Introduction

During the fifth millennium BC, in the eastern Balkans, several social innovations emerged for the first time on the European continent. Notably, the widespread dispersion of cemeteries as standard burial practices was a novel phenomenon, along with the rise of social hierarchy within the targeted area, the expansion of an extensive Eurasian trade network, the advent of metallurgy, and the establishment of substantial permanent tell settlements1. These events were concomitant with the spread of a common material culture over a strikingly large area due to a second wave of migration of farmers from northwestern Anatolian Neolithic-related ancestry2.

In contrast to the fragmented cultural landscape of the first half of the fifth millennium BC, the Kodjadermen-Gumelnița-Karonovo VI archaeological signal (KGK VI) eventually covered most of the Balkan Peninsula (northeastern Greece, eastern side of Bulgaria, southeastern Romania3,4,5 and even extended to neighboring Moldova and to southern Ukraine for a few generations, as the KGK VI communities reached its maximal territorial extension5,6. The ceramic production of the KGK VI human groups borrowed many characteristic features of the previous archaeological pottery styles (cultures) found in the lower Danube River Valley while appearing to be a natural evolution of the former Boian-Vidra and Boian-Spanţov human groups of southern Romania. Consequently, this area seems to have played a key role in the formation of the subsequent KGK VI material culture identity. In addition, a recent study by Popescu et al.5 based on the probability distributions of radiocarbon dates, investigated potential regional and local variation population dynamics; the authors concluded that the KGK VI human population density grew steadily for the first approximately 300 years after their original appearance in the early fifth millennium BC (4901 − 4751 cal. BC), with at least two initial cores (north and south of the Danube), one of them being located in the area of the Mostiştea River Valley. Following an expansion stage from the two cores, within two distinct timespans (4751 − 4665 cal. BC and 4665 − 4580 cal. BC), both spread in the south/southwest direction. Eventually, during a third and final expansion phase (4580 − 4531 cal. BC), those human communities covered the maximum area of their territorial extension. These growth episodes were followed by an equally fast decline, without any significant recovery, until it vanished from the archaeological record toward the end of the fifth – early fourth millennium BC5. It should be underscored that the site studied here was home to one of the most resilient communities of south Romania, and was occupied until the very end of the KGK VI civilization.

Several authors have alluded to the possible role of migration as a driving force behind social change in the eastern Balkans during the Eneolithic, especially concerning the start7 and the end of the period1,8. In addition, Merkyte and Albek9 insisted on the importance of kinship and marital alliances on the patterning of material culture, a factor that Ştefan10 also pointed out as a possible mechanism for the spread of the KGK VI communities.

The decline of this flourishing civilization was triggered by an economic and cultural collapse due to significant climatic changes around 4300 cal. BC, known as the 5.9 ka BP event, which was characterized by cooling temperatures, increased aridity, and decreased precipitation. The KGK VI decline unfolded gradually over approximately 550 years, spanning from roughly 4350 BC to 3800 BC. While these environmental changes are linked to the civilization’s downturn, the underlying drivers of these trends—particularly the mechanisms behind the prolonged decline—remain uncertain and necessitate further investigation5,24,59.

However, recent aDNA investigations in southeastern Europe have shown that the populations responsible for the above mentioned “cultures” of the second half of the sixth millennium and fifth millennium BC, KGK VI included, had similar genetic features and common ancestry due to their common origin in northwestern Anatolia11. In addition to the northwestern Anatolian Neolithic-related ancestry, the KGK VI groups had a consistent Balkan hunter-gatherer related ancestry12along with sporadic evidence of steppe-related ancestry (Varna I and Smyadovo cemeteries), which points toward a genetic admixture between these populations2,13,14,15.

Fig. 1
figure 1

Elevation map showing the sites cited in this study, together with the locations of the plants sampled to establish the Biologically available strontium (BASr) baseline (see Table 4 below).

Full size image

So far, direct evidence for past mobility through strontium isotope analyses of human remains belonging to KGK VI groups has only been gathered at Pietrele, Romania16 Smyadovo, Bulgaria17 and Yunatsite, Bulgaria18. The results from these studies pointed at a certain amount of mobility taking place within these communities, but their interpretation was impaired by the lack of systematic endeavors to establish a solid baseline around the sites under inquiry. Sporadic data on individuals from earlier Neolithic groups of the Danube River Valley are also available at Ajmana, Cârcea, Gârlești, and Căscioarele19,20 (Fig. 1).

The recent resumption of the excavations at the KGK VI cemetery of the Gumelnița eponymous site, in southern Romania, offers an outstanding opportunity to improve our understanding of mobility on this site and in the northern Balkans during the Eneolithic through strontium isotope analyses (87Sr/86Sr) on tooth enamel. Material retrieved during previous excavation campaigns21,22,23,24 are also included in this study. In addition, proteomic analysis of amelogenin supplements morphological methods for sex identification, greatly improving our knowledge of the life history of the individuals being studied.

Fig. 2
figure 2

Investigated zones on the Gumelnița site and location of the excavation areas (marked with red) made between 2017 and 2022 by our team.

Full size image
Fig. 3
figure 3

Graves and features from the Gumelnița cemetery included in this study: (a) M1/1962; (b) C12-I3 (the mandible); (c) M10; (d) M7; (e) M9; (f) M1; (g) M15; (h) M16; (i) M3/1962.

Full size image

The Gumelnița site

The site consists of a tell settlement and a cemetery, along with an off-tell area that was also in use during the ancient occupation of the site (Figures S1, 2). The Gumelnița tell is the biggest KGK VI tell settlement north of the Danube. The site is located on the Danube River floodplain, immediately south of the confluence area of the Argeş River, in southeastern Romania21,22,23,25.

The tell rests on a rectangular erosion remnant (272 × 227 m) of the high terrace of the Danube and is made up of loess deposits (Figures S1, 2). The maximum altitude of the tell is 38 m.a.s.l, and its area is about 6 ha21,22,24. The archaeological deposits are over four meters thick in the top area and five meters thick in the southeastern corner of the tell mound where our excavations took place. It was occupied by KGK VI communities during the entire relative chronology of the Gumelnița archaeological ‘culture’ (A1, A2 and B1 phases)21,25. Based on the 14C data obtained from animal bones and seeds for the tell settlement (n = 14), the absolute chronology is 4782 − 4000 cal. BC with a 95.4% probability22.

The cemetery is located c. 200 m east of the tell settlement, on a high terrace of the Danube River (Fig. 2). It was excavated for the first time in the 1960 s, when five graves were excavated and another seven or eight graves were destroyed by public works21,26,27. After that, rescue excavation of 1977 allowed for the retrieval of three other KGK VI graves, but several other graves accidentally were again destroyed by public work28. In 2017, planned excavations were resumed in the cemetery of Gumelnița site, leading to the discovery of another 23 inhumation graves (Figures S3-S4), 21 of which belong to KGK VI communities21,22,23,24.

Generally, throughout the KGK VI occupation of the site, the Gumelnița cemetery contained single inhumation graves, and several pits (Figs. 3 and 4). Among the latter, two pits (C12 and C21) represented special situations: containing human bones without anatomical connection deposited in the filling levels of the respective pits, in association with Gumelnița ceramic fragments, burnt adobe, animal bones and plant seeds. In the case of the C12 pit, two skeletons in anatomical connection (M15 and M16) were discovered close to the bottom of the pit but in overlapped layers (Fig. 4.h-i), deposited face to face on the sides of the pit, and human bones from at least six other individuals were also discovered in the upper part of the pit filling (Fig. 4.b). Grave M9 was almost destroyed by contemporary works (Fig. 4.e).

Another atypical situation is represented by the presence in some graves (e.g., M1, M2, M3) or other features that contains human bones/skeletons (e.g., C12, C21) of additional bones from other human individuals that those buried in anatomical connection21,22,23,24,26,27. This custom involves the deliberate incorporation of bones from another skeleton into a regular burial29 a custom that has also been documented in other KGK VI cemeteries from the Balkans (e.g., Varna I, Devnja, Durankulak, Goljamo Delchevo, Vinitsa, Vărăști, etc.)24,27,29,30,31.

The graves from Gumelnița contain skeletons laid in a crouched position, usually on the left side (Fig. 4.a, c-d, f, h-i), rarely on the right one (Fig. 4.e, g), oriented with the head in an easterly direction (in most of the cases). Only the individual from grave M3/1962 (Fig. 4.j) was found in an extended dorsal position, a typical burial position for the KGK VI communities that lived on the Black Sea coast (the so-called “Varna culture”), with similar cases being documented in the Palazu Mare, Varna I, Durankulak, and Devnja cemeteries3,27,32. The grave goods are not very spectacular (e.g., clay pots, stone or copper axes, flint blades, bone figurines, copper chisel or shell beads), and have been documented in 11 graves (Table 1). In addition, plant remains were identified in some graves21,22,23,24,26,28. Based on the 14C data obtained for the features investigated in the cemetery area (n = 12), the absolute chronology is 4545 − 4054 cal. BC (Table 1; Figure S6) with a 95.4% probability22.

Fig. 4
figure 4

Graves and features from the Gumelnița cemetery included in this study: (a) M17; (b) M18; (c) M21; (d) M22; (e) M23; (f) C21.

Full size image

Considering the impact of the aquatic resources in the diet of Gumelnița inhabitants, the radiocarbon dates were calibrated with the percentage of freshwater resources indicated by the FRUITS software. As a result, the corrected dates for the reservoir effect (FRE) have a delay of an average of 147 years24 which represents an acceptable deviation that does not influence the conclusions related to the KGK VI community from Gumelnița (Table 1; Figure S6).

Human population

The state of preservation of the human skeletal remains is generally bad or poor. Among the taphonomic markers identified on the surface of some of the bones, the presence of a calcareous crust, exfoliated periosteum, whitish-colored periosteum, together with traces of a red pigment (ochre), periosteum showing imprints of plant roots, or traces of secondary burning, have all been mentioned33.

The minimum number of individuals is 38, including the skeletons from the old excavations from the 1960 s (Table 1). The majority of the remains were discovered in the cemetery (graves and pits), but a child’s skeleton was also found in the same period in the tell settlement21,22,23,24,25,26,33.

The degree of synostosis of the cranial sutures34 the aspect of the sternal ends of the ribs35 the aspect of the pubic symphysis and the aspect of the auricular surfaces34 were used to estimate the age of adult skeletons. The age of sub-adult individuals was estimated based on recording the stage of dental eruption36. The age groups comprise children (n = 5), adolescents (n = 3), young adults (n = 11), adults (n = 4), mature adults (n = 12) and old adults (n = 3)22,23,33. Based on the currently available data (Table 1), the death frequency of the young population segment (Young adults = 29% and Adolescents = 8%) appears three times as high as the one of children (13%). Old age (over 50 years) is also reached frequently (8%), and the mortality within the mature adult population (over 35 years) is 39%.

The skeletal sex of adults was determined using cranial indicators (supraorbital margin, glabella, mastoid, mental eminence and nuchal line according to Buikstra and Ubelaker37) and postcranial (subpubic concavity, subpubic angle, ischiopubic ramus, ventral arch, compound arch, large sciatic incision after Steckel et al.36). The sex categories show a balanced situation: 15 females, 11 males, and 12 indeterminable individuals, represented by children (n = 4), adolescents (n = 3) and adults (n = 5)22,23,33. Sex cannot be assessed for these adult individuals because only a few pieces of their skeletal remains were retrieved, often in a fragmented state, either in pit C12 or as an added element in other primary graves (Table 1).

The health status of the KGK VI groups that lived at Gumelnița was determined based on the paleopathological investigations. Thus, a series of dental and bone diseases were identified. We mention antemortem tooth loss for 5 individuals (M3-I1, M6, M10, M11, M16) and dental caries for 2 individuals (M10, M12) in the first category. The most frequently observed bone disease was periostitis, identified in 6 individuals: three children (M2-I1, M3-I2, M13) and three adults (M5, M6, M9). Only in two cases (M10 and M11) manifestations of osteoarthrosis were found. Periostitis is generally regarded as an osteological indicator of infectious or traumatic diseases. Still, it can also appear due to a nutritional imbalance38. It can also represent a disease or results from specific health conditions39. In the individual from M4, a woman no older than 30, possible manifestations of Paget’s disease were identified, a condition of the bone system also known as osteitis deformans - a disease with a metabolic and endocrine substrate characterized by bone resorption phenomena (lytic), simultaneously with new bone tissue formation processes (sclerotic)40. Among the identified markers of the disease, we mention the hyperostosis of the left frontal and parietal bones and the thickening of a part of the right femur. However, we remain reserved regarding the diagnosis due to the absence of bone deformations (bows), which, even if they were present, cannot be observed due to the high degree of fragmentation of the individual’s long bones22,23,33.

Table 1 KGK VI graves from the Gumelnița site, with the indication of sex, age, pathology, grave goods, radiocarbon and stable isotopic data. “M” indicates the grave, and “I” is the individual. *Next to the grave number denotes the main individual of the grave. *Next to the sex indicates that proteomic analysis of amelogenin in tooth enamel was performed.
Full size table

Dietary profiles of Gumelnița inhabitants based on the analysis of δ15N and δ13C (Figure S4) indicate that around three quarters of the diet of the KGK VI people came from terrestrial resources, with half coming from plants and one quarter coming from animals. The remaining quarter of the diet was composed of freshwater resources, which could be further divided into shellfish (around two thirds of it) and fish (accounting for a third of that component)24. The dietary profiles of the Gumelnița community are like those from other Neolithic and Eneolithic sites in Bulgaria and Romania that have been investigated isotopically (Figure S5).

Results

Strontium isotope results

The 87Sr/86Sr values of the human samples from the Gumelnița site vary from 0.7077 to 0.7103, with an interquartile range (IQR) of 0.7091–0.7096 (Table 2). A visual inspection of a 87Sr/86Sr and [Sr] biplot suggests the existence of two groups, one being constituted by M15 and M16, and the other one comprising all the other individuals. A Grubb’s test performed on the individuals’ average strontium isotopic values reveals that M15 constitutes a significant outlier in this dataset at p < 0.05. Once the average of the two values for M15 are removed from the set, no outlier was identified, even though M16 is further from the rest of the values (G = 2.907, p = 0.009142). The strontium concentrations ([Sr]) range from 69 to 280 ppm (Fig. 5). A Grubb’s test performed on the individuals’ average strontium concentrations reveals no statistical outliers in this dataset (G = 2.474, p = 0.1003).

Fig. 5
figure 5

Biplot showing the relationship between 87Sr/86Sr and [Sr].

Full size image

Individuals for which two different teeth were analyzed (M09, M15, M16, and C12-I3) have similar 87Sr/86Sr values on both teeth (difference lower than 0.0001) (Table 2), suggesting an absence of mobility during their childhood.

To create the baseline, we chose to use modern plants as recent studies have underlined that they accurately represent the bioavailable strontium at a given location41,42. In our area of study, they also offer the advantage of being easily accessible in all geologically distinct backgrounds. Consequently, we were able to collect samples all directions and at consistent length intervals around the site, the furthest sampling point being located at approximately 85 km away from the necropolis. Such an extensive and systematic coverage of the grounds surrounding the sites would not have possible if we had opted to rely on other types of samples. The 87Sr/86Sr values of the modern plants sampled in the vicinity of the Gumelnița site range between 0.7092 and 0.7112, with an IQR of 0.7096–0.7106 (Table 3). Plants closer to the Danube River have the lowest 87Sr/86Sr values, and fall in the lower end of the range with an average value of 0.7092 ± 0.0001 (2SD). We extracted the distance to the nearest river and the altitude of the sampling locations with a GIS software to examine the relationship between these variables and the 87Sr/86Sr values; in the former case, the correlation was neglectable (R2 = 0.06), while in the latter we observed a weak positive correlation (R2 = 0.36). We also classified the sampling locations into four different categories according to their geological background (Fluvial Valley: 3 locations; Holocene Terrace: 4 locations; Northern Bank Pleistocene Terrace: 8 locations; Southern Bank Pleistocene Terrace: 5 locations) and performed a One-Way ANOVA to investigate a possible effect of the geology on the 87Sr/86Sr values. The test revealed that there was a statistically significant difference in between at least two categories (F(3, 56) = [14.18], p < 0.00001). Tukey’s HSD Test for multiple comparisons found that the mean 87Sr/86Sr value was significantly different between Fluvial Valley locations and all other groups (for Holocene Terrace, p = 0.02731, 95% C.I. = 0.0000594, 0.001363; for Northern Bank Pleistocene Terrace, p = 0.000002, 95% C.I. = 0.0006804, 0.001836; for Southern Bank Pleistocene Terrace, p = 0.000003, 95% C.I. = 0.000711, 0.001957). There was also a significant difference between Holocene Terrace locations and Northern Bank Pleistocene Terrace locations (p = 0.037, 95% C.I. = 0.00002456, 0.001069), and between Holocene Terrace locations and Southern Bank Pleistocene Terrace locations (p = 0.0278, 95% C.I. = 0.0000507, 0.001195). Conversely, the test did not find any significant difference between Northern Bank Pleistocene Terrace locations and Southern Bank Pleistocene Terrace locations (p = 0.9759, 95% C.I. = −0.0004104, 0.0005624).

Table 2 Biological sex, tooth type, 87[sr]r/86[sr]r with two standard errors, and [Sr] for the individuals analyzed in this study. The biological sex of individuals marked with * were determined through proteomic analysis of amelogenin in tooth enamel.
Full size table
Table 3 Geographic coordinates (WGS 84), type, and 87Sr/86Sr with two standard errors for the plant samples analyzed in this study.
Full size table

Proteomic results

After analyzing ion chromatograms, we found that the AMELX peptide SIRPPYPSY was present in all the samples, while the AMELY peptide SM(ox)IRPPY, together with the other three AMELY peptides monitored, was present in 4 out of 7 of the individuals (Fig. 6 and S7).

Fig. 6
figure 6

Extracted ion chromatograms (XICs) for amelogenin peptides. Four different XICs monitoring AMELX peptides are displayed on the left panels (sample C12-I3 on top panel, sample M1/1960 on bottom panel), whereas four different XICs monitoring AMELY peptides are displayed to the right. Precursor m/z and normalized ion count levels (NL) are displayed next to the corresponding XICs. Each ion count represents the sum of the intensities of fragment ions indicated in Table 2. All four amelogenin Y peptides are clearly detectable in sample C12-I3, whereas they are absent in sample M1/1960. AMELX peptides are monitored as positive controls.

Full size image

Discussion

The plant samples collected in 20 different locations around the site indicate a range of 87Sr/86Sr BASr included between 0.7092 and 0.7112, which is in line with the expected values for Cenozoic sediments according to Ventresca Miller et al.43. It should be noted that the 87Sr/86Sr values of the plants sampled nearer to the Danube River tend to be lower than those sampled further away, and that overall, Pleistocene deposits exhibit higher Sr ratios than Holocene deposits (Fig. 7). It is likely that the original Neolithic landscape of the river valley has been heavily remodeled since then, and that a thick amount of more recent sediments sits on top of the soils accessible to prehistoric populations. As a consequence, sampling the area currently occupied by fluvial deposits is problematic. However, the location we selected used to be an island surrounded by a lake until the second half of the twentieth century, when considerable efforts were exerted to drain the residual water and turn those wetlands into agricultural fields. Several neolithic settlements and a large KGK VI necropolis were excavated there in the 1960s27,44. It is therefore more likely to provide a strontium isotopic signature similar to those found in the area in the past.

Fig. 7
figure 7

Simplified geological map showing the median 87Sr/86Sr value for plants at each sampled location.

Full size image

The 87Sr/86Sr found in human tissues results from the mixing of the 87Sr/86Sr signature of the food sources and, to a lesser extent, drinks45,46. Therefore, food items with a high content in strontium will have a relatively larger impact on the final 87Sr/86Sr of human tooth enamel47,48,49. It is likely that such plants were locally grown, and it can be assumed that individuals that have 87Sr/86Sr outside the range of the baseline defined by the plant samples consumed food that was not grown locally, and are likely to have spent their childhood in a location with a different BASr signature.

At Gumelnița, the lower end of the 87Sr/86Sr baseline defined by the plant samples overlaps with the higher values of the range of the human, which is consistent with the hypothesis that 13 out of the 17 individuals analyzed in this study spent their childhood years in the area (Fig. 8). Conversely, four individuals, 1960-M1, M3/1962, M15 and M16 have 87Sr/86Sr that are lower than 0.7092. Consequently, they should be considered as non-local based on the current plant data available. However, it should be underscored that only four of the local individuals display strontium ratios consistent with the range defined by the location sampled right next to the site, and where the cemetery could be found; the rest are deemed local uniquely on the premise that the values found on the island of Vărăşti, which is located a little bit less than 30 km, are representative for the fluvial deposits found around the site. The concentration of the 87Sr/86Sr values of human individuals in the lower end of the BASr baseline may also be seen as evidence for mobility taking place exclusively alongside river valleys. In addition, C21-A exhibits a strontium isotopic that is only slightly within the range of the local baseline.

Fig. 8
figure 8

Plot of the 87Sr/86Sr values for humans at Gumelnița, with the BASr derived from the plant 87Sr/86Sr values shown in green. The plant samples collected in the immediate vicinity of the site (site BASr) are indicated in dark green.

Full size image

Within the KGK VI archaeological group, 87Sr/86Sr values lower than 0.7090 are found on the southern bank of the Danube River, at a considerable distance from Gumelnița. At Smyadovo, the IQR for the nine Eneolithic human tooth enamel samples were 0.7074–0.7089, while the IQR for the three human dentine samples was 0.7082–0.708517. At Yunatsite, out of twelve individuals, only one was considered as local, with an 87Sr/86Sr of 0.7111 similar to the value of the soil found at the foundation of the tell settlement (0.7113)18. The IQR for the non-local individuals is 0.7087–0.7092, with two individuals considered as outliers. Once they are both removed as well, the IQR stands at 0.7087–0.709. In contrast, the IQR for Pietrele, on the northern bank of the Danube, is 0.7097–0.710116. If we also consider the 87Sr/86Sr values for individuals from earlier Neolithic cultural horizons, we can observe a rather large IQR range at Ajmana (0.7087–0.7097 for 11 individuals)19; it should be underscored that according to the authors, the degree of mobility of the population was quite high during the period, and these numbers may include a fair proportion of non-locally born individuals. Regarding the other Neolithic sites recently reported by Price and Borić20 at Gârlești, the average 87Sr/86Sr value (n = 4) was 0.7109, while at Cârcea, the authors give an average 87Sr/86Sr value of 0.7109 for a group of individuals, and another value of 0.7127 for a single individual presumably from the same site spelt differently. The only individual analyzed at Căscioarele, displays a value of 0.7089, which does not fit well with our baseline. Unfortunately, there is no information regarding the archaeological context where this individual was found. Still, these ranges have to be considered with caution as no adequate BASr baseline exists for these regions.

This trend of lower 87Sr/86Sr values south of the Danube River is supported by the analysis of soil leachates in Bulgaria50. It is also suggested by the 87Sr/86Sr values for regional rivers reported by Zitek et al.51. In effect, the values for the Danube River are comprised between 0.7091 and 0.7095 in a 50 km radius around Gumelnița, the 87Sr/86Sr of the Argeș river, on the northern side, rises to 0.7100, while those of the Yantra and Russenski Lom Rivers on the southern side are around 0.7085. However, it should be noted that surface water is not always considered as a reliable indicator of the local BASr52.

It is likely that M15 and M16 grew up in a region with a geological background quite different from the one around the Gumelnița site, while 1960-M1 and M3/1962 spent their childhood in an area whose BASr values only slightly differed. Furthermore, it remains intriguing that most other individuals analyzed in this study also exhibit lower 87Sr/86Sr values compared to those of the plants found in the immediate vicinity of the site. Such a situation may result from the community focusing on the nearby floodplains rather than on the terraces to produce their food. An alternative explanation could be the consumption of imported food items with a different 87Sr/86Sr signature, even though isotopic analyses point at a relatively homogeneous diet profile within the Gumelnița community24. Besides plants that are more likely to have been grown locally, salt and salted foods represent one of the food items that have a high enough content in strontium to have an impact on the final 87Sr/86Sr found in human tissues53.

Existing isotopic studies indicate that Eneolithic populations of the northern Balkans had a mixed diet of C3-C4 plants and terrestrial animals24,54 including in the community that lived at Gumelnița, even though material evidence suggest that at least some of these communities were heavily focused on the exploitation of river and marine resources55. Salt is a key ingredient for the preservation of fish, and salted fish has been an important food item in several cultures across the world56. It has also been used as a trading good and corresponds to the expansion of the Hamangia and Cucuteni human groups, which were preceding and coeval with the KGK VI communities, respectively57. Brigand and Weller’s58 spatial analysis showed the interest of the Cucuteni populations of Moldova for the saline springs. Evidence for the large-scale exploitation of brine derived from rock salt has also been discovered at the Provadia-Solnitsata, in Dobruja, Bulgaria59 which is contemporary with KGK VI. The salt produced in these regions may have an 87Sr/86Sr signature significantly different from the one found in plants grown around the site of Gumelnița, altering the 87Sr/86Sr value of the individuals who consumed a certain amount of salt originating from these sources.

Quite remarkably, the 87Sr/86Sr data seem to indicate that the individuals that were not born locally were given a deviant type of burial in comparison with the standard funerary customs of the group. Proteomic analyses have also contributed to better understand those unique identities by providing us with reliable biological sex attribution for individuals for whom morphological methods would have produced unreliable results at best. Thus, as mentioned previously, both the position of M3/1962 (extended dorsal) and the goods found in his grave are strongly reminiscent of those associated with populations of the eastern steppes, but also in Varna region near the Black Sea. The little girl labeled as 1960-M1 is the only individual found buried on the tell rather than in the necropolis so far. The two women found together in a pit, deposited in the pit base, on its opposite sides (face to face, but with bodies being oriented in opposite directions) associated with isolated human remains, M15 and M16, also appeared to have been born in a different community. Conversely, a foreign origin cannot be retained as a valid explanation for the seemingly different post-mortem treatment of the isolated jaws.

The proteomic analyses of amelogenin in tooth enamel performed for this study have allowed reliably assessing the sex of every individual for which strontium isotopic analysis was performed. Consequently, it makes it possible to explore whether the biological sex of an individual may have been related to their mobility in the KGK VI community of Gumelnița. 27% (3 out of 11) of the females analyzed in this study were non-local, while 17% (1 out of 6) of the males were non-local according to their 87Sr/86Sr values. These numbers contrast with those available for the other sites in the KGK VI area, where the results of strontium isotopic studies indicated that males were more mobile than females. For instance, at Smyadovo, three males out of the five studied were non-local, while the only female was deemed as local17; at Yunatsite, all the males (n = 5) were identified as non-local, and only one female (n = 6) was deemed as local18. It should be underscored that, contrary to the present study, the interpretation of the data for these two sites did not benefit from an extensive BASr baseline, and that the sample population is rather low.

The higher proportion of non-local females at Gumelnița site may suggest a preference for patrilocality in this group, and for the fact that female exogamy was more commonly practiced. However, it should be emphasized that these observations must be taken with precaution, considering the low sample size of this study. Nevertheless, since such an interpretation is commonly upheld by authors dealing with data on individual mobility in the Neolithic Linear Band Keramik groups of Central Europe60 the possibility that it was also practiced at Gumelnița site should also be entertained. The fact that non-local individuals of both sexes were buried at Gumelnița, but with a higher proportion of non-local females, is also compatible with other systems of post-marital residency, such as ambilocality (which here means that individuals of both sexes relocated in their spouse’s settlement after marriage), avunculocality, shifting residence, since some individuals may have moved between different settlements during their lifetime61 or virilocality (the couple staying at the estate of their groom’s kindred)62. However, as mentioned above, it should be noted that the only non-local male may have come from a culturally different community, which may have followed different rules when it comes to post-marital residency; besides, the non-local female individuals may also have been buried in Gumelnița cemetery for the same reason.

Last but not least, the radiocarbon dates of the four non-local individuals (1960-M1, M3/1962, M15, and M16) are of significant importance. When considering the freshwater reservoir effect (FRE) (Table 1), we can see that all these individuals date to a period following the peak of KGK VI civilization. Specifically, they fall after 4350 cal. BC, marking the onset of a prolonged decline lasting approximately 550 years (ca. 4350–3800 BC), ultimately leading to the disappearance of these groups from the archaeological record5. By the late 5th millennium BC, as this decline continued, the climate began to cool and dry, corresponding with the transition to the Subboreal period, marked by the Bond event 4 (5.9 ka BP event)24,63,64. This event, detected through studies of lipid biomarkers, pollen, and speleothems in the Carpathian Mountains65,66,67 coincides with the disappearance of many KGK VI tell settlements24. This pattern of decline, transformation, and collapse is also observed in other contemporary communities, such as Cucuteni and Tiszapolgar, which were similarly affected by the influx of new populations from the North Pontic steppes (e.g., Decea Mureșului, Cucuteni C, Csongrád-Kettőshalom)24,64,68. These migrations are evidenced by individuals with steppe-related genetic ancestry buried in cemeteries such as Varna I and Smyadovo in Bulgaria2. Naturally, this hypothesis needs to be confirmed by further studies on a larger number of individuals.

The economic collapse of KGK VI communities, which seems to have been primarily driven by climate change, led to significant transformations in their socio-economic systems. The presence of outsiders at the Gumelnița site may reflect a small-scale migration process triggered by these changes. The resilience of the Gumelnița community, which, based on 14C data, survived several hundred years longer than other tell settlements until the early 4th millennium BC, may explain why these individuals settled there. The community’s ability to adapt, using intensive soil fertilization, irrigation, and cultivating climate-resistant crops like barley24 allowed it to endure despite the challenges posed by Bond Event 4. This resilience could account for the presence of ‘refugees’ from other collapsing settlements. This hypothesis may explain the mobility observed in certain individuals whose burial practices suggest a different origin from the majority population at Gumelnița.

Conclusion

While the sample size is not large enough to generalize the results of this study to the whole KGK VI area, it should be noted that four out of seventeen individuals could be considered non-local, and that three of them were females (1960-M1, M15 and M16) while one was a male (M3/1962). Their 87Sr/86Sr value is consistent with having spent their childhood on the right bank of the Danube River. This means that these individuals may have been born in present-day Bulgaria and travelled a considerable distance to settle and eventually die in the Gumelnița settlement. These facts may reflect the existence of a complex phenomenon of migration that reflects the decline and collapse of the KGK VI civilization5 and may be connected to migrations driven by climate changes associated with the Bond event 4.

Considering the limited number of strontium isotopic data within the KGK VI area, and the difficulty to assess the biological sex of fragmented remains using standard morphological methods, it would be beneficial to extend similar investigations to other contemporaneous cemeteries in order to deepen our understanding of mobility pattern in the northern Balkans during the Eneolithic.

Finally, few studies have explored ancient mobility in the Lower Danube River Valley using strontium isotope ratios, leaving uncertainty about this area’s expected bioavailable strontium values. This lack of established baseline data complicates the interpretation of strontium isotope analyses from human or animal tooth enamel in the region. To address this gap, the current research focuses on developing a bioavailable strontium isotopic baseline for the Lower Danube River Valley, a potential game-changer in improving the accuracy of mobility studies in this part of the world.

Materials and methods

Strontium isotope analyses

Strontium is a geochemical tracer to study the mobility of past populations, as its isotopic value (87Sr/86Sr) is directly related to a geological source. This element has four stable isotopes; of particular interest is 87Sr, which is radiogenic and results from the decay of rubidium 87Rb). As such, the amount of 87Sr in a rock depends on the original amount of rubidium as well as the age of its formation69. Strontium cycles through a living ecosystem through weathering of the rock into soil and its absorption by plants, and their subsequent consumption by animals and humans70.

Strontium substitutes for calcium in bioapatite (the inorganic fraction of bones and teeth), allowing for the study of ancient human remains, where soft tissues are seldom preserved. Human tooth enamel is formed during childhood71 and its composition does not change afterward72. It is also considerably more resistant to diagenesis than bones and dentine73. Consequently, it is possible to compare the 87Sr/86Sr found in the tooth enamel of an individual with the isotopic value of the local geological background, and infer whether the deceased individual was born and lived in an area with a similar background or not. Little fractionation occurs when the element cycles through the ecosystem, making it a great geochemical tracer74. There are, however, several limitations that obfuscate the direct relationship between the strontium isotope ratio of a rock and the one present in the tissues of a human being. Different minerals found in an individual rock may have different rates of weathering, which means that the isotope ratio of the bioavailable strontium may differ radically from the one measured in the local geological background75. The sea spray effect, which occurs when strontium of marine origin reaches coastal soils, could also change the isotopic value of the bioavailable strontium76,77 as well as the use of modern fertilizers78,79.

Consequently, it is advised to constitute a baseline of the bioavailable strontium (BASr) at a local level80. Among the different types of samples used, modern plant samples from non-cultivated areas seem to better represent the ratio of the BASr at a given location81. Since there is no BASr reference in Romania, three plant species with different root lengths in a wood located near the site were analyzed, as well as seven other locations within a 30 km radius around the site, which should reflect the diversity of geological background found around the site. The procedure described in Snoeck et al.82 was followed. In total, for this study, we analyzed 21 tooth enamel samples from the 17 individuals whose teeth had been preserved. All of them could be securely dated to the KGK VI period. In addition, 60 plant samples from 20 different locations were analyzed to create the BASr baseline.

Proteomic analysis of amelogenin

The biological sex of a person is intrinsically linked to their gender identity and their role in society83. As such, knowing the sex of an individual offers some foundational information to explore issues of gender-patterned mobility and potential exchange of spouses in past societies. In most archaeological contexts, the absence of soft tissues implies that sex estimation has to rely on the sexually dimorphic characteristics of the human skeleton84. Following such techniques, the sex of adult individuals can be assessed with a confidence level superior to ninety-five per cent when several morphological traits are present85. Unfortunately, most human remains of the period are found fragmented and in a poor state of preservation. In addition, post-mortem manipulation of human remains and secondary burials were frequent during the Eneolithic, resulting in finds of isolated bones or teeth86. In these conditions, sex assessment can be challenging, if not impossible.

These challenges can be overcome through the use of biomolecular techniques. In effect, proteomic analysis, a method already used successfully in the field of archaeological research for the characterization of ancient food residues87,88,89 has recently been applied to profile a group of proteins known as amelogenins, which contribute to enamel maturation, and subsequently remain within the enamel matrix90. In humans, since the AMEL gene is found on the sex chromosomes, only AMELX will be found in female samples, while both AMELX and AMELY will be present in male samples91. This approach has been applied successfully to archaeological samples92,93,94 and has shown to be significantly more reliable than morphological approaches for sex assessment of fragmented materials95. In this study, we chose to apply an improved technique96 on seven individuals for which sex estimation was difficult or impossible: two adults whose post-cranial elements were either poorly preserved or absent, two isolated jaws, and a child.

Analytical methods

In order to perform the strontium isotope analyses, after having mechanically removed the outer layer of the tooth crown which may be contaminated with exogenous strontium97a small quantity (c. 20 mg) of enamel was collected. The samples were treated with 0.1 M acetic acid for 30 min and then rinsed three times with milliQ water95,98. For plant samples, 0.25 g of material was first microwave-digested in 4 mL 14 M HNO3 and 0.5 mL HF in a Milestone UltraWAVE99; 0.5 mL of H2O were added and evaporated to dryness100. Strontium extraction from the samples was performed following the protocol adapted from Snoeck et al.101 detailed in Gerritzen et al.102 and measured on a Nu Plasma 3 MC-ICP Mass Spectrometer (PD017 from Nu Instruments, Wrexham, UK) at the Vrije Universiteit Brussel (VUB). The Sr isotopes were measured by static multi-collection. Each analysis consisted of 60 ratio measurements (3 blocks of 20 cycles), resulting in a data collection duration for each individual sample of 12–13 min. All the Sr isotopes (84, 86, 87, 88) were measured, while the masses 85 (Rb) and 83 (Krypton) were simultaneously monitored, allowing for interference corrections on masses 84, 86 (Kr) and 87 (Rb). The Sr isotopic ratios were automatically normalized to ⁸⁷Sr/⁸⁶Sr = 0.1194 using an exponential law. Strontium concentrations were also obtained by comparing the intensity of 88Sr (in V) of the sample to that of the NBS987. Repeated measurements of NBS987, SRM1400 and SRM1515 returned ratios of 0.710254 ± 0.000017 (n = 12; 2SD), 0.713111 ± 0.000026 (n = 5; 2SD) and 0.713943 ± 0.000032 (n = 6; 2SD) respectively. There are consistent with the values reported in the literature (NBS987–0.710252 ± 0.000013103; SRM1400–0.713139 ± 0.000087104; SRM1515–0.713950 ± 0.000008105). Repeated measurements of SRM1400 for [Sr] returned a value of 251 ppm ± 3% RSD (n = 4) which is close to the certified value of 248ppm. All the results have been made available online on the open access and collaborative isotope database IsoArcH106,107.

For proteomics analysis, samples were processed essentially as described in Lugli et al.92. Teeth samples were cleaned in HPLC water. Enamel chunks weighing 10–50 mg were subjected to three washing cycles: (i) HPLC water (ultrasonic bath); (ii) 1 M HCl (5 min), (iii) HPLC water. To extract enamel proteins, specimens were soaked in 1 M HCl (1:20 w/v ratio) for 3 h in a glass vial at room temperature. Peptides were subjected to StageTip purification before mass spectrometric analysis. Offline StageTip purification was performed as described in Taverna et al.108. Briefly, ten µL of each sample were diluted to 200 µL in 0.1% TFA and purified by C18 Stage Tips. After sample loading and washing with 0.1% TFA, peptides were eluted by adding 10 µL of water: acetonitrile 1:1 to the StageTip. The eluate was diluted 15-fold in mobile phase A (see below) and injected for nanoLC-MS/MS analysis (2 µL).

NanoLC analysis was performed as described in Taverna et al.105. Chromatography was performed on an EasyLC 1200 instrument (Thermo Fisher Scientific) coupled to a Q-Exactive “classic” (Thermo Fisher Scientific). The nanoLC column was a pulled capillary, 0.075 × 130 mm (i.d. and column length, respectively), in-house packed with C18 silica particles (Dr. Maisch). Peptides were loaded directly on-column at 500 nL/min in mobile phase A (2% acetonitrile, 0.1% formic acid) and eluted at 230 nL/min by a linear gradient (from 10% B to 50% B in 20 min, with additional 5-min ramping to 100% B and 5 min of column regeneration at 100% B; mobile phase B was 80% acetonitrile, 0.1% formic acid v/v). Peptides were electrosprayed in positive ion mode using a spray voltage of 1800 V; MS acquisition was performed in mixed data-dependent acquisition/parallel reaction monitoring (DDA/PRM) mode. A full MS1 spectrum (resolution 35,000, scan range 350–1000 m/z) was followed by six data-dependent (top-6) MS2 scans using 6 s as dynamic exclusion. Eight consecutive precursors were then monitored in PRM mode (see Table 2). For all MS2 events, max. injection time was 60 ms, resolution was 17,500, isolation window was 1.6 m/z and collision energy was 25 in normalized units. The only function of DDA scans was to evaluate the possible presence of contaminants. Amelogenin analysis was based on PRM scans exclusively.

We specifically focused on peptides SM(ox)IRPPY (AMELY; [M + 2 H] + 2 440.2233 m/z) and SIRPPYPSY (AMELX; [M + 2 H] + 2 540.2796 m/z), previously demonstrated as abundant amelogenin peptides91,92,93. Additional peptides reported in Table 4 were also searched to corroborate the presence (or the absence) of AMELY. Four precursors were monitored for both AMELY and AMELX.

Table 4 Precursors monitored in PRM scans. Fragment ions used for producing extracted ion chromatograms (XICs) are also reported.
Full size table

PRM data were used to construct specific extracted ion chromatograms for each precursor monitored. Ion chromatograms were constructed using Xcalibur (Thermo Scientific) with a mass tolerance of 50 ppm.