Exploring Hypotheses on Early Holocene Caspian Seafaring Through Personal Ornaments: A Study of Changing Styles and Symbols in Western Central Asia

1.1 Kaylu Rockshelter

Previous investigations into the Final Pleistocene/Early Holocene transition in Western Turkmenistan were conducted during the mid-20th century by A.P. Okladnikov, who discovered dozens of sites in the Krasnovodsk peninsula in 1947 (Okladnikov, 1949, 1951). The Kaylu rockshelter (Figure 1) is important because it preserves stratified deposits spanning the late Mesolithic and Neolithic. Since its discovery, the site has not been comprehensively reassessed or directly dated (Shnaider et al., 2018, 2021).

The Kaylu rockshelter is located at the eastern end of Kubadag, 20 km east of Krasnovodsk, at an altitude of 23 m above the current Caspian Sea level. During excavations conducted in 1952, Okladnikov recorded 4 m of stratified deposits. Seven cultural layers were identified, with the first six upper layers attributed to the Neolithic. The absence of pottery remains in layer 7 led Okladnikov to attribute this layer to the Mesolithic (Okladnikov, 1953, 1966). Kaylu was rediscovered during fieldwork in 2018, and surface artifact assemblages were subjected to a preliminary study (Shnaider et al., 2018). Until now, excavated lithic, pottery, and faunal remains had been poorly described. Pottery is present in layers 1–6, but more frequent in layers 3 and 4 (Table 1). Their analysis is still ongoing. No faunal data are available for these layers.

No precise spatial field records were taken during Okladnikov’s excavation. The absence of systematic sediment sieving has most likely introduced a bias toward larger artifacts. The stone tool assemblages in layers 1–6 comprise debris, flakes, flat cores for flake production, volumetric cores for bladelet production, and retouched flake tools (Table 2). The absence of bladelets may result from the transport of finished tools away from the site, or from the poor field methods.

The lithic assemblage from layer 7 comprises 219 artifacts. Bladelet cores are semi-cylindrical with one striking platform (Table 2 and Figure 1). Core preparation includes semi-tablets, corner and lateral flakes, and front core-trimming. The debitage includes 79 flakes, 29 blades, and 45 bladelets. The retouched tool assemblage totals 17 artifacts, including end-scrapers, notches, splintered pieces, awls, and backed lunates.

Two burials were discovered at the site, just a few meters from the entrance of the rockshelter (Figure 1). The skeletons were not deeply buried and showed many alterations, including a high level of breakage, cracks, decalcification, and concretion. Some remains were cemented in a very hard breccia, which was probably responsible for the high level of material breakage during excavation. The individual from burial 1 had postcranial bones with fused epiphyses (Figure S2). Wear patterns on the two upper premolars suggest that it was a young adult. The skeleton from burial 2 was almost exclusively represented by small upper body bone fragments, one premolar, and one lower incisor (Figure S2). The wear pattern of the incisor suggests that this individual was an adult. Sex determination was performed using peptides in the tooth enamel, following the method of Stewart et al. (2017) (Supplementary Text A). The young adult was identified as a male and the adult as a female (Tables S1 and S2). No photographs or drawings showing the position of the bodies were kept with the collection, but descriptions of the burials in Okladnikov’s field report indicate that the skeletons were supine, with the head orientated north-west and the feet south-east, and their two arms were joined on the pelvic area. Both the bodies had been covered with ochre and were accompanied by dozens of discoid shell beads. Some lithic tools, similar to the material from layer 7 of the rockshelter, were found during the burials’ excavation (Okladnikov, 1951). However, the lack of spatial information regarding the association of the lithic material and the human remains does not secure their strict contemporaneity, and the presence of lithic material within the filling sediment of the graves cannot be excluded.

2.1 Taxonomic Identification of Shells

Taxonomic identification of the shells involved two steps: the characterization of the shell’s general shape for class determination (e.g. Scaphopoda, Gastropoda, Bivalvia) followed by an examination of the number of whorls (shape of the aperture, lip, ventral and dorsal sides) and ornamentation to determine the genus or species (Harasewych & Moretzsohn, 2010; Poppe & Goto, 1993). The nomenclature employed adopted classifications from the Caspian Sea Biodiversity Project (Checklist for Caspian Sea mollusks by P. Kijashko https://www.zin.ru/projects/caspdiv/caspian_molluscs.html, search performed on April 2018) and the WoRMS database (World Register of Marine Species http://www.marinespecies.org/index.php, search performed April 2018). A consideration of the configuration and distribution of the regional biotopes particular to each shell species and fossil outcrop revealed their probable procurement location(s) (Bar-Yosef Mayer, Gümüs, & Islamoglu, 2010; Rasilla et al., 2020; Rigaud, O’Hara, Charles, Man-Estier, & Paillet, 2022; Rigaud, 2013; Vanhaeren, 2002).

2.2 Functional and Morphometric Analyses of the Shells

A Motic SMZ-168 microscope equipped with a Canon 1100 D digital camera was used to document the surface modifications of each shell. The surfaces show microscopic alterations testifying to events that occurred either during the life of the mollusk or post mortem. In the case of shells collected and/or modified by prehistoric groups, microscopic analyses provided information relevant to the environment in which the shells were collected, as well as subsequent taphonomic and anthropogenic modifications (d’Errico, Henshilwood, Vanhaeren, & Van Niekerk, 2005; Dupont, 2006; Taborin, 1998; Vanhaeren, d’Errico, van Niekerk, Henshilwood, & Erasmus, 2013). The presence, location, and degree of natural modifications impeding microscopic analysis (calcite deposits, cracks) were recorded for each specimen, along with the degree of preservation of the shell’s original shape and features. Anthropogenic modifications such as fractures, use-wear, or those produced by suspension (e.g. perforations, residues, incisions) were also systematically recorded.

Morphometric variables (shell maximum and minimum diameter, maximum and minimum diameter of the perforation and thickness of the shell) were recorded to characterize size selection and shaping standardization. The K-means Cluster Analysis (KCA) was used to explore the size variation of the beads and to partition data into relatively homogeneous groups. Because the primary role of personal ornamentation is to be seen, to transmit symbolic messages (Erikson, 1969; Meisch, 1998), the most important characteristics of the beads are those related to their visual impact (Beck, 1926; Kassam, 1988). We therefore chose the number of groups used for the K-mean clustering based on macroscopic observations performed on the full bead assemblage. PAST software (Hammer, Harper, & Ryan, 2001) was used for the analysis, performed with K = 3 and 1,000 iterations.

2.3 Residue Analysis

The location, color, and microscopic features of the shell bead residues were systematically recorded. Their mineralogical composition was subsequently explored with Raman Spectroscopy and SEM microprobe analyses. Raman spectra were obtained on six samples with a SENTERRA confocal microspectrometer (Bruker Optics, Ettlingen, Germany) equipped with a 532 nm excitation line. The spectra, acquired with a 5 mW laser power, 50 co-additions of 5 s excitation, were compared with the RRUFF™ database (Lafuente et al., 2015). Four samples were characterized using a JEOL JMS 6460 LV scanning electron microscope, operating at an accelerated voltage of 20 kV, coupled with an Oxford X-Max EDS Silicon Drift Detector in a 20 mm² active area and a 125 eV resolution for the Mn Kα emission line (Le Bourdonnec et al., 2010; Mulazzani et al., 2010).

2.4 3D Characterization of the Beads

Beads embedded in carbonate concretions were 3D scanned using a General Electric (GE) Vtome x|s microtomograph. For each scan (length = 90 mn), the cubic voxel size and beam parameters were adapted to fit the sample characteristics (Table 3). A copper filter of 0.1 mm was used and 2,550 projections were taken at 360°. Reconstruction of the volumes was performed using the Datos | Acq software, GE. 3D visualizations of the beads were then performed within the Avizo 9.1 (FEI) workspace.

2.5 Radiocarbon Dating

One premolar from burial 2 at Kaylu was directly dated by enamel carbonate radiocarbon (14C) dating. Tooth enamel samples were leached under vacuum with acetic acid in the Environmental Isotope Paleobiogeochemistry Laboratory at the University of Illinois to remove exogenous soil carbonate, using a protocol based on the Vacuum Milling technique of Krueger (1991). At the Keck Carbon Cycle AMS (KCCAMS) laboratory, University of California, Irvine, the tooth enamel was further purified with 0.1 M HCl to remove 10% of the sample weight, prior to hydrolysis with 85% phosphoric acid to generate CO₂ for conversion to AMS targets. Sample preparation backgrounds were subtracted, based on measurements of 14C-free calcite. All results have been corrected for isotopic fractionation according to the conventions of Stuiver and Polach (1977), with δ13C values measured on prepared graphite using the UCIAMS spectrometer at the KCCAMS laboratory, University of California, Irvine. However, as these can differ from the δ13C of the original material they are not reported. Radiocarbon concentrations are given as fractions of the modern standard, D14C, and radiocarbon years before present (RCYBP), following the conventions of Stuiver and Polach (1977) (Table 4).

3.1 Identification and Sourcing of the Shells

Forty-six shells were discovered within the stratigraphic layers of the rockshelter, including three bivalve species and two gastropod species (Table 5, Figure 2). Six unmodified shells, attributed to four different species, were also present in the material from the burials (Table 5). Four shells were not clearly attributed to any of the site’s sectors.

Figure 2

Material discovered in the rockshelter (a) and the two burials (b and c): fragments of discoid beads (b) and intact discoid beads (c).

Personal ornaments discovered in the burials included 189 discoid shell beads. Parallel and straight large ribs – naturally present on the dorsal side of many bivalve species – are present on the convex side of all these beads. Some also present shorter ribs on the concave side corresponding to the natural ribs present in the inner part of bivalves.

The evolution of the cardiid bivalves of the Ponto-Caspian region remains unclear because of stratigraphic gaps, unsettled taxonomy, and the unclear status of bivalve groups. Due to these issues, the shell beads have not been taxonomically attributed, but the thickness of the shells and the morphology of the ribs suggest the genus Didacna belonging to the Cardiidae family.

3.2 Taphonomic and Functional Analyses of the Shells

The bivalves from the rockshelter (Unionidae, Didacna sp., and Dreissena polymorpha) do not show any anthropic modification, though a dorsal perforation is observable on one Theodoxus pallasi. The perforation is irregular and the edge of the hole shows fresh breakages not covered with patina, indicating the perforation is post-depositional, recent, and most likely occurred during excavation. The six complete shells from the burial do not show any anthropic modifications.

Several sets of discoid beads were cemented in their original association by calcite concretions. The position of the beads shows that at least some of the ornaments were aligned and suspended by a string through the central perforation at the time of their deposit (Figure 3).

Figure 3

Discoid shell beads maintained in their original configuration by calcite concretions (a–d). µCT-based three-dimensional visualization of the beads: (e and f) 3D-CT of cemented beads (b); (g) 3D-CT of cemented beads (c); (h and i) 3D-CT of cemented beads (d).

Fragmentation prevented the measurement of metric variables on many bead specimens; however, 64 well-preserved beads and their perforations were measured (thickness, maximum and minimum diameter) and showed patterns in size variation, with three main size classes being clearly visible to the naked eye (Figure 2). The KCA performed on the minimum and maximum diameter confirms the presence of three statistically different groups of shell bead diameters (Mann–Whitney U test p < 0.01). A scatterplot of the minimum and maximum diameters clearly differentiates the three groups identified by KCA with the larger beads clearly separated from the small and mid-sized beads (Figure 4).

Figure 4

Scatterplot of the maximum and minimum diameters of the discoid shell beads. Confidence ellipse 95%. Each ellipse corresponds to one of the groups identified by the K-mean cluster analysis.

The technological analysis was limited by the multiple alterations present on the shells’ surface. Several post-depositional processes including dissolution, breakages, and concretions have contributed to the destruction of anthropic modifications.

Natural ribs are visible on both sides of the large discoid beads (Figure 5). Ribs on the concave side are short, do not extend across the full surface of the beads, and correspond to natural ribs located at the shells’ ventral margin. The presence of short ventral ribs on the large archeological specimens indicates that the beads were shaped from a portion of the shell located near the ventral margin. Natural ribs still visible on the discoid beads suggest that the surface of the large beads was not regularized by grinding (Figure 6). The absence of short ribs on one side of the mid-sized and small beads may be due to the removal of the ventral margin during manufacture. No other clear technological features were observed on the mid-sized or large beads.

Figure 5

View of the natural ribs on the surface of the bivalves. (a) Dorsal and ventral side of an archeological shell bead and (b) location of the anatomical portion of the bivalve used for the discoid bead manufacture in red.

Figure 6

Modifications on the ventral and dorsal side of the mid-sized (a and b) and large discoid beads (c–f). Taphonomic alterations: exfoliation (a, b, d, e, and f), concretion (c, g, and h); anthropic modifications: red residue (b–f).

Many small and mid-sized beads were affected by exfoliation, resulting in a laminated breakage pattern (Figures 2b and 3). The thickness of the foliated beads was not recorded. The measurements recorded on the unbroken beads show that the large beads have a mean thickness of 3.24 mm (n = 15), mid-sized beads a mean thickness of 2.37 mm (n = 17), and small beads a mean thickness of 2.36 mm (n = 53). Thickness differences between the large and the mid-sized/small beads and an absence of grinding evidence on the ventral and distal sides of all the shell beads suggest the use of two different raw shell materials.

Circular striations on the perforations of the small beads indicate that the shells were perforated by the rotation of a lithic point (Figure 7c–g). The perforations are biconic, suggesting that the rotation was performed from both sides of the bivalves (Figure 7i and j). The edge between the two perforation cones is closer to the dorsal side of the valve, evidencing that the perforation was processed first from the ventral side and the opening was subsequently enlarged from the dorsal side. Considering the natural convexity of the valves, this process may have limited accidental breakages by applying rudimentary pressure on the dorsal convex side during the final stage of perforation manufacture. Use-wear analysis was not conclusive due to the poor preservation of the material (Figures 6a–h and 7h).

Figure 7

Taphonomic and anthropic modifications observed on the small discoid beads. (a, b, d, e, and g) Red residue in the perforations embedded in calcite concretions; (c–g) circular striations resulting from perforation by the rotation of a lithic point; (h) discoid bead altered by dissolution process; and (i and j) sharp edge on the biconic perforation showing that the circular motions were made from both sides of the shell.

3.3 Residue Analysis

Microscopic analysis identifies a red residue concentration within the perforation of the small and mid-sized shell beads and at the surface of both sides of the large beads (Figures 6b–f and 7a, b, d, e, and g). Raman spectroscopy and SEM analyses were performed on specimens from each bead size category (Figures 8 and 9, Figures S5 and S6). Analysis of the red compound revealed fragments and fine particles of iron oxide, explaining the residue’s red coloration. Hematite (Fe₂O₃) was present as finely grained powder; no particular variation in structure or orientation was visible in the samples, besides the presence of rare, shiny and larger fragments in one sample (Figure 9, Figures S5 and S6). With regards the ochre particles, analyses identified calcite (CaCO₃), gypsum (CaSO₄·2H₂O), quartz (SiO₂), and halite (NaCl). The proximity of the site to the Caspian Sea, and the local dry sandy environment, suggests that quartz, gypsum, and halite were part of the surrounding sediment. Calcite corresponds to the shells’ composition; therefore, its presence within the residue itself may also result from the post-depositional carbonate formation that contributed to cementing some of the beads together and to preserving and hardening the residue (Figure 3).

Figure 8

Raman spectra obtained for the red residue present on the beads: (a) small bead from burial 1, (b) mid-sized bead from burial 2, (c) mid-sized bead from burial 1, and (d) large bead from burial 1. C′ labeled peak corresponds to calcite/aragonite.

Figure 9

SEM-EDS analysis performed on one large bead. AI: area of interest; S: spectrum. AI2-S2: ochre (iron-rich and aluminosilicate-rich material); AI2-S1: halite; AI3-S1: calcite; AI3-S2: ochre and calcite concretion and/or aragonite from the shell.

3.4 New Chronological Data

Recent data obtained on the Late Quaternary Caspian Sea Khvalynian transgression (Kurbanov et al., 2021; Yanina et al., 2017) show that between 13 and 15 ky ago the Caspian Sea level was 76 m above the modern sea level and that part of the Kubadag cliff, including the Kaylu site, was underwater (Kurbanov, Svitoch, & Yanina, 2014). The Kaylu rockshelter was presumably formed during one particular sea-level stabilization phase attributed to the Late Khvalynian period (Kumskaya stage). The rockshelter then became accessible for settlement after the beginning of the Mangyshlak regression, about 9–10 ka BP (Svitoch, 2012). These geomorphological data provide an approximate terminus ante quem for possible early human occupations of the site.

The results of the radiocarbon dating of burial 2 indicates a median age of 6300 ± 27 cal BP (Table 6).

4.1 Acquisition Selection, Transformation, and Use of the Shells

The shell species identified in the rockshelter and the burials are naturally present in the Ponto-Caspian area. The Didacna sp. bivalves are an endemic marine species currently present in the Caspian Sea and are widely distributed in local fossil outcrops (Albrecht, von Rintelen, Sereda, & Riedel, 2014). They are classically used as a paleontological marker of the different Caspian Sea transgressive and regressive episodes (Yanina, 2005). The D. polymorpha bivalve is present in the Caspian Sea and also in freshwater rivers and brackish water estuaries (Son, 2007). Unionidae shells are freshwater bivalves (Graf, 2007). Turricaspia sp. gastropods are mostly marine shells, but some rare species can be found in brackish water (Vinarskij, & Kantor, 2016). The T. pallasi gastropod is a freshwater species present in the region and is also observable in the brackish environment near Caspian estuaries (Anistratenko, Zettler, & Anistratenko, 2017; Neubauer, van de Velde, Yanina, & Wesselingh, 2018). The location, about 500 m from today’s Caspian shore, and less than 20 m during the transgression some 13 ky ago, suggests that the raw shell materials used for bead manufacture were collected locally.

The shells present in the archeological layers of the rockshelter do not show any anthropic modifications, are diversified, and not abundant in each layer. The field methods used during the Okladnikov excavation did not include sieving and have therefore affected the collection of small items. Regional comparisons show that a perforated umbo fragment, attributed to Didacna sp., has been documented as a bead in one Epipaleolithic layer at Altappeh (Iran), located in the southern Caspian Sea (Manca, Mashkour, Shidrang, Averbouh, & Biglari, 2018). Perforated T. pallasi were also present in Caspian Mesolithic layer V of Dam-Dam Cheshme 2 (Figure 10, personal data, eastern Caspian region, Turkmenistan).

Figure 10

Shell beads documented within the Epipaleolithic layers of Dam-Dam Cheshme 2 (Turkmenistan): (a[a]) T. pallasi perforated by percussion on the dorsal side of the last spire whorl; (a[b]) T. pallasi perforated by abrasion on the dorsal side of the last spire whorl; (b) Dentalium sp. with use-wear on the distal and proximal extremities; (c[a–c]) anthropic perforation and use-wear on a fragment of Didacna sp. naturally eroded by surf action; and (c[d–f]) anthropic perforation and use-wear on a marginal fragment of Didacna sp. naturally eroded by surf action.

Beads made from Didacna sp. and T. pallasi from two contemporaneous sites in the southern Caspian region suggest that these same shell species may have been brought to the Kaylu rockshelter for the same purpose, sometime during the Caspian Mesolithic. The other shell species found at Kaylu were not documented as beads, although they were unlikely to have been used for consumption. The lack of geoarchaeological data concerning the impact of the Caspian Sea regression/transgression episodes on the estuary and river systems, surrounding the rockshelter and on the layers formed within the site, prohibits further discussion on the origins of the shell accumulation, though a natural origin cannot be excluded.

The three D. polymorpha, the Didacna sp. valve, the Turricaspia sp., and the Theodoxus pallisi, present in burial 1 show no anthropic modifications, and no data are available regarding their location in the burial or their relation to the body (Okladnikov, 1951). This lack of spatial information renders their strict contemporaneity with the skeleton inconclusive, and their presence within the filling sediment of the grave is a possibility.

The discoid shell beads from burials 1 and 2 were shaped on bivalves attributed to the Didacna genus. The difference in shell thickness between the shell beads, the thinner Didacna sp. from the rockshelter, and the unmodified valve unclearly associated with burial 1 may be due to their various states of maturation (juvenile versus adult specimen) or their collection from differing environments, an external factor well known to impact shell size, or to the presence of different Didacna shell species in the material (Fisher, Rhile, Liu, & Petraitis, 2009; Reimchen, 1982). The thickness of the large shell beads compared to the mid-/small-sized beads may also reflect similar factors.

Our results show that the portion of the shell dedicated to be perforated was not randomly selected and was made using a precise percussion process. After perforation, the shell flake was shaped into a disc corresponding to one of the three size categories identified in the collection. The nearly complete absence of overlap between the various size categories indicates that the shaping of these distinct sets of beads was performed separately to obtain a highly standardized morphology and size (Figure 11). The abrasion of shell flakes assembled on a string would have allowed for such standardization.

Figure 11

Manufacturing process of the discoid shell beads from the Kalyu burials.

The presence of three size categories suggests that the beads may have been suspended in many different configurations. However, preservation of the beads maintained in the concretion hull shows homogeneous sets exclusively composed of beads attributed to the same size categories (Figure 3). This result echoes observations made by the excavators in the field who identified three types of beads: some big and flat and some smaller and thicker of two different sizes. Many of the smallest beads were observed located next to the skulls and interpreted as part of a headdress. The biggest ones were mainly observed on the right side of the skeletons next to the pelvis and right femur (Okladnikov, 1951). No indication is available concerning the mid-sized beads. Our morphometric results also show that the large beads are the thickest, indicating that the field observations were somewhat inaccurate.

The presence of a red sediment covering the skeletons was reported during excavation, suggesting that ochre was spread on the bodies during the funerary rites (Okladnikov, 1951). Observation of the human remains shows the presence of a red compound on the surface of the two skeletons, particularly visible through the slightly red coloration of the skull, hand and long bones (Figures S1 and S2). The residual sediment present in the storage boxes shows that the deposit surrounding the bones was also reddish and rich in iron oxide. The residue observed on the beads adheres more firmly and is redder and probably richer in hematite. Discrepancies between the sediment and light-red coloration of the bones and the residue observed on the beads support two different uses of ochre – the intentional use of a hematite compound for the beadwork and the use of another red pigment for the funerary rites.

The red compound on the beads may result from the use of a pigment to change the natural color of the beads, the use of hematite as a fine-grained abrasive during bead manufacture (Rigaud et al., 2014), or the use of iron oxide as part of an adhesive compound used to attach the beads (Dayet, Erasmus, Val, Feyfant, & Porraz, 2017; d’Errico et al., 2005; Orton, 2008; Rigaud et al., 2014). However, the high concentration and the thickness of the red residue on the beads from the two burials does not plea for the use of hematite solely as an abrasive during bead manufacture. The location of the residue strictly distributed on the ventral and dorsal sides of the large beads and its near complete absence from the perforations suggest that both sides of the beads were coated. The disposition of the beads aligned on a single string indicate that they may have been tightly attached to each other with an adhesive. The pattern is different for the small- and mid-sized beads that show a high concentration of red residue cemented within the perforations. The location of the residue supports the hypothesis that a red pigment was used to deliberately stain the strings used to attach the beads. The presence of large beads near the pelvis and small beads near the skull indicates that two different systems of suspension (applying glue or using red string) were used for two different parts of the body to create disparate visual effects.

4.2 Regional Comparisons

Movius (1953) described Okladnikov’s (1949) excavated finds of early Neolithic Didacna sp. shell beads in several stages of manufacture at Cape Kuba-Sengir, located approximately 8 km ESE of the Kaylu rockshelter. Movius reports that Okladnikov (1949) considered this site to be a Neolithic workshop and that it “suggests a stage of industrial specialization in the production of ornaments that probably indicates trade relations over a fairly wide area.” Analysis of the 843 shell beads from the Okladnikov collection shows significant differences with the material from Kaylu (Figure 12 and Table 7). At Kuba-Sengir, the beads are represented by perforated Didacna sp. valves naturally eroded by sand (Figure 12a–c) and commonly found within thanatocenoses (Rogalla, Amler, & München, 2007). Perforations were made by bifacial rotation (Figure 12c–i) and the natural shape of the eroded valves does not show any anthropic modification. The presence of unperforated eroded valves in the material (Figure 12b) supports the workshop hypothesis previously mentioned by Okladnikov.

Figure 12

Shell beads documented at Kunda-Sungir: (a) Didacna sp. shell bead from the Okladnikov collection, (b) intact, naturally eroded Didacna sp. valves discovered at the site, (c) perforated Didacna sp. valves, and (d–i) perforations made by bifacial rotation.

Didacna sp. discoid shell beads documented at sites in the Caspian Sea region, Uzboi, Ustyurt, and the lower reaches of the Amu Darya are dated from the late 5th to the late 3rd millennia BC (Vinogradov, Itina, & Yablonsky, 1986). Small discoid shell beads, similar to the small beads recovered from the Kaylu burials, are documented at the Chalcolithic burial site Tokmak, located near the Caspian Sea in the Mangyshlak peninsula (Astaf’ev, 2014). At Tokmak, the discoid beads were found associated to Dentalium sp. shells, discoid stone beads, several small clay cylinders, and perforated Theodoxus sp. (Table 7 and Figure 14). Similar shell beads were discovered at Aktau 1 and Koskuduk 1 and 2, all sites belonging to the Oyukly culture (Astaf’ev, 2014). The discoid shell beads recovered from the Kaylu burials also show correspondences with those documented at the northern Caspian Chalcolithic cemetery, Khvalynsk (Kirillova et al., 2018), which had been shaped from Pleistocene fossil species of Didacnoides cf. caucasicus and belong to one size category (Kirillova et al., 2018). The location of the fossil outcrops indicates contact with the southeastern Caspian region near the Buzachi peninsula, currently in Kazakhstan (Kirillova et al., 2018). Okladnikov found some analogies between the material from the Kaylu burials and the Late Neolithic Mariupol cemetery located in North Azov sea (Okladnikov, 1966). One Early Neolithic site located in the southwestern Caspian area, MPS 4 (Azerbaijan, Heit, 2014), also shows similarities with Kaylu. The presence of Didacna sp. shells, fragments, blanks, and finished beads indicates the production of discoid shell beads at the site (Heit, 2014). The morphometric data available in the literature indicate that only one bead size category was present at MPS 4 (Heit, 2017), probably corresponding to the small beads identified in the Kaylu burials. The large discoid shell beads from Kaylu also echo the bead material from the Tuzovskiye Bugry-1 burials in Altai (Kiryushin, Kiryushin, Schmidt, Kuzmenkin, & Abdulganeyev, 2011). These discoid beads are made from Colletopterum sp. bivalve fragments, originating from the Ob Bassin (Kiryushin et al., 2011). The presence of Dentalium sp. and Corbicula sp. shell beads suggests contact between Tuzovskiye Bugry-1 and the southern areas of Western Central Asia (Kiryushin, Kiryushin, Schmidt, & Abdulganeyev, 2012). Large and small discoid shell beads were also documented at the Tumek-Kichidjik cemetery (Northern Turkmenistan, Vinogradov et al., 1986). Stone beads and pendants, clay plugs, bone beads, bivalves, and gastropods originating from the Persian Gulf or the Red Sea have been documented in the Early Neolithic village of Jeitun (Southwest Turkmenistan, Figure 13), but no discoid shell beads have been reported so far (Harris et al., 2010). At the Jebel Neolithic site (Great Balkan Turkmenistan), naturally eroded Didacna sp. shell valves were documented, but discoid shell beads are absent from the material (personal data and Okladnikov, 1956). Near the southern Caspian Sea, in Iran, perforated Didacna sp. associated with various stone beads are documented in the Early Neolithic Tappeh Sang-e Chakhmaq East Mound (Roustaei et al., 2015); however, no discoid shell beads have as yet been documented in the Early Neolithic occupations (Table 7 and Figure 14).

Figure 13

Material from Jeitun (Turkmenistan, Masson excavations): 1–5 limestone pendants, 6–8 calcite pendants, 9–16 clay plugs, 17–18 calcite plugs, 19 green stone tubular bead, 20 and 22–25 discoid calcite beads, 21 chalcedony tubular bead, 29–39 tubular bird/small carnivorous bone beads, 40–43 Cypraea sp. shell beads, 44–48 Mitrella cf. agatha, and 49–50 Cardiidae.

Figure 14

Map of the sites mentioned in the text: (1) Kaylu; (2) Kuba-Sengir; (3) Jebel; (4) Dam-Dam Chesme-2; (5) Jeitun; (6) Aktau-1; (7) Koskuduk-1, 2; (8) Tokmak; (9) Tumek-kichidjik; (10) Tuzovskiye Bugry-1; (11) Khvalynsk; (12) Mariupol cemetery; (13) MPS-4; (14) Belt-Kamarband; (15) Altappeh; and (16) Tappeh Sang-e Chakhmaq.

No discoid shell beads were found in the Caspian Mesolithic layers from the Kaylu rockshelter (Okladnikov, 1951). Only the presence of the freshwater gastropod T. pallasi echoes another Caspian Mesolithic occupation at Dam-Dam-Cheshme 2, where the same species was perforated and used (personal data, eastern Caspian region, Turkmenistan, Figure 10). Other pendants present in the Caspian Mesolithic layers of Dam-Dam-Cheshme 2 include Dentalium sp. and the perforated vestigial marginal edge of Didacna sp. valves. Similar bivalve fragments are commonly found within thanatocenoses and result from extreme sand abrasion by surf action (Rogalla et al., 2007). Broken perforations with use-wear have been observed on two eroded marginal shell fragments at Dam-Dam-Cheshme 2. Similar shell pendants are documented within the Epipaleolithic occupation of Altappeh (Iran) located in the southern Caspian Sea (Manca et al., 2018). Personal ornaments, including one perforated fragment of Didacna sp., were documented in the Mesolithic layers of the Belt Kamarband cave (Coon, 1951).

The shell beads associated with the Kaylu burials do not correspondence with the material from the rockshelter, or with other Epipaleolithic/Caspian Mesolithic occupations documented in the southern Caspian region. Differences between the Neolithic shell beads of the regions, including the Early Neolithic material from the two burials documented at Kaylu and the Epipaleolithic/Caspian Mesolithic ornaments documented in the southern Caspian, indicate a sharp typo-stylistic discontinuity between the two periods.