurements were carried out on an automated Risø TL/OSL DA 20 reader equipped with a calibrated 90 Sr beta source. Post infrared stimulation measured at 290°C (pIRIR 290 , Thiel et al. 2011) was carried out with infrared diodes (870 nm, FWHM = 40), and the signals were detected through an interference filter (410 nm; SI1a-d). The initial 4 s of the signal minus a background of the last 20 s was used for pIRIR 290 dating. Laboratory experiments included prior-IR stimulation temperature tests (Buylaert et al. 2012) using a range of temperatures from 50 to 220°C and dose recovery tests (SI2; SI3a-b). For this dose recovery test, samples were illuminated for 24 h in a Hönle SOL2 solar simulator, and a laboratory dose in the range of the natural dose was given to the samples. Additionally, infrared stimulation measured at 50°(IR 50 , Wallinga et al. 2000;Preusser 2003) was carried out for the three samples (SI1a-d). Equivalent doses were calculated with an arithmetic mean. Fading tests (Auclair et al. 2003) were carried out for all samples for the IR 50 and the pIRIR 290 measurement protocols. Only the IR 50 ages were corrected for fading following the approach of Huntley and Lamothe (2001). The environmental dose rate was measured using high-resolution gamma ray spectrometry. We have calculated the gamma dose rate for sample C-L4260 after Aitken (1985, Tab. H1) to allow for dose rate heterogeneity arising from the underlying gravels and by assuming that the gravels are inert. The environmental dose rate was calculated using the DRAC v.1.2 (Durcan et al. 2015) using conversion factors of Guérin et al. (2011), an a value of 0.07 ± 0.02 (IR 50 , Preusser 2003) and 0.12 ± 0.02 (pIRIR 290 , Schmidt et al. 2018) and the measured water content. The internal beta dose rate contribution of the polymineral fine grain samples was calculated by assuming a potassium content of 12.5 ± 0.5% (Huntley and Baril 1997). Alpha and beta attenuation factors of Bell (1980) and Guérin et al. (2012) were used. The cosmic dose rate was calculated following Prescott and Hutton (1994). Results are presented in Table 1.  Table 1 Dose rate data, equivalent dose values (9.8 mm aliquots), and luminescence ages. Mbs meters below surface, W% measured water content (water mass over dry sediment mass), U uranium, Th thorium, K potassium; age calculation is based on cosmic dose calculation after Prescott and Hutton (1994), conversion factors of Guérin et al. (2011), and the measured water content; De equivalent dose. The internal beta dose rate contribution of the feldspar samples was calculated by assuming a potassium content of 12.5 ± 0.5% (Huntley and Baril 1997). Samples taken from trench 3 eastern wall (see Fig. 2 Additionally, AMS 14 C dating of three micro-charcoal (342-0.02 mg) samples found above the find level was carried out by BETA analytics, and the results using the INTCAL13 calibration curve (CalPal) are displayed in Table 2. Samples were pretreated by gentle crushing and dispersal in deionized water where they were then washed in hot acid (HCl) then alkali (NaOH) then acid solutions and dried.

Stratigraphy and Sedimentology
There was considerable spatial variability of the preserved sediment cover in spite of the same geological succession. While trench 1 showed~50 cm of sediments above the terrace gravels (with the interval from 45 to 55 cm being a mixture of gravels and sediment), trench 2 showed 75 cm of sediments. The northern wall of trench 3 had the thickest sediment cover of 85 cm. Trench 4 exhibited 25-30 cm of sediment covering the gravel.
Here, we focus on trench 3 because it was the largest excavated trench with the thickest sediment cover and the highest artifact density (Fig. 2). At Seňa I, trench 3, the excavations revealed an upper 30 cm representing a middle-brown plow horizon of a soil with a partly sub-polyhedral and partly friable structure and ∼ 3-10% small to medium pebbles (up to ∼ 4 cm diameter) embedded (level 1A). The upper 20 cm contained more pebbles than the interval from 20 to 30 cm depth where their contribution decreases to~1%. The interval from~30 to 44 cm was a lighter brown plow horizon slightly enriched with clay (level 1B) with no pebbles. From~44 to 64 cm, a layer of loose loess (−like) sediment occurred containing a few small pebbles (level 2). The ochre-light brown color likely indicates a weak pedogenic overprint. From~64 to 85 cm, a slightly reddish-dark brown paleosol was observed containing about 5% pebbles (level 2/3), followed by the terrace gravels (level 3). Grain size data show a multimodal distribution (Chu et al. 2018, their Fig. 9). Besides a peak in the grain size around 20-40 μm, Seňa I also shows a coarser sandy component and an elevated clay content.
The results of the EMMA are 2-4 end members that describe most of the data variability, reducing data dimensions from the original 113 grain size classes (with variable data). Because the explained overall variability only slightly increases with a 4th end member, we use three end members here that in sum explain 58%, 81%, and 84% of overall data variability, respectively. The three end members show (a) an end member with a dominant peak in the sand fraction centered at~550 μm explaining4 0% of end members, (b) an end member with dominance of a peak at~20-30 μm explaining~30% of end members, and (c) a component including side peaks of the previous components in the sand and silt fractions along with a strong clay/fine silt content in the~0.2-10 μm grain size fraction representing~29% of end members.

Excavation
The excavated assemblage consists of over 278 artifacts (21 found in the wet sieving) along with two faunal remains. The latter were fragments smaller than 20 mm that were not further identifiable, and their poor surface preservation prohibited any assessment of anthropogenic modifications. No malacofauna remains were recovered. Three small ochre fragments (c. 10 mm in diameter) were also found in the excavation. Lithic artifacts were mostly flakes and blade fragments or knapping debris made of heavily patinated limosilicites though some small pieces (N < 5) of radiolarite and obsidian were also recovered. Lithic artifacts indicative of an Aurignacian assemblage were three prismatic blade cores (Fig. 3), a nosed endscraper, various other forms of endscrapers, burins (Fig. 4), blade fragments (Fig. 5), and retouched flakes. Furthermore, the lithic artifacts varied in size (7-83 mm in maximum length) and weight (> 0.01-115.6 g) suggesting minimal redeposition reflected in typical differential movement of objects of different sizes (i.e., no size sorting). There was no clear pattern to horizontal artifact distribution, and there were no evidences for pits or other features.
The three presented radiocarbon ages show no chronological pattern and are found higher than the luminescence samples, in or at the interface of plowed sediments (Table 2; Fig. 2). The small size (324 mg) and infinite age of Beta-444,100 suggest that it is likely geological (e.g., much older coal) as no visible striations were observed. The other samples were unidentifiable micro-charcoals (< 2 mg) that can be mobile in sediments and were not retrieved from in situ clearly visible hearths and were from later events that were incorporated into the plow zone during tillage. Therefore, these radiocarbon samples have no secure association with an archeologically recognizable event that could be related to an Aurignacian occupation.

Stratigraphy and Sedimentology
At Seňa I, the grain size distribution (maximum distribution of ∼ 20-40 μm) of the sediment cover is highly similar to the pure loess sample from the well-studied reference site of Semlac, Romania (c. 200 km south; Chu et al. 2018;Zeeden et al. 2016a;Zeeden et al. 2016b). Eolian deposition from nearby river plains is common during cold climatic periods in the vicinity of rivers at the northern edge of the Carpathian Basin (see e.g., the literature on loess around Tokaj; e.g., Sümegi and Hertelendi 1998). Due to the site's position on a water permeable terrace, eolian sediment erosion probably did not occur during initial sedimentation on top of the gravels where the majority of the artifacts were found, but may have developed later with increased eolian deposition and heavy precipitation. There is an additional eolian coarser sandy component interpreted to come from the close-by Hornád River flood plain. The topographic position on a river terrace makes a major contribution of slope material unlikely because of both low slope angle and high water permeability of the underlying terrace and the eolian sediment. A smaller clay component was also present that is typical for pedogenetically overprinted loess in the region (e.g., Bodrogkeresztúr; 45 km south; Bösken et al. 2017). In the lower levels 2 and 2/3, pebbles (5%) occur in the eolian sediment mixed in from the underlying terrace gravels based on similar size, shape, and lithology. Larger stones are present in level 1A and are abundant on the surface that were brought into the soil by regular plowing from the terrace edge where sediment cover is thin and the topsoil is mixed with stones from the terrace through plowing. The three end members of the grain size distribution are depicted in Fig. 6 and describe fine (clay and fine silt; EM3), silt (EM1), and sand fractions (EM2). The topmost plow zone samples are explained through a combination of the end members representing clay and sand indicating eolian sediment with some sand that may originate from stronger winds during the Holocene (Obreht et al. 2015). A dark gray chernozem-like soil is found at the top, but is disturbed through plowing in the upper 45 cm. While the boundaries between sediment packages were spatially undulating, bioturbation was minor, and no crotovinas and major root channels were observed. The lowermost sample (level 2) can be approximated by the end members representing sand and silt, where sand from the underlying terrace was mixed with the eolian sediment.
The sedimentological findings and the here presented interpretation suggest that lithic artifacts were deposited close to the gravel surface and later covered by more eolian sediment deposits. Whether artifacts were deposited directly on the gravel surface or on an existing weak paleosol cannot be reconstructed from the limited excavation and also due to apparently slow sedimentation. Post-depositional processes later affected the vertical distribution of material through both mixing with the terrace gravels and sand (lowermost of the eolian sediment sample contains sand probably from terrace material), and much later plowing of the profile top which distributed stones including artifacts from the terrace edge over a wider area. This would be consistent with the open-air nature of the site and can explain the low density of lithics and the heavy patina. The upward fining of artifacts and pebbles in the lower profile corresponds with this scenario.

Excavation
The results of our excavation were similar to those of Bánesz in terms of position, artifact density, raw materials, and typo-technology, allowing a straightforward correlation with the 1955 excavation. Disregarding the subdivisions of our stratigraphy and varying depths, the geological levels agree with those previously observed. While our main find concentration was in contact with the terrace gravels, we also found that artifacts continued 15 cm upwards and that errant typologically analogous artifacts were found higher up in the stratigraphy and on the surface. The latter is due to truncation and redeposition by current and historical plowing of zones (c. 45 cm) where the sedimentary sequence was notably shallower (e.g., trench 4). While excavated finds may have migrated as a result of bioturbation within the paleosol and also due to mixing with underlying gravels, the precision and coherent chronological succession of our luminescence ages, combined with sedimentostratigraphy, demonstrate an integrity of the main find level sediments and indicate a discrete archeological deposition with limited reworking. Furthermore, the spread in ages over a shallow sequence suggests a sustained period of occupation under relatively low sedimentary deposition. The low sedimentation rates compared with sites within the Carpathian Basin can be explained by the limited sediment input from the Hornád River floodplain.
While the excavated trenches were too small to reliably identify horizontal artifact concentrations, they did reveal variable artifact density with some areas yielding higher artifact density than others. This suggests differences in spatial variability potentially representing activity areas, but the stochastic distribution of artifact types does not allow us to make further inference.
Artifacts were primarily manufactured on local limnosilicite that was later heavily patinated possibly as a result of long aerial exposure under fluctuating climate and environmental conditions (Glauberman and Thorson 2012;Thiry et al. 2014) that may be inferred from our short sedimentary sequence over a long timeframe. There was a single radiolarite core and a single knapped obsidian nodule from the site; it is remarkable that obsidian was so infrequently used as the raw material source is semilocal (40 km) and as, curiously, obsidian from the same source is frequently found at Aurignacian sites in Romania some 170 km away (Dobrescu et al. 2018).
The artifact types throughout the sequence are consistent with Bánesz's collection, an assemblage with flake and blade production and many endscrapers including a nosed endscraper. Combined with the artifacts from Bánesz's excavation, we can (re-)ascribe this assemblage to the Aurignacian from central Europe based on blade production, burins, and endscrapers (including typical carinated/−nosed endscrapers; Davies 2001) and the absence of both earlier and later fossiles directeurs such as leaf points for the Szeletian, Levallois-like core technology for the Bohunician, and backed bladelets for the Gravettian (Moreau et al. 2015;Škrdla 2017b;Svoboda 2006). Due to the relative absence of further diagnostic tools (e.g., specific burin types, Góra Puławska II-type microliths; Demidenko et al. 2017), it is currently not possible to ascribe it further to subindustries.

Age and Chronostratigraphy
The chronostratigraphic position of Seňa I indicates by proxy that modern humans were present in the northern Carpathian Basin by 33.5 ± 2.4 ka ago based on the lithics overlying the gravels and the associated luminescence ages of the embedding sediments (Table 1, C-L4260). As the luminescence ages represent the date of the sediment deposition, however, the human occupation may be older. Artifacts are also found higher in the same level at the date of 24.0 ± 1.6 ka (C-L4259) and may indicate a sustained occupation of the site, though it is possible that they may have in part migrated upwards due to frost heaving or other processes. No artifacts were found in level 2 which was dated to 15.5 ± 1.1 ka ago (C-L4258). As occasional artifacts were found higher up in the contact zone between 1b and 2 as well as on the surface, we attribute these to reworking by plowing of areas where the sequences were notably shallower (e.g., trench 4).
In the Transcarpathian Ukraine, blade industries at Korolevo (I level Ia; II/II) and Sokirnitsa IA level 3 have been dated to around 39 ka B.P. through pollen, radiocarbon, TL, and paleomagnetic dating, though the various proxies are not always in agreement and the conflicting evidence suggests that they may be younger (Monigal et al. 2006;Usik et al. 2003). Additionally, the absence of diagnostic tools at both sites makes it difficult to ascribe them to the Aurignacian and they may be culturally related to the Szeletian or Bohunician. However, if the assemblages are indeed younger, Transcarpathia may be connected to the Hornád Valley sites by Ukrainian Volhynia flint at several other sites in the Hornád and by vitrophyric dacite (i.e., andesite) from Korolevo at the Aurignacian site of Kladzany in the Ondava river basin (Kaminská et al. 2000).
West of the northern Carpathian Basin in Lower Austria, the earliest Aurignacian in Europe at Willendorf II AH3 was modeled to 39.5 ky (uncal) B.P. based on climatostratigraphy and radiocarbon ages (Haesaerts et al. 1996;Nigst et al. 2014), but both the chronology and cultural attribution of the assemblage have been contested (Teyssandier and Zilhão 2018;Zilhão and d'Errico 1999). Furthermore, the ages are old for other dated Aurignacian assemblages in the area (e.g., Krems-Hundssteig, Senftenberg, Grossweikersdorf) that are assigned to between 34 and 31 ky (uncal) B.P. (Nigst and Haesaerts 2012;Wild et al. 2008).
In sum, while there are a range of varying ages for the early Upper Paleolithic in the northern Carpathian Basin and its immediate surroundings, direct numerical chronologies for hominin remains and secure Aurignacian assemblages all center around 34-29 ky (uncal) B.P.

Implications
Our results, along with the high concentration of other large, close-by sites (e.g., Barca and Kechnec localities and other findspots), indicate that by 34 ka B.P., the Aurignacian technocomplex was well established along the Hornád Valley as part of an expanding population in eastern central Europe with the site reflecting a potential seasonal population (Schmidt and Zimmermann 2019;Svoboda 2006) in an area where abundant resources were located within a modest walking radius (< 5 h; Hauck et al. 2018). If the purported habitation structures, storage pits, and other "site furniture" from Barca (I, II, and III), Seňa I and Tibava are indeed supported (which is still controversial); they would also strongly intimate a sustained occupation at least on a semi-regular basis (Binford 1979).
For the moment, our ages contribute to a minimum age for the Aurignacian assemblage at Seňa I, and consequently cannot contribute to the question of synchrony with preceding regional industries such as the Szeletian or Bohunician. However, the notable absence of these earlier technocomplexes from the Hornád Basin that are well known from neighboring regions suggests that Aurignacian settlements in the region do not clearly map onto preceding (or subsequent) technocomplexes and may be the result of preferential landscape use, possibly mitigated by changes in climate and environment (Staubwasser et al. 2018).

Conclusions and Future Perspectives
The presence and timing of the early Upper Paleolithic in eastern central Europe remains a vital part of understanding the nature of transitional industries and the first modern human occupation of Europe during the mid-Pleniglacial. New excavations combined with sedimentological analysis at the site of Seňa I reconfirm the presence of a rare stratified Aurignacian site in the region. Robust radiometric ages from the main find level confirm a chronology for the assemblage before 33.5 ± 2.4 ka.
These results corroborate earlier radiometric ages on hominin remains and archeological assemblages from the surrounding regions. While our current work adds to the corpus of early Upper Paleolithic research in central Europe, future efforts should continue to focus on the identification and dating of stratified archeological sites where lithic/faunal studies can contribute to our understanding of the relationship between the various early Upper Paleolithic assemblages where clear delineations are either blurred or not fully understood. However, such sites are exceedingly rare in the Carpathian Basin, and in their absence, future research at sites such as Seňa I should exploit geoarchaeological advances in physical and chemical techniques such as micromorphology, magnetic susceptibility, and geochemistry that may help to further elaborate site formation processes and their implications for early hominin behavioral patterns.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.