Introduction
Discussion about the Paleolithic rock art in Siberia and Central Asia is still ongoing (Sher, Reference Sher1980; Zotkina et al., Reference Zotkina, Tekhterekov, Kharevich and Plisson2014, Reference Zotkina, Kolobova, Sutugin and Olsen2023; Molodin et al., Reference Molodin, Geneste, Zotkina, Cheremisin and Cretin2019, Reference Mylnikova, Parkhomchuk, Molodin, Menshanov, Babina, Nenakhov and T.A2024; Miklashevich, Reference Miklashevich2020; Ponomareva, Reference Ponomareva2021; Zotkina and Sutugin, Reference Zotkina and Sutugin2023; Abolonkova et al., Reference Abolonkova, Zotkina and K.Ain press). Since a group of naturalistic zoomorphic images from open-air sites of Southern Siberia remains undated, there is still room for debate (Supplementary Information 1).
The rock art of the Minusinsk Depression (Krasnoyarsk Region, Southern Siberia, Russia) (Fig. 1) includes stylistic assemblages reliably dated from about 4.5 ka to the ethnographic period (Zotkina et al., Reference Zotkina, Kolobova, Sutugin and Olsen2023, fig. 2). All of them have been attributed based on proven relationships between the images and related archaeological materials (Supplementary Information 1.B). The earliest known rock art in the region is the so-called “Minusinsk Style” (Podolskiy, Reference Podolskiy1973). However, due to the lack of established association between these images and archaeological data, the Minusinsk Style has neither been dated nor attributed (Miklashevich, Reference Miklashevich2020). The Minusinsk Style exclusively features zoomorphic images distinguished by their archaic character and naturalism (Supplementary Information 1, Figs. 1 and 2). In multilayered compositions, they are always overlain by younger carvings, and no instances of the reverse have been recorded so far (Zotkina, Reference Zotkina2019). These are the main indicators of the earliest age of the Minusinsk Style, but it has never been accurately defined (Miklashevich, Reference Miklashevich2020, p. 93; Zotkina et al., Reference Zotkina, Kolobova, Sutugin and Olsen2023, fig. 3, 6) (Fig. 2; Supplementary Information 1, Figs. 1 and 2).
Location of the main sites (red dots) with the earliest rock images in the region.
The earliest stylistic assemblage rock art examples. Maydashy rock art site (Minusinsk Depression).
The archaeological context of the earliest rock art in the region is rather unclear. This territory has been inhabited since the Middle Paleolithic, with the Dvuglazka cave (80–40 ka) being the only site known for its Levallois industry. Malaya Syya (38–33 ka), Sabanikha (31–30 ka), and other Early Upper Paleolithic sites were discovered in the Minusinsk Depression (Abramova et al., Reference Abramova, Astakhov, Vasil’ev, Ermolova and N.F1991; Lisitsyn, Reference Lisitsyn2000; Lbova et al., Reference Lbova, Panov, Zenin and Barkov2015; Kharevich et al., Reference Kharevich, Kharevich, Zolnikov, Klementiev, Zotkina, Anoikin and Akimova2024). The Final Upper Paleolithic sites are more abundant in the region and form two archaeological cultures: Kokorev (15.5–11.5 ka) and Afontov (15–12.5 ka) (Abramova, Reference Abramova1979a, b). As we can see, the Upper Paleolithic archaeological context does not exclude the possibility of occurrence of the Pleistocene rock art in the region (Sher, Reference Sher1980).
Unfortunately, due to extremely limited data, the period between 11 ka and 6 ka appears as a blank spot in the local cultural chronology. However, the presence of Neolithic hunter-gatherers was recorded around 6 ka (Unuk, Maliy Kyzykul’, and Vyuzhnoye settlements). In addition, a new population with the same economic type connected to the Western Siberian Neolithic cultures came to the region between 6 ka and 5 ka (Karasevo, Tagarsky Ostrov and Malaya Minusa) (Vinogradov, Reference Vinogradov2022). Hence, the earliest rock art also could have been produced by Neolithic hunters-gatherers (Sovetova and Miklashevich, Reference Sovetova, Miklashevich and Bobrov1999).
Around 5 ka years ago (2900–2500 BCE), the cattle-herding population of the Afanasiev culture that was related to the pastoral societies in the west (Yamnaya cultural-historical community) appeared abruptly in the Minusinsk Depression (Polyakov, Reference Polyakov2019; Polyakov et al., Reference Polyakov, Svyatko and Stepanova2019). For a while, this population coexisted with the autochthonous societies. Despite the abundance of excavated sites and data on the Afanasiev culture (Stepanova and Polyakov, Reference Stepanova, Polyakov, N.F. and Polyakov2010), no significant examples of its art in situ have been discovered, not to mention the rock art specifically. Therefore, the hypothesis that the Minusinsk Style belongs to the Early Bronze Age (Afanasiev culture) (Kovtun, Reference Kovtun2001; Rusakova, Reference Rusakova and Devlet2005; Esin, Reference Esin, Stepanova and Polyakov2010) has neither been confirmed nor disproved.
Nevertheless, researchers agree on one event that indicates the upper chronological boundary of the earliest rock art in the region: the Okunev culture appearance in the Minusinsk Depression about 2600–2500 BCE (Lazaretov, Reference Lazaretov2019). Its pastoral population also came from the west and brought a well-established material and spiritual culture (Lazaretov, Reference Lazaretov and Soyenov2017, Reference Lazaretov2019), including the peculiar tradition of monumental art that carries no resemblance to the preceding style (Miklashevich, Reference Miklashevich2020) (Supplementary Information 1).
Thus, the hypothetical age of the Minusinsk Style ranges from Upper Paleolithic to Neolithic (15–5 ka) and Early Bronze Age (4.9–4.5 ka), although none of them has been proven so far (Zotkina et al., Reference Zotkina, Kolobova, Sutugin and Olsen2023; Abolonkova et al., Reference Abolonkova, Zotkina and K.Ain press) (Supplemental Information 1). The only reliable chronological boundary for the petroglyphs has been the Okunev rock art tradition (started as early as 4.5 ka), which is completely different in style (Supplementary Information 1.B).
Since the earliest rock art has not been properly dated based on archaeological evidence, the focus was shifted to alternative methods. Geomorphological dating, which considers geological events, sedimentation and erosion processes as chronological markers of lower and upper age limits for the images left on rock surfaces, is an effective strategy for dating rock art research (Philips et al., Reference Phillips, Flinsch, Elmore and Sharma1997; Stuart, Reference Stuart, Pollard and Brothwell2001; Aubry et al., Reference Aubry, Dimuccio, Mercè Bergadà, Sampaio and Sellami2010; Huyge et al., Reference Huyge, Vandenberghe, De Dapper, Mees, Claes and Darnell2011; Delannoy et al., Reference Delannoy, Bruno David, Genuite, Gunn, Finch, Ouzman and Green2020; Deline et al., Reference Deline, Ravanel, Delannoy, Le Roy, Molodin, Cheremisim, Zotkina, Cretin, Geneste and Plisson2020, Reference Deline, Ravanel, Delannoy, Le Roy, Team, Léanni and Cheremisin2023).
In order to date the earliest rock art known in the region, we studied the geomorphological context of the sites where the Minusinsk Style petroglyphs were found. The Krasnoyarsk water reservoir built on the Yenisei River in 1970s has had a great effect on the valley having deeply damaged its nature. Since most of the Minusinsk Style images belong to the sites located on the reservoir’s shoreline (Fig. 1), they could not be studied by geomorphological methods, unlike the Maydashy site located upstream from the reservoir in the non-affected part of the Yenisei Valley (Fig. 1).
To date the surfaces on which the Maydashy images were carved, we studied their geomorphologic position and the geological nature of the sediments in the surrounding area. The site is located on a rocky outcrop of about 250 m long and up to 50 m high. The surface is up to 5 m in height and located on the right bank of the Yenisei River, across from Sokhatik Island (Fig. 3). The images are 1–1.5 m above the current water level, so from time to time they are covered in sediments during high floods, which have been repeatedly recorded by researchers (Fig. 4).
Hypsometric map of the Maydashy site (marked with a star) and studied geological sections; the figures mark the observation points (OPs) described in the paper (a) General view of the site; (b) floodplain terrace development area; (c) locations of altitude profiles through OP 4; (d) OPs 1, 5, and 6; (f) OP 2. Altitude profiles shown in Figure 5.
One of the surfaces with the earliest images (aurochs) cleaned from modern sediments.
The Maydashy panels have no direct connection with the ancient unconsolidated sediments to enable geological dating of the images’ upper age limit. However, the area where they are located is bounded on both sides by small ravines containing unconsolidated sediments deposited by diluvial–proluvial fans (Fig. 3, observation points [OPs] 3–6). To assume the age of the earliest Maydashy images, descriptions of the sediments and the river terrace (Fig. 3, OPs 1 and 2) are given in Supplementary Information 2 along with their interpretation from the perspective of the geologic history of this part of the Yenisei valley.
Ravine formation is an important phenomenon that makes it possible to date various events in the geologic history of particular regions. Erosion–accumulation processes are well preserved in ancient ravines, which have been repeatedly confirmed by multiple authors working in Eastern Europe (Dotterweich, Reference Dotterweich2005, Reference Dotterweich2008; Superson et al., Reference Superson, Rodzik and Reder2014; Poręba et al., Reference Poręba, Śnieszko, Moska and Mroczek2019). In this study, a similar approach has been applied to establish the lower chronological boundary of the earliest Maydashy petroglyphs. This approach, which can be useful in rock art research for dating carvings from open-air sites, is applicable for the rock art sites of other regions where the ravines network had an effect on their natural context.
Materials and methods
Maydashy rock art site
The Maydashy rock art site is located near the town of Minusinsk in the southern part of the Krasnoyarsk Region. The petroglyphs facing the river were carved in outcrops of red Devonian sandstone of the Maydashy Mountain. The mountain is located on the right bank of the Yenisei River below the confluence of the Minusa River, 10 km NW of Minusinsk and 3 km WNW of Bystraya village (Fig. 1).
The carvings have been known since the nineteenth century (Spasskiy, Reference Spasskiy1857, pp. 146–147, table IV). The site was surveyed and most of the petroglyphs were documented by Adrianov (Reference Adrianov1904). He impressed them on paper to produce very accurate reproductions of the relief images that rival reproductions of modern facsimiles (Miklashevich, Reference Miklashevich2019). He made 53 impressions and 13 photographs, described all the copied surfaces, and wrote the history of the site’s investigation (Adrianov, Reference Adrianov1904, p. 27, Reference Adrianov1906, pp. 58–69). He reported that there had been 363 carved images and a painted one. The petroglyphs were scattered in different parts of the mountain’s foot for a verst (a Russian unit of distance equal to 1.067 km) to portray people in high hats, horsemen, archers, people in boats, cows, sheep, goats, running and lying deer, female deer with calves in the womb, elk, birds, horses, bears, etc. (Adrianov, Reference Adrianov1904, Reference Adrianov1906).
According to Savenkov (Reference Savenkov, Lissner and Sobko1910, p. 99), the largest and most valuable part of the images was destroyed in 1907–1908 while quarrying. The survey conducted by Leontiev (Reference Leontiev1976, p. 131) in the early 1970s found the site to be completely destroyed . At the same time, within the same rocky outcrops downstream, he found several previously unnoticed surfaces with petroglyphs (Fig. 2), which he showed in 2003 to one of the co-authors of this article. Later, he found another surface that had to be cleared from river sediments (Fig. 4).
In the 2010s, a new section of the site was documented (Miklashevich, Reference Miklashevich2015, p. 69, figs. 4.1 and 5.2; Miklashevich et al., Reference Miklashevich, Mukhareva and Bove2016, p. 34, fig. 4). The preserved surfaces are located in two small groups 120 meters from each other, on the lowest tier of rocky outcrops, close to the water’s edge, so that during floods they are sometimes hidden under water and were partially or completely covered in sediments. Most of these images represent the earliest iconographic layer (Fig. 2) (Supplementary Information 1.A and 1. Figs. 1–3).
Geological sections
Materials for this study were obtained during fieldwork in 2023 and included geological descriptions of a series of clearings in riverside outcrops and diluvial–proluvial fans. All observation points (OPs) are indicated on Fig. 3. The layers were described using lithological and stratigraphic methods to indicate lithology, textural and structural features of rocks, geological boundaries, and sediment thickness (Supplementary Information 2).
To establish a reliable correlation between the studied geological sections, geodetic geo-positioning was performed, and a series of altitude profiles were obtained throughout the studied sections (Fig. 5), the only exception being OP 3 located downstream from the rock art panels. The altitude profile was greatly distorted at that point due to obstructive vegetation, and eventually altitude marks at OP 3 had to be identified using global navigation satellite system (GNSS).
Altitude profiles: (c) OP 4; (d) OPs 1, 5, and 6; (e) OP 2.
Radiocarbon and OSL dating
Chemical sample preparation (collagen isolation and purification) and stable isotope analysis were carried out at Cenozoic Geochronology Shared Use Center of Institute of Archaeology and Ethnography of the Serbian Branch of the Russian Academy of Sciences (SB RAS). Sample preparation and isotope analytical procedures were described earlier (Mylnikova et al., Reference Mylnikova, Parkhomchuk, Molodin, Menshanov, Babina, Nenakhov and T.A2024). δ13C and δ15N measurements were performed in a Delta-V-Advantage mass spectrometer, and the reproducibility of the obtained values was ± 0.2 %.
The samples of isolated collagen were sent for graphitization and accelerator mass-spectrometry (AMS) dating to the Laboratory “AMS Golden Valley”, Novosibirsk. Graphite probes were prepared using AGE-3 (Ionplus), AMS dating on MICADAS-28. The sample age was determined in the BATS software by normalizing for the IAEA C-6 ANU sucrose (150.61 ± 0.11 pMC) and subtracting a background value determined for PE STD (BN 268530, Thermo Scientific). The obtained value accounted for the isotopic shift measured for 13C in graphites in MICADAS-28.
Reliability of the isotopic signatures of the collagen extracts was tested using their chemical composition. These values had to be similar to those for the collagen extracted from a fresh bone to be considered reliable for isotopic measurements and radiocarbon dating. The collagen samples whose atomic C/N ratios were lower than 2.9 or higher than 3.6 were regarded as altered or contaminated and excluded from the study (DeNiro, Reference DeNiro1985; Ambrose, Reference Ambrose1990).
The sample 14C age was calculated using the 14C half-life of 5570 years calculated since 1950. The dates were calibrated using the OxCal 4.4 program (https://c14.arch.ox.ac.uk/oxcal/OxCal.html) and the IntCal20 calibration curve (Reimer et al., Reference Reimer, Austin, Bard, Bayliss, Blackwell, Bronk Ramsey and Butzin2020).
Samples for OSL dating were prepared using standard methods (Murray et al., Reference Murray, Arnold, Buylaert, Guérin, Qi, Singhvi, Smedley and Thomsen2021) under dimmed red light in the OSL Laboratory of the All-Russian Geological Research Institute of A. P. Karpinsky in St. Petersburg. The samples were wet-sieved and subsequently treated with HCl (10%) and H2O2 (10%) to remove carbonates and organics. To obtain quartz grains, 90- to 250-μm fractions were density separated using a heavy liquid solution (2.58 g/cm3). The separated fraction (> 2.58 g/cm3) was then etched with concentrated HF (40%) to remove any remaining feldspar and an alpha-irradiated outer layer of quartz grains and then treated with HCl (10%) to remove any fluorides. The etched quartz grains were sieved again to remove the portions reduced to less than 90 μm.
OSL measurements were performed in a Risø TL/OSL-DA-20 C/D reader equipped with a 90Sr beta source. Five-mm-diameter aliquots of quartz grains were mounted on 9.7-mm-diameter stainless steel discs using silicone oil. Signals were stimulated by blue (470 nm) and infrared light-emitting diodes (IR LEDs) (850 nm) and separated from stimulation light using a Hoya U340 filter and a blue filter pack, respectively. Each sample’s purity was evaluated using an OSL-IR depletion test (Duller, Reference Duller2003). Quartz was measured using a single-aliquot regenerative-dose (SAR) protocol (Murray and Wintle, Reference Murray and Wintle2000) employing IR stimulation at 125 ºC for 40 seconds prior to blue stimulation for samples with the OSL-IR depletion ratio deviating from unity by more than 10% (Banerjee et al., Reference Banerjee, Murray, Bøtter-Jensen and Lang2001; Roberts and Wintle, Reference Roberts and Wintle2003). The net OSL signal used for equivalent-dose (De) determination was calculated as a sum of counts during the signal’s initial 0.32 s minus an early background from the following 0.8 s of the decay curve (Cunningham and Wallinga, Reference Cunningham and Wallinga2010). For dose estimation, each dose-response curve (DRC) was derived by fitting the regenerated dose points by a single-saturating exponential function forced through the origin, and the fitting was carefully checked. Instrumental reproducibility errors of 1.5% were incorporated into curve-fitting models.
Eighteen aliquots of each sample were measured. De values for age calculation were extracted using an arithmetical mean model. The De values from individual aliquots were always rejected if they were obtained by DRC extrapolation beyond the largest regenerative dose or the initial Tn signal was less than 3σ higher than a corresponding background. Different rejection criteria (recycling ratio differs from unity by more than 2σ, recuperation >5%, relative standard error (RSE) on Tn signal >25%, RSE on De value >30%) were only applied if this resulted in a change in mean De by more than 1σ or in a decrease in the corresponding standard error. A systematic error of 3% was also added to the measurement errors for possible bias in the calibration of the laboratory beta source (Hansen et al., Reference Hansen, Murray, Buylaert, Yeo and Thomsen2015).
Activity of potassium (40K), thorium (232Th), uranium (238U), and radium (226Ra) were determined in a high-resolution gamma spectrometer with a broad-energy high-purity germanium detector Canberra BE3825 from the bulk sediments surrounding the OSL samples. All samples were dried, ground in a mortar and pestle, and then sealed with wax and stored for at least 20 days (5 222Rn half-live) before counting (Murray et al., Reference Murray, Helsted, Autzen, Jain and Buylaert2018). The measured activities of radioactive elements were converted to annual dose rates using the data from Guérin et al. (Reference Guérin, Mercier and Adamiec2011), which included the attenuation factors relating to sediment grain size, density, and measured water content. The latter included a ± 5% error to incorporate fluctuations through time. The cosmic ray dose rate of each sample was calculated based on the latitude, longitude, altitude, and burial depth of each sample (Prescott and Hutton, Reference Prescott and Hutton1994). Ages are quoted in thousands of years (ka) from the present day. All uncertainties are expressed at 1σ confidence levels and contain both random and systematic components.
Hypsometric mapping
To build the area’s digital elevation model (DEM) and hypsometric map, images were taken using a DJI Mavic Mini 2 unmanned aerial vehicle (UAV) (CMOS 1/2.3-inch sensor, 12 MP, 35 mm lens, image size 4000 × 3000 pixels) sent to an altitude of 50–100 m to provide an average frame overlap of 30%. The obtained images used to produce orthophotomaps were processed in the Agisoft Metashape software (1 cm/pix resolution). The orthophotomaps were georeferenced to the WGS-84 coordinate system to build a DEM in Agisoft Metashape Professional. To increase the accuracy of geodetic geo-positioning, the control points were georeferenced using E-survey E600 GNSS receivers in the regional coordinate system MSK-166.
To clarify the features of relief formation in the study area, a DEM-based hypsometric map was devised in the ArcGIS Pro software (Esri, USA). HillShade, a feature making a landscape’s characteristic properties visible, was added for the DEM using the Tools function > Spatial Analyst Tools > HillShade, with the necessary parameters entered beforehand. This feature gave the reconstruction a 3D effect. The following display parameters were set for the created layer: contrast (19%), brightness (12%), transparency (30%). Hydrographic digitization was performed in the QGIS software package. The map’s design was finalized in ArcGIS Pro.
Results
In the part of the valley between Funtikov Mount and Strelka Ridge is a section of the pebble terrace, whose absolute elevations vary 255–263 m, which corresponds to the water-edge height of 10–17 m (in August, Yenisei’s water level rises to ∼246.5 m). The terrace structure was studied at two points (Fig. 3), the detailed geological descriptions, photos, and drawings of the sections are appended in Supplementary Information 2.
The terrace in the study area has the same structure and approximately equal thickness (12–15 m) of sediments. It is composed of a bed (6–12 m) of well-rounded and washed pebbles of medium and large size deposited in the riverbed by a rather active current. These sediments are overlaid by 3–3.5 m of well-washed medium- to coarse-grained sands and clays deposited by inundations at 15.6 ± 2.3 ka (Table 1). At the top, the terrace is covered by a thin sedimentary cover (0.6–1.5 m) composed of red-gray silts with medium-grained sands that accumulated at about 12.4 ± 1.5 ka (Table 1). The studied terrace sections show a rather good correlation (Fig. 6), with some variations in sediment structure and thickness probably due to differences in facies contexts during their formation.
Correlation scheme for the geological sections studied. 1 = Devonian sandstone bedrock; 2 = pebbles; 3 = sands; 4 = loose silt; 5 = rubble–boulder; 6 = soil and paleosol; 7 = OSL date (ka); 8 = 14C date (cal yr BP); 9 = bone remains; 10 = shell remains; 11 = modern household waste; 12 = scree.
OSL dates of sediments of the first floodplain terrace.
1 Water content was assumed to be 75% of the measured water saturation in the samples.
2 No natural or regenerated OSL signal from quartz was detected.
In addition to the alluvial sediments of the terrace, there are ravines cutting through the terrace and a rock massif where the Maydashy rock art site is located. These ravines reveal the strata of diluvial–proluvial sediments studied in a series of sections (Fig. 3).
A similar structure of diluvial–proluvial sediments was observed at all the studied OPs (Supplemental Information 2). The lower parts of geological sections are represented by poorly rounded sandstone rubble of irregular sizes (up to 40 cm) with loose, light-gray silts and fine-grained sandy fill. These sediments were overlain by silts with rock-slack lenses or interlayers. The upper third of most sections included fossil soil interbeds. Although not properly dated, these paleosols could be attributed to the Holocene Climatic Optimum, which was confirmed by 14C dates obtained for the overlying slope deposits in OP 4 (Table 2).
Radiocarbon dates of the bone remains from the diluvial–proluvial deposits.
The earliest ravine sedimentation age was recorded at OP 3 and was dated as Early Holocene based on the 14C date of 10,246–9964 cal yr BP determined for a bone fragment of a large mammal (Table 2). There was only one section (OP 5), where we observed these sediments cutting directly into the terrace sediments. Nevertheless, the correlation study of the geological sections refined using altitude profiles made it possible to obtain a reliable match between the OPs (Fig. 6). Ravine sediments dipping towards the river have been discovered in all sections, which confirms their diluvial–proluvial genesis and the presence of a local base level during the sedimentation.
Therefore, studying the terrace and ravine deposits in the vicinity of the Maydashy rock art site enables one to trace the geological history of the area and estimate the time when the images were created. The edge of the eroded river terrace (OP 3) and the upper part of the sediments of the studied ravines are located at approximately the same hypsometric level (∼12–15 m from the river level) at 254–255 meters absolute height. At Bystraya village, the terrace height reaches 263 m, which suggests the existence of a single terrace level in the area from the western end of the modern village to the mouth of the ravine near the rock art site (OP 3). The preserved surfaces with petroglyphs are located significantly below the marked level (∼248 m), meaning they could not have been created earlier than the time that the terrace level was eroded (Fig. 7).
Therefore, studying the terrace and ravine deposits in the vicinity of the Maydashy rock art site enables one to trace the geological history of the area and estimate the time when the images were created. The edge of the eroded river terrace (OP 3) and the upper part of the sediments of the studied ravines are located at approximately the same hypsometric level (∼12–15 m from the river level) at 254–255 meters absolute height. At Bystraya village, the terrace height reaches 263 m, which suggests the existence of a single terrace level in the area from the western end of the modern village to the mouth of the ravine near the rock art site (OP 3). The preserved surfaces with petroglyphs are located significantly below the marked level (∼248 m), meaning they could not have been created earlier than the time that the terrace level was eroded (Fig. 7).
Discussion
Geological history of the area and the Maydashy site age
In the part of the valley between Funtikov Mount and Strelka Ridge is a section of the pebble terrace with elevation as high as 255–263 m that corresponds to a height of 10–17 m above present water level (in August, Yenisei’s water level rises to ∼246.5 m). The terrace has a uniform structure and approximately equal sediment thickness of 12–15 m at all locations (Fig. 3). It is composed of a bed (6–12 m) of well-rounded and washed medium to large pebbles, deposited by a fairly active current (Fig. 5). These sediments are covered by a layer (3–3.5 m) of well-washed, medium- to coarse-grained alluvial sands and clays. On top is a shallow sedimentary cover (0.6–1.5 m) composed of red-gray silts with medium-grained sands, sometimes interbedded with paleosols.
The structure and composition of the studied terrace section match those of the known alluvial sections of the river’s first floodplain terrace. According to Puminov and Buzulutskov (Reference Puminov, Buzulutskov and Saks1968) the average height of the first floodplain terrace in the Minusinsk Depression is 5–6 m, with the total thickness of sediments up to 25 m, especially when it comes to pebble sequences. The structure of the second floodplain terrace of the Yenisei River shows similar patterns but with slightly higher elevations (14–18 m) and is overlain in the upper part by loess-like loams (Drozdov et al., Reference Drozdov, Chekha and Haesaerts2005), which we did not observe in the studied terrace. The level of the high floodplain composed of sands, loams, and sandy loams in the Minusinsk Depression corresponds to heights of 2–3 m (Puminov and Buzulutskov, Reference Puminov, Buzulutskov and Saks1968).
For the Yenisei Valley, the time during which the alluvial part of the first floodplain terrace accumulated is defined as the second half of the Sartanian stadial, which corresponds to MIS 2. According to numerous 14C dates, the terrace formed between 19.5 ka and 12.7 ka or within a somewhat wider interval (Zubakov, Reference Zubakov and Gromov1965; Kind, Reference Kind1967).
To specify the terrace’s age, we collected three samples for OSL dating (OP 1) and obtained dates for two of them. The date was 15.6 ± 2.3 ka for the river-bed sands from the middle part of the terrace and 12.4 ± 1.5 ka for the silts and floodplain sands in the upper part of the section. Considering the confidence intervals for both dates, we can confidently say that formation of the terrace took place in the second half of the Sartanian stage (MIS 2), which agrees well with the interpretations proposed by other authors. It means that accumulation of the alluvium of the first floodplain terrace correlates with the period of sea level rise at the end of the Late Pleistocene that terminated at the final stage of Allerød from ca. 14 to ca. 12.5 ka (Lambeck et al., Reference Lambeck, Rouby, Purcell, Sun and Sambridge2014).
The terrace itself is covered in continuous diluvial deposits with numerous deep ravines graded to the present-day river level. The ravines are filled with diluvial–proluvial sediments that were studied in a series of sections (Fig. 3, OPs 3–6). As for those in the study area, they have a rather monotonous structure: diluvial loams blanket the terrace surface; along the river are deep washouts in the terrace body filled with strata of porous, red-brown silts and interlayers of rock slack and sandstone clasts of different sizes from 1–2 cm to 50–70 cm (Fig. 6). At the mouth of the ravine, located immediately downstream of the rock art site, a fragment of a large mammal bone was found in the rock slack (OP 3). A 14C date of 9013 ± 39 (GV-04545) obtained for this fragment corresponds to an age of 10,200 ± 60 cal yr BP. This date is interpreted as indicating the time when the first floodplain terrace was eroded and the floodplain formed.
Erosion of the first floodplain terrace might be related to a significant decrease in the rate of global sea level rise (Lambeck et al., Reference Lambeck, Rouby, Purcell, Sun and Sambridge2014) during the Late Dryas stadial (ca. 12.5–11.5 ka), which led to formation of a new erosion base level and transformation of the river network. The short-term pause in sea level rise contributed to development of the ravine network in the area around the Maydashy rock art site. The ravines were subsequently filled with proluvial sediments that are associated with the warming and moistening during transition from the Preboreal to the Boreal phase of the Holocene at about 11–10 ka, which is well documented in many sequences in Siberia (Velichko et al., Reference Velichko, Catto, Drenova, Klimanov, Kremenetski and Nechaev2002), Baikal (Bobrov et al., Reference Bobrov, Maltsev, Krivonogov, Preis, Klimin and Leonova2023), and the Far East of Russia (Razjigaeva et al., Reference Razjigaeva, Grebennikova, Ganzey, Chakov, Klimin, Mokhova and Zakharchenko2021) and confirmed by our 14C date (10.2 ka). The sharp increase in water flow at this time is evinced by a significant increase in the intensity of chemical weathering and the productivity of terrestrial ecosystems in the vicinity of Lake Baikal (Shimaraev and Mizandrontsev, Reference Shimaraev and Mizandrontsev2004), which correlates well with the increase in the rate of global sea level rise at the end of the Dryas cooling (Lambeck et al., Reference Lambeck, Rouby, Purcell, Sun and Sambridge2014).
Upward along the section, these rock-slack strata are replaced by interlayers of sandy silts embedded with single slack fragments or small lenses. Sometimes 1–2 humic interlayers can be found in the upper part of the section (Fig. 7).
3D model of the Maydashy rock art site showing the assumed sedimentation level of the first floodplain terrace (dotted line). The continuous line depicts the lower part of the clearance at OP 1.
We think the paleosol layers correlate with the Holocene climatic optimum soils formed between 7.9 ka and 4.5 ka (Kind, Reference Kind1967). They are the most frequent type of the region’s Holocene soils that are often represented by scattered interlayers (Malikov et al., Reference Malikov, Volvakh, Volvakh, Saveleva, Ovchinnikov and Malikova2023). Above these humus-rich interlayers are layers of fine-grained sand (see OP 4), which probably correspond to similar sand interlayers in layer 2 of the Chernousov Log section (Malikov et al., Reference Malikov, Volvakh, Volvakh, Saveleva, Ovchinnikov and Malikova2023). The time when these interlayers formed may correlate with the wetting phase of Subboreal to Subatlantic transition that occurred in the region during the first millennium B.P. (ca. 2.8 ka) (van Geel et al., Reference van Geel, Bokovenko, Burova, Chugunov, Dergachev, Dirksen and Kulkova2004). The dates obtained for the bone remains from these deposits confirm this correlation: 2205 ± 32 cal yr BP (GV-04547) and 2229 ± 32 cal yr BP (GV-04548) (OP 4, Fig. 6, Table 2), which correspond to a time interval of 2340–2120 cal yr BP. Thus, the geological history of the Yenisei River valley around the Maydashy rock art site seems to have developed in a sequence of stages (Fig. 8).
Ravine network development in the vicinity of the Maydashy rock art site.
Stage 1
At the end of the Late Pleistocene, in the second half of the Sartanian stage (MIS 2), the first floodplain terrace of channel pebbles and sands began to form, reaching 12 m in height. According to many authors (Zubakov, Reference Zubakov and Gromov1965; Kind, Reference Kind1967; and others), formation ends at the Pleistocene–Holocene transition, which agrees well with the OSL dates obtained for the terrace in the studied section (OP 1). Considering the hypsometric level of the terrace, the ravines cutting through it, and the location of the Maydashy rock art site at that time, the terrace sediments could have covered the rock surface. The area may have been a rocky base for the terrace and remained below the water’s edge during this period. One way or the other, the images could not have been carved during this period (Fig. 8, stage 1).
Stage 2
At the beginning of the Boreal stage of the Holocene, when the high floodplain of the Yenisei River began to form at about 10–7 ka (Kind, Reference Kind1967), the river cut into the first floodplain terrace to form an erosional base at the site. This produced numerous slack-filled washouts in the preserved section of the pebble terrace, their oldest age being 10.2 ka. During this time, the ancient inhabitants of the Minusinsk Depression might have had an opportunity to carve images on the rock surfaces of the mountain foot (Fig. 8, stage 2).
Stage 3
The ravines formed during stage 2 were later infilled, so the soil surface levelled to that of the terrace. The reduced sedimentation intensity was reflected in the predominance of silty and fine- to medium-grained sands with poorly developed interbeds of fossil soils (Fig. 8, Stage 3). We think these soils formed during the Holocene climatic optimum (7.9–4.5 ka), which means that the erosion stage of the first floodplain terrace and ravines formation (stage 2) occurred earlier than this time.
Stages 4–6
Erosion increased again after the end of the soil formation period (stage 3) at the onset of the Holocene Subatlantic period. At this time, new ravines developed along the existing runoff troughs. This process was repeated several times (Fig. 8, stages 4–6).
The geomorphologic position and obtained geological data allow us to conclude that during the Pleistocene the surfaces of the foot of Maydashy Mountain were covered in sediments, which made it impossible to carve petroglyphs on them. At the very beginning of the Holocene, before the onset of the optimum, the Yenisei River eroded the terrace and exposed the rock surface, which gives us a foundation to estimate the earliest time when the images could have been created. Based on the assumptions given above, their age is less than 11 ka.
One could assume that the images were created prior to the first phase of sedimentation before the second half of the Sartanian stage. However, in that case the later extreme processes (terrace formation with well-rounded pebbles and its subsequent erosion) likely would have destroyed or severely damaged the rock carvings, especially the painted one reported by Adrianov (Reference Adrianov1904, Reference Adrianov1906). Nevertheless, the images from Maydashy are preserved relatively well (Fig. 2), which prevents us from attributing them to the Pleistocene. Unfortunately, the absence of the sediments overlying the Minusinsk Style images does not allow us to estimate the upper age limit of the earliest imagery using the same methods.
Zoological context of dating of the Maydashy rock art
According to Adrianov (Reference Adrianov1904, Reference Adrianov1906), the Maydashy rock art initially featured images of animals, such as cows, sheep, goats, running and lying deer, female deer with calves in the womb, elk, birds, horses, bears, etc. Only a few of those have survived to the present day, including the ones stylistically attributed to the earliest phase of the regional rock art.
Among the surviving images of the Maydashy rock art was the image of a large bull (Fig. 2) morphologically similar to the genus Bos Linnaeus, represented in the wild by Bos primigenius Bojanus. Taking a close look at the images present in the earliest rock art of the Minusinsk Depression, one can identify bulls, various species of deer, goats, sheep, horses, wild boars, and bears.
All the listed species have been known in the region from Pleistocene to Late Holocene and the present, except for two: the aurochs and the wild boar. Despite the fact that the fossil remains of Bos primigenius have been found in many regions of Siberia, including alongside the Pleistocene species, there has never been a single find with a Pleistocene 14C date. In recent years, there have been five published dates of B. primigenius remains in Southern Siberia, three of those from the Prealtai plain, one from Tyva, and one from the Minusinsk Depression (Lavrov and Zabelin, Reference Lavrov and Zabelin2007; Vasiliev et al., Reference Vasiliev, Parkhomchuk, Serednyov, Milutin, Kuzmin, Kalinkin and Rastigeev2018, Reference Vasiliev, Parkhomchuk, Serednyov, Milutin, Rastigeev and Parkhomchuk2020). These dating efforts produced a maximum age scatter of 12.7–6.2 ka.
Occasional finds of wild boar remains alongside Pleistocene mammals have been recorded in Siberia in the caves of Altai (Galkin and Ovodov, Reference Galkina, Ovodov and Yudin1975), Kurtun-1 cave near Lake Baikal (Filippov et al., Reference Filippov, Grebnev, Henzyhenova, Erbaeva, Martynovich and Popov2012), Chumysh River on the Prealtai plain (Vasiliev et al., Reference Vasiliev, Serednyov and Milutin2019), and the Novoselovo locality in the Minusinsk Depression (Malikov, Reference Malikov2016). However, the ages have not been confirmed in any of those finds. Dating of wild boar from the Chumysh River provided an age of about 770 cal yr BP (Vasiliev et al., Reference Vasiliev, Parkhomchuk, Serednyov, Milutin, Rastigeev and Parkhomchuk2020), while the wild boar from the Minusinsk Depression has a 14C age of 175 ± 30 yr BP (Poz-78196), which does not exceed 290 cal yr BP. Thus, the paleontological data do not make it possible to attribute the petroglyphs of the Maydashy rock art to a time before the Early Holocene.
Archaeological context and the upper boundary of the Minusinsk Style
The archaeological context of the earliest rock art known in the region has been understudied so far; however, the general archaeological data allow us to suggest that the earliest rock art age ranges from the Upper Paleolithic (38–11.5 ka) to the Afanasiev culture (until 4.5 ka), which is quite a broad period. Moreover, the oldest evidence of human presence in the region dates back to the Early Upper Paleolithic and even earlier. The results of geological context dating of the Maydashy rock panels open a new path for the chronological delimitation of the entire body of the earliest rock art in the Minusinsk Depression.
Summarizing the data obtained, we judge that the Maydashy rock surfaces have been exposed since 11 ka, and the Minusinsk Style carvings on the site could not have been created earlier, which is confirmed by their relatively good preservation. This shows that the Minusinsk Style, known as the earliest in the region, does not belong to the Upper Paleolithic; this is also confirmed by the species of the depicted animals (which belong only to the Holocene).
Even if it is impossible to estimate the upper age boundary of the earliest imagery using the same methods, the statement still holds that the Okunev rock art tradition, which is well dated by archaeological materials (from 4.5 ka onward), draws a rather sharp boundary between the earliest phase in the local rock art and the further stylistics of the Bronze Age. However, the assumption that the earliest images belong to the Afanasiev culture remains up for debate.
Conclusions
Unlike other earliest rock art sites, Maydashy is located in the part of the Yenisei River valley that has not undergone significant anthropogenic changes. At the same time, the geological context of the valley and the presence of a ravine network has enabled us to reconstruct the main stages of its geological history that are essential for determining the age of the rock art in question.
It has been found that the area around the site was covered in alluvial sediments of the first floodplain terrace formed in the second half of MIS 2, which has been confirmed by OSL dating. At that time, the foot of Maydashy Mountain, which was covered in sediments, could not have been used for image carving. Later, the accumulated sediments were extensively eroded, which led to ravine formation, a process that first started 11–10 ka and then was repeated numerous times. These assumptions have made it possible to estimate the potential lower age of the images, which is less than 11 ka.
Study of the geological context of the rock art site has allowed us to obtain the first relative dates for the earliest rock carvings in the Minusinsk Depression. The obtained results are in good agreement with the historical and cultural context of the Minusinsk Style, which presents exclusively zoomorphic images made in an archaic and naturalistic manner. At the same time, there are no images of fossil species, meaning the Minusinsk Style cannot be attributed to the Pleistocene, which indirectly confirms our findings.
The absence of the sediments overlying the images does not allow us to estimate the upper age limit of the earliest petroglyphs. However, well-dated subsequent stylistic assemblages (the Okunev Style dated back to 4.5 ka), and the lower position of the earliest rock art in multilayered compositions have made it possible to establish the chronological boundaries of the earliest petroglyphs between 11 ka and 4.5 ka. Thus, the results of our research provide fundamental evidence for the Holocene age of the rock images that are considered nowadays as the earliest in the region. Consequently, the hypothesis of a Paleolithic age of the Minusinsk Style in the rock art of Southern Siberia can be considered disproved.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/qua.2025.5.
Acknowledgments
This work was supported by the Russian Science Foundation [grant number 22-18-00070], https://rscf.ru/project/22-18-00070/ The authors thank Yan V. Reznichenko for the translation and proofreading. Special thanks to Dr. Krzysztof Stefaniak (Department of Palaeozoology, Faculty of Biological Sciences, University of Wrocław) for dating the boar remains from the Novoselovo site in 2016.
Data availability
All data used in this article are available in this paper.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Author contributions
Dmitry G. Malikov: fieldwork, collecting and processing geological material, summarizing the results and writing the manuscript. Lydia V. Zotkina: research leadership, writing the manuscript (archaeological part), corresponding author. Roman V. Davydov: 3D-model of the valley creation. Ekaterina L. Malikova: hypsometric map creation. Elena A. Miklashevich: writing the manuscript (history of research). Aleksey V. Petrozhitskiy: AMS measurements. Olesya V. Ershova: sample preparation for AMS. Anna Cherezova: OSL measurements.
