Sequencing the Southern Iberian Late Neolithic hypogeum cemetery of La Beleña through radiocarbon dating and Bayesian modeling

Abstract

This study aims to determine the chronological sequence of the collective burials in the hypogea of the prehistoric cemetery of La Beleña (Cabra, Córdoba) through Bayesian analyses of ¹⁴C dates obtained from human remains. The data from this site are not only key to grasping the phenomenon of the introduction and spread of hypogea throughout the western Mediterranean, but to gain insight into multi-stage funerary practices during the Late Neolithic/Chalcolithic. The dataset comprises ¹⁴C dates of 71 of the 79 individuals placed in five of La Beleña’s six hypogea. The findings suggest: (i) La Beleña is one of the oldest assemblages of hypogea in Iberia, (ii) that this type of collective burial spread rapidly throughout the western Mediterranean area, (iii) that La Beleña is marked by two main phases of funerary activity interspersed by brief burial surges, (iv) funerary intensity at La Beleña increased between cal BC 3400–2900 (2σ), and (v) the cemetery saw a very brief surge of burials potentially related to a catastrophic event. The results of this analysis thus shed light on the little-known chronological sequence of prehistoric hypogea or rock-cut tombs in Iberia, their spread, and their relation to other Late Neolithic collective burials in western Europe.

Introduction

Collective burials are characteristic of western Europe’s Late Neolithic (Chambon et al 2018). This practice appears to have its origins in the late 5th millennium BC in Brittany from where it expanded in the early 4th millennium BC to Britain and Ireland, northern Europe and Iberia (Balsera et al 2015; Scarre et al 2003; Schulz Paulsson 2019; Whittle et al 2011). Collective burials are defined as single features containing multiple corpses that contrary to mass burials were deposited over periods of time ranging from several decades to hundreds of years. The dead were often accompanied by a variety of goods such as pottery, stone tools and ornaments, items potentially reflecting differences of social status (Chambon et al 2018). The reasons behind the phenomenon remain unclear. Certain scholars argue they reflect the emergence of complex societies and their need to reinforce and display social identity and alliances through funerary practices (Boulestin 2016; Sánchez-Quinto et al 2019). Others suggest they relate to a change in religious beliefs, with an emphasis on community and collective afterlife (Whittle 2017).

The tradition of collective burials intensified from around cal BC 3500 to 2500 throughout western Europe as well among different central Mediterranean islands (Aranda Jiménez et al 2022; Chambon et al 2018; Thompson et al 2020). They are associated with megalithic features such as barrows, cairns, as well as natural and artificial caves (rock-cut tombs/hypogea) (Chambon et al 2018; Pardo-Gordó and Carvalho 2020; Scarre 2010). However, grasping the nexus between their different architectural features is complex due to multiple factors such as chronology, local traditions and geomorphological conditions (Guilaine 2021; Sauzade 2021). Furthermore, the tradition of collective burials persisted, albeit with changes, into the Early Bronze Age. A notably difference is a decrease in the number of individuals in the tombs and that the dead tended to receive a more individualised treatment suggesting a shift in this direction by society (Aranda Jiménez et al 2017; Chambon et al 2018).

Hypogea are among the least known types of prehistoric collective burials. Their origin and spread throughout western Europe are still the subject of debate due to the scarcity of high-precision ¹⁴C dates (Guilaine 2015; Sauzade 2021). Rock-cut tombs in the central Mediterranean such as at Cuccuru S’Arriu in Sardinia (Robin et al 2021) and Scintilia in Sicily (Gullì and Terrasi 2020) date to approximately cal BC 4500. These types of features subsequently became widespread throughout these two islands around cal BC 4000–3400 (Robin et al 2021). Similar features are recorded in the Iberian Peninsula and in Provence in the first half of the 4th millennium BC (Carvalho 2014; Guilaine 2015; Sauzade 2021). However, how this practice spread throughout continental western Europe remains obscure. The current radiometric data (cal BC 3700–3550) suggest the Portuguese cemeteries of Quinta da Abóboda, Barrancas I and Sobreira de Cima as the earliest hypogea in the Iberian Peninsula (Valera 2013, 2020a). Furthermore, these collective burials became more common around cal BC 3400–2900 during southern Iberia’s transition from the Late Neolithic to the Chalcolithic (Guilaine 2015; Lillios et al 2014). The end of the tradition of rock-cut tomb cemeteries is likewise unclear as certain hypogea, subsequent to a hiatus, saw reuse by Bell Beaker groups (Valera et al 2014).

As other prehistoric collective burials, hypogeum cemeteries are highly dynamic spaces characterised over time by multiple depositional and postdepositional events yielding complex palimpsests of funerary practices (Chambon et al 2018). The disturbance of their primary and secondary burials in the form of trampling, intentional fragmentation, bone removal, translocation and secondary disposal have yielded intricate, multi-chronological archaeological contexts complicating their sequencing (Aranda Jiménez et al 2020a; Lillios et al 2014; Valera 2013). In spite of these complications, fine-grained chronometric analyses of prehistoric collective burials have enabled gaining an unprecedented understanding of their dynamics (Scarre 2010). These approaches consist of interdisciplinary analyses based on three main strategies: 1) radiocarbon (¹⁴C) datings of human remains so as to associate the isotopic events reflected by the radiometric measurements (that is, the moment of death) to the depositional event (when the individual was placed in the tomb); 2) applying sampling criteria based on dating a minimum number of individuals (MNI) whose selection stems from osteological analyses; and 3) designing Bayesian models to estimate the outset, duration and end of the events (Aranda Jiménez et al 2017, 2021; Bayliss 2009; Blank et al 2020).

The exponential growth of direct accelerator mass spectrometry (AMS) datings of human remains has revitalised this research as it offers solid foundations to delve into the lifespan and abandonment of these features. The new data also shed light on the links of the manipulations of human remains and other funerary practices such as the opening and closing of these spaces (Aranda Jiménez et al 2022). The new approaches likewise allow renewing work on older tombs that at their moment of their excavation did not benefit from these techniques (Aranda Jiménez et al 2017, 2020b; 2021; Schulting et al 2017). Recent work on prehistoric megalithic collective burials in the Iberian Peninsula clearly demonstrates the impact of both ¹⁴C dates and Bayesian modelings on their interpretation (Aranda Jiménez et al 2017, 2020a, 2021, 2022; García Sanjuán et al 2018; Linares-Catela and Vera-Rodríguez 2021; Santa Cruz del Barrio et al 2020; Valera 2020a; Valera et al 2019). These methods have led to unravelling the use, closure, and reuse of these monuments as well clarifying their multi-stage burial practices. These new techniques have likewise contributed to gaining a finer grasp of the chronological framework of the Iberian phenomenon as they offer clues to the origin, spread and interaction of the different types of megalithic monuments and hypogea from the architectural standpoint.

The radiometric analyses and Bayesian modeling of the hypogea of La Beleña specifically offer a chronological framework shedding light on its multi-stage funerary practices as well as the spread of this phenomenon throughout the western Mediterranean landscape. Furthermore, this study counts on one of Europe’s largest ¹⁴C datasets which help to fill the research gap concerning the spread of prehistoric hypogea or rock-cut tombs throughout Iberia and the ties of these features to other Late Neolithic collective burials of western Europe.

Archaeological background

The cemetery of La Beleña was initially discovered in 1973 during agricultural work which led to the collapse of the dome-shaped ceiling one of its tombs (Hypogeum 1). The exact location of this hypogeum is still uncertain due to modifications made to the area following its discovery. However, locals who visited the site in the 1970s indicated that the hypogeum was situated just a few meters south of Hypogeum 2. The opening in the ground revealed human bones and grave goods that were removed by the local archaeologist and deposited in the Museum of Archaeology of Cabra. A second well-preserved hypogeum was later discovered by an agricultural worker in 2015. This find led to a research project aimed at identifying the site’s burial practices, chronological framework and osteobiography of the dead. Since then, five new rock-cut tombs or hypogea were identified and excavated (Figure 1b) (Camalich et al 2023).

Figure 1. Map of the Iberian Peninsula with the location of the cemetery of La Beleña (left) and aerial view of the five hypogea (right).

The hypogea of Beleña contain primary and secondary human depositions and grave goods. Their burial chambers are hemispherical connected to west-facing corridors cut into a compact marl substrate. Although the ceiling of three collapsed during recent agricultural work (20th century), their contents suffered no damage (Figure 2). Moreover, the archaeological fieldwork identified evidence of intentional sealing in Late Neolithic/Chalcolithic ending their use (Camalich et al 2023).

Figure 2. View of Hypogeum 2 containing displaced commingled human remains in secondary position. The cranial and infracranial elements were arranged in separate groups.

Material and methods

Material

La Beleña comprises four primary inhumations linked to Hypogea 3 and 5. Otherwise all hypogea contain commingled human remains in secondary position. The primary inhumations, determined by the articulated arrangement of their bones, consist of individuals buried soon after their death. This means their ¹⁴C dates closely align with the time when their remains were deposited. The other dispersed commingled bones, in turn, stem from secondary multi-stage burial practices meaning there is no direct chronological link between their ¹⁴C dates and the moment/s of their disposal.

The sampling strategy intended to grasp the dynamics of the funerary activity was intended to analyse the largest number of human remains (Santana 2020). The approach, based on the minimum number of individuals, was to ensure that the same individual was not sampled twice. The osteological collection was systematically analysed to establish the paleodemographic profile and to identify other features, including paleopathological disorders and trauma. Teeth analyses yielded an MNI of 79 individuals for the five tombs, a total that slightly surpasses that of the cranial and infracranial analyses, respectively 62 and 67 (Table 1). As several individuals were only represented by teeth, not all were subjected to ¹⁴C analyses as certain were reserved for ancient DNA analyses. Ultimately ¹⁴C analyses were carried out on a total 71 individuals, broken down into 52 bone fragments and 19 teeth (Tables 1–2).

Table 1. La Beleña: minimum number of individuals by hypogeum based on teeth and skeletal region analyses. The final column indicates the number retained for AMS ¹⁴C dating

Table 2. La Beleña: results of the ¹⁴C datings and isotope analyses

Methods

Radiocarbon analyses

The ¹⁴C dates were obtained by Accelerator Mass Spectrometry (AMS) at the Beta Analytic Testing Laboratory (USA) and at the Centro Nacional de Aceleradores (Spain). The results were then subjected to Bayesian modeling. OxCal software version 4.4 (Bronk Ramsey 2009a) served to design the calibrations, plots and modelings. The 2σ levels of confidence, recommended by Millard (2014), served as the base when discussing the ¹⁴C measurements. 1σ probability intervals were added to the figures and tables. The modeled dates are rounded to the nearest five years since the modeled results vary from run to run. The ¹⁴C dates of individuals linked to terrestrial diets were corrected by the international IntCal20 atmospheric calibration curve (Reimer et al 2020).

Stable isotope analyses

Stable isotope analyses (δ¹³C and δ¹⁵N) were likewise carried out to evaluate marine/freshwater intake. The consumption of significant amounts of these types of resources can produce what is known as the reservoir effect, skewing the precision of the datings, that is, yielding earlier measurements (Cook et al 2015). Stable isotope analyses were thus carried out by elemental analysis – isotope ratio mass spectrometry (EA-IRMS) from samples of bone and dentine collagen (50 bone and 19 dentine samples). The isotopic value analyses were undertaken by Beta Analytic Inc. (USA), ETH Zurich Laboratory of Ion Beam Physics (CH) and the IsoAnalytical Laboratory (UK) (Table 2). The three laboratories adhered to rigorous quality control guidelines applying international reference standards. Differences between laboratories can influence the comparability of isotope values from one lab to another (Pestle et al 2014). Nevertheless, collagen isotope readings tend to be consistent across various labs, especially when identical preparation techniques are employed. The isotopic values were reported according to the international V-PDB standards for δ¹³C and atmospheric air (AIR) for δ¹⁵N. The quality of the collagen extracted from the bones was controlled by international criteria yielding > 1% and a C:N ratio of 2.9–3.6 (van Klinken 1999).

Bayesian modeling

Bayesian models yield rigorous estimates of the start, end and duration of events (Bayliss 2015; Bayliss et al 2007a). The current study assumes that the human remains were deposited soon after their death. Long-term curation and/or retrieval of human bones can also be detected through these types of models in the form of outliers (Schulting et al 2017).

Several models were drawn up with the OxCal software based on the different interpretations of the ¹⁴C dataset. A Single Uniform Phase Model (Bronk Ramsey 2009a) determined whether all the burials were devoid of discontinuity and align with a single chronological phase. This therefore equates with a uniform distribution model based on the hypothesis that all the events are likely to occur at any time at the start and end of the potential phases. Hence calibrated models were created for each hypogeum using OxCal tools (Oxcal’s Sequence, Phase, Boundary, Duration, Interval, and Difference commands; Bronk Ramsey 2009a, 2009b). The efficacy of each model is signified by the agreement indices, specifically A_model and A_overall, both of which should not fall below a threshold of 60%. A_model furnishes a comprehensive metric evaluating the congruence of the entire model, while A_overall operates as a composite function of the individual agreement indices for each specific date (Bronk Ramsey 2009a).

This statistical study also identified outliers among each hypogeum as well as for the overall sample by applying OxCal’s Outlier Model (Bronk Ramsey 2009b). The level of contemporaneity between the different ¹⁴C measurements was tested by Chi square tests (Ward and Wilson 1978) which assessed the degree of overlapping among the ranges of probability. Non-parametric statistical methods based on kernel density estimation (KDE) (Bronk Ramsey 2017) were then applied as exploratory devices to characterise the potential phases of La Beleña’s burials. This is a widely used frequentist method with no formal priors for the distribution. Its advantage when compared to that of the sum function is that it reduces the noise from the calibration procedure allowing KDE distribution to serve as a prior in the Bayesian model (Bronk Ramsey 2017). Furthermore, this method is more precise when lacking reliable information as to the stratigraphic relationship of the samples (Bronk Ramsey 2017). This is the case of La Beleña’s hypogea marked by little reliable stratigraphic or spatial information due to the commingling and disarticulation of the human remains during multi-stage burial practices (Blank et al 2020). This was specifically carried out with KDE_Plot OxCal tools (Bronk Ramsey 2017).

Results

Isotopes analyses

Bone collagen samples was successfully extracted from 71 humans (52 bones and 19 teeth). The carbon and nitrogen content of both bone and dentine collagen ranged respectively from 20.7% to 45.48% (40.8% ± 3.2) and 7.4 to 16.46% (14.6% ± 1.2) (Table 2). The atomic C/N ratio ranged from 3.1 to 3.6 (3.2 mean) yielding acceptable atomic carbon/nitrogen ratios (van Klinken 1999; DeNiro 1985). The δ¹³C values were between –20.49‰ and –18.4‰ (–19.5‰ ± 0.5, n = 71). Seven individuals yielded δ¹³C values above –19‰. The δ¹⁵N values, in turn, ranged between +6.98‰ and +11.62‰ (8.7‰ ± 1.1‰, n = 71). Three samples revealed δ¹⁵N values >10‰ (Table 2). These results suggest that the diet was predominately terrestrial as δ¹³C values were more negative than –18‰ and δ¹⁵N values falling below 2‰ (Schulting et al 2024). Such isotopic signatures are consistent with those found in other Late Neolithic Iberian populations with terrestrial diets (Cubas et al 2019). It is noteworthy to mention that La Beleña is located at a distance of approximately 84 km from the Mediterranean Sea (Figure 1).

The general chronology of the Cemetery of La Beleña

Analyses of the overall sample (n = 71) yielded four early ¹⁴C dates from Hypogeum 5: Beta-593540, Beta-593546, Beta-593548 and Beta-593547 (Table 2; Figure S1). However, when considering only the ¹⁴C dates of Hypogeum 5, the results suggest that three ¹⁴C dates pertain to an initial phase of burial activity, while the sample Beta-593540 appears to be an earlier outlier (Figure S2). Interestingly, the associated stable isotopic values for this individual do not indicate any reservoir effect on the ¹⁴C date (δ¹³C = –19.9‰ and δ¹⁵N = +9.4‰).

The KDE model depicts a first phase of funerary activity around cal BC 3700 (Figure 3) followed by a short gap preceding another second phase of activity. The first phase comprises the three oldest ¹⁴C dates of Hypogeum 5 (Beta-593546, Beta-593548 and Beta-593547) ranging from cal BC 3700–3450. The second cluster, after the short hiatus, comprises the group of datings from cal BC 3400 to 2900. The end of this second range most likely corresponds to the abandonment of the cemetery (Table 3). The KDE analyses therefore indicate a first phase linked to Hypogeum 5 and a second intensive burial phase between cal BC 3400 to 2900 comprising all the tombs. Thus, the Bayesian single model of the overall sample, when excluding the outlier, suggests that burials at Beleña began in cal BC 3595–3540 (2σ; median = 3560) and ended in cal BC 2915-2865 (2σ; median = 2895) separated by an interval of about 670 years (635–712 years, 2σ) (Table 3, Figure S3).

Figure 3. La Beleña: KDE modeling of the ¹⁴C datings. The graph depicts the two main burial phases and the outlier Beta-593540. The start boundaries are indicated in green, and the end boundaries in red. The outlier 14C date is highlighted in blue. Upper brackets below each age estimate represent the 68.2% and the lower brackets the 95.4% confidence interval.

Table 3. Multi-phase Bayesian ranges pertaining to the estimated start, span, interval and end of burials at La Beleña

Bayesian analyses of the two funerary phases

A model illustrating the overlap of the two phases was drafted to further delve into the two main clusters of ¹⁴C dates yielded by the KDE model (Figure 3). As the ¹⁴C dates of Phase 1 pass the chi-square test (T’ = 4.5 [T(5 % = 6.0)]) (Ward and Wilson 1978), it is possible to assume that the individuals likely died over a relatively short timespan (one or two generations) (Figure S4). Phase 1 thus begins around cal BC 3975–3540 and ends between cal BC 3645 and 3390 (2σ; median = 3710) within an interval of 0–515 (2σ; median = 150 years) (Table 3). The range between the end of Phase 1 and the outset of Phase 2 is estimated at between 40 and 415 years (2σ; median = 250 years) (Table 3). Phase 2 began around cal BC 3380–3175 (2σ; median = 3305) and ended between cal BC 2995–2870 (2σ; median = 2905) with an estimated interval of 185 and 500 years (2σ; median = 400 years) (Table 3).

Bayesian analyses of each hypogeum

Hypogeum 2

The ¹⁴C dataset of Hypogeum 2 (n = 21) did not pass the chi-square test of contemporaneity (T’ = 96.27 [T(5 % = 31.4)]) (Ward and Wilson 1978). This suggests that this ¹⁴C dataset likely reflects multiple depositional events of human remains. Furthermore, no hiatus was discerned as in the KDE model (Figure 4; Figure S5). The Bayesian model indicates that the burial activity began in cal BC 3370–3130 (2σ; median = 3250) and ended in cal BC 3000–2835 (2σ; median = 2890) with an interval of 130–500 years (2σ; median = 360 years) (Table 4).

Figure 4. Bayesian chronological ranges depicting the start and end of each hypogeum combined with individual KDE plots. The start boundaries are indicated in green, and the end boundaries in red. The outlier ¹⁴C date is highlighted in blue. Upper brackets below each age estimate represent the 68.2% and the lower brackets the 95.4% confidence interval.

Table 4. Single phase ranges for the estimated start, span, interval and end of each hypogeum

Hypogeum 3

The KDE model (n = 7) and the test of contemporaneity suggest that all the individuals in Hypogeum 3 were buried over a period of one or two generations (T’ = 10.5 [T(5 % = 12.6)]) (Ward and Wilson 1978) (Figure 4; Figure S6). The burial practice began between cal BC 3420–3100 (2σ; median = 3290) and ended between cal BC 3310–2900 (2σ; median = 3070) with an interval of about 0–460 years (2σ, median = 150) (Table 4).

Hypogeum 4

The nine ¹⁴C dates of Hypogeum 4 also did not pass the test of statistical contemporaneity (T’ = 21.5 [T(5 % = 15.5)]) (Ward and Wilson 1978) (Figure 4; Figure S7). The model tends to offer younger dates for samples such as Beta-465699 and Beta-593538 which coincide with a plateau in the calibration curve. These factors may in fact affect the accuracy of the start of the depositions in this tomb. The model places the start of the burial activity sometime between cal BC 3160–2940 (2σ; median = 2940) and the end around cal BC 3015-2855 (2σ; median = 3065). The interval is estimated to between 0–270 years (2σ; median = 120 years) (Table 4).

Hypogeum 5

The results of the ¹⁴C analyses of Hypogeum 5 did not pass the chi square test of contemporaneity (T’ = 1105.1 [T(5 % = 23.7)]). This remained the case even when the outlier was excluded (T’ = 520.3 [T(5 % = 22.5)]) (Ward and Wilson 1978). Moreover, the KDE model highlights the outlier and two distinct sets of datings (Figure 4). This led to the design of two Bayesian models, one of a single phase (Figure S8) and a second of two phases (Figure S9), excluding the outlier (Beta-593540). The first indicates an outset of burial activity between cal BC 3840–3550 (2σ; median = 3690) and an end between cal BC 2980–2730 (2σ; median = 2890) within an interval of 650–1060 years (2σ; median = 820 years) (Table 4).

The Bayesian model of the two phases of Hypogeum 5, excluding the outlier, reveals a first set of datings (Beta-593546, Beta-593547 and Beta-593548) corresponding to Phase 1 and the remaining to Phase 2. Those of Phase 1 pass the statistical test of contemporaneity (T’ = 4.5 [T(5 % = 6.0)]) (Ward and Wilson, 1978). Those of Phase 2, on the contrary, do not (T’ = 84.6 [T(5 % = 18.3)]) (Ward and Wilson 1978). The Bayesian model suggests that Phase 1 began between cal BC 3980–3540 (2σ; median = 3670) and ended between cal BC 3650–3260 (2σ; median = 3560) within an interval of 0–650 years (2σ, median = 110 years) (Table 5). Phase 2, in turn, began between cal BC 3430–3120 (2σ; median = 3240) and ended between cal BC 3000–2810 (2σ; median = 2910), with an interval of 150–590 years (2σ; median = 330 years). The difference between phases 1 and 2 was estimated at between 40–500 years (2σ; median = 305 years) (Table 5, Figure 5).

Table 5. Results of the Bayesian multi-phase model of Hypogeum 5 indicating the ranges for its estimated start, span, interval and end

Figure 5. Multi-phase Bayesian chronological ranges indicating the start and end of each phase and KDE plots of the overall distribution of dated events within each collective burial. Hypogeum 6 also includes high posterior densities of the oldest and youngest ¹⁴C dates of the dataset (Table 6). The start boundaries are indicated in green, and the end boundaries in red. The outlier ¹⁴C date is highlighted in blue. Upper brackets below each age estimate represent the 68.2% and the lower brackets the 95.4% confidence interval.

Hypogeum 6

The assemblage of ¹⁴C dates (n = 19) of Hypogeum 6 also did not pass the chi-square test of contemporaneity (T’ = 31.1 [T(5 % = 28.9)]) (Ward and Wilson 1978). This may be due to the results of samples Beta-5333965 and Beta-5333959, yielding respectively the oldest and youngest dates. It is noteworthy that the KDE models depict a first peak based on sample Beta-5333965, a short gap, and then, the probabilistic distributions of the other ¹⁴C results (Figs. 4–5). In any case, the Bayesian single model places the start of burial activity in this hypogeum between cal BC 3440–3100 (2σ; median = 3360), the end between cal BC 3310–2930 (2σ; median = 3070) and the interval between 0–400 years (2σ; median = 270 years) (Table 4; Figure S10).

The results indicate that 17 ¹⁴C dates (89% of those of Hypogeum 6) yielded similar ranges and thus passing the chi-square test of contemporaneity (T’ = 6.2 [T(5 % = 26.3)]) (Ward and Wilson 1978). They therefore suggest that the 17 individuals in its chamber died over a very short period of time. A Bayesian model was thus designed to delve into this cluster as a phase while maintaining Beta-533965 and Beta-533959 within the general model (Table 6; Figure 5). The results of Beta-533965 suggest the range of cal BC 3500–3330 (2σ; median = 3370). Phase 1, in turn, began between cal BC 3350–3170 (2σ; median = 3370) and ended sometime around cal BC 3320–3085 (2σ; median = 3250) with an interval of 0–180 years (2σ; median = 40 years) (Table 6), whereas the modeled range of Beta-5333959 is cal BC 3090–2910 (2σ; median = 2985) (Figure S11). Therefore, the model reinforces the notion that these 17 individuals met their death in a brief timeframe.

Table 6. Results of the Bayesian model for Hypogeum 6 indicating the ranges of the start, span, interval and end of its main phase of burial activity. The list includes the ranges of the oldest and youngest datings as well as the gaps between the main phase of burial activity, the end of the oldest (difference 1) and start of the youngest (difference 2)

Discussion

The chronological sequence of the hypogea of La Beleña

Maximising the number of AMS ¹⁴C dates (71 of 79 individuals, 90%) was a deliberate choice as they facilitate the understanding the prehistoric cemetery’s different depositional events. Applying the fine-grained radiometric protocol enabled garnering notions as to patterns of burial activity with a precision corresponding to few human generations (Blank et al 2020; Scarre 2010; Wysocki et al 2013). Moreover, combining the data with the Bayesian modelings also increased the precision of the sequence offering a better understanding of the succession of the burial activity. Thus, the cemetery of La Beleña was inaugurated by hollowing out Hypogeum 5 and the deposit in it of the first human remains around cal BC 3700 (start of Phase 1). However, the findings include a ¹⁴C date on a cranium from the same hypogeum that is distinctly earlier than the primary dataset of La Beleña. This date is identified as an outlier because it does not correspond with the remaining ¹⁴C dates associated with the site’s sequential deposition history. A viable hypothesis for this inconsistency is that the cranium might be a bone relic utilised in the funerary practices of La Beleña. Bone relics are curated human remains that are subjected to intricate processes of dissemination, circulation, and re-disposal. Although the employment of bone relics in the funerary contexts of European Late Prehistory is relatively infrequent, it is not exceptional, becoming more common during the Bronze Age (Bradley 1998; Brück 2006; Esparza-Arroyo et al 2018; Fowler 2004). Such relics often play a crucial role in shaping identities and preserving social memory (Borić 2003; Lillios 1999; Weiss-Krejci 2011). Fundamentally, relics can be interpreted as tangible manifestations of memory, serving as a bridge between past and present, and between the realms of the living and the deceased (Walsham 2010). In the context of La Beleña, it is essential to note that this bone relic was found in the earliest hypogeum of the cemetery. Furthermore, the multi-stage burial practice at this site affords significant consideration to cranial remains, leading to the formation of skull caches within the burial space, as observed in hypogea 2, 4, and 6 (Figure 2; Camalich et al 2023). Therefore, it is possible that this cranium was involved in ritualistic practices intended to establish a connection between the deceased and their forebears by incorporating ancestral remains.

The initial surge of burial activity in La Beleña lasting around 40 years ended around cal BC 3550 (first phase of Hypogeum 5). Phase 2 saw the placing of new human remains around cal BC 3300 after a hiatus of ca. 250 years. It likewise coincided with another surge of burials in the form of the hollowing out of Hypogea 2, 3 and 6, and the reuse of Hypogeum 5. Individuals then continued to be deposited in Hypogea 2, 3 and 5, but not in Hypogeum 6. Ultimately, Hypogeum 4 was opened around cal BC 3065 cal while all the collective tombs, excepting Hypogeum 3, received new dead. The results therefore suggest a scaled construction of hypogea over time with new chambers were only inaugurated during more intense funerary activity. Both the archaeological evidence and the Bayesian models also reveal that all the tombs, except Hypogeum 3, were closed at about the same time (cal BC 2900, 2σ). The cemetery of La Beleña, undisturbed until its recent discovery in the 20^th century, thus serves as a reliable indicator of the cycles of funerary activity linked to prehistoric collective tombs.

The statistical models indicate that Hypogea 2 and 5 saw use for hundreds of years by several generations. The funerary activity of Hypogeum 5 extended for about 730 years, a timespan significantly greater than that of Hypogeum 2 (ca. 330 years). On the contrary, the lifespans of Hypogeum 3 (ca. 145 years), Hypogeum 4 (ca. 100 years) and Hypogeum 6 (ca. 90 years) were relatively brief. The five collective tombs can, based on their duration, be classified into two groups: 1) Hypogea 2 and 5 followed by 2) Hypogea 3, 4 and 6. However, although group 1 saw a longer lifespan, the Bayesian analyses suggest that human remains arrived in the tombs during short, intense phases.

Hypogeum 5 saw two short-lived phases separated by a long hiatus (ca. 280 years). Phase 1 represents the initiation of funerary activity at La Beleña while Phase 2 is thought to correspond to a very intense period featuring the construction of Hypogea 2, 3 and 6. This surge is likewise observed at other Iberian prehistoric collective burials such as Alto Reinoso (Alt et al 2016), Cardim 6 (Valera et al 2019), Montelirio (García Sanjuán et al 2018), Los Zumacales (Santa Cruz del Barrio et al 2020), Panoría (Aranda Jiménez et al 2022) or Perdigoes 4 (Valera 2020b), among others. Other European megalithic assemblages such as the British monuments of Ascott-under-Wychwood (Bayliss et al 2007b) and Hazleton North (Meadows et al 2007), and the Swiss dolmen of Oberbipp (Ramstein et al 2022), also follow analogous patterns.

Up-to-date archaeological evidence suggests that this area saw limited settlement prior to the time frame associated with La Beleña (Martín Socas et al 2018). Yet, the significant funerary activity at this cemetery suggests emerging patterns of human occupation in the area during the latter half of the 4th millennium BC. While evidence for contemporary settlements in the immediate vicinity remains scarce, burial sites often serve as primary indicators of human activity (Camalich et al 2020; Martín Socas et al 2018). It is noteworthy that new sites contemporaneous with La Beleña have been discovered in the region, including Torreparedones (Martínez Sánchez et al 2014) and the Cave of Los Cuarenta (Vera Rodríguez et al 2014). Additional examples from more distant settings include the Polideportivo de Martos-La Alberquilla (Cámara et al 2010), Marroquíes Bajos (Portero et al 2010; Rodríguez Ariza 2011), and Loma de las Eras del Alcázar (Nocete et al 2009).

Other studies focusing on western and southwestern Iberia have likewise identified a rapid population growth between cal BC 3350 and 2900 based on Summed Probability Distributions (SPD) of ¹⁴C dates (Balsera et al 2015; Lillios et al 2014; Pardo-Gordó and Carvalho 2020; Sweeney et al 2022). This growth coincides with the transition from the Late Neolithic to the Chalcolithic in southern Iberia, a process characterized by an intensification of food production (Cubas et al 2019), a development of permanent settlements (Díaz-del-Río 2023; Valera et al 2017) and an emergence of more complex social systems (Castro et al 1996; Díaz-del-Río 2023; García Sanjuán and Murillo-Barroso 2013; Gilman 1987a, 1987b). Palaeogenomic research also suggests biological exchanges between southern Iberia and northern Africa around cal BC 3000 (Fregel et al 2018). The dental non-metric trait analyses of southwestern Iberian populations of the Late Neolithic/Chalcolithic also reinforce the notion of arrivals from northwestern Africa and/or the eastern Mediterranean (Irish et al 2017). These biological findings are reinforced by discoveries of African ivory at Iberian Late Neolithic/Early Chalcolithic sites (Schuhmacher et al 2009). Therefore, the intensity of funerary activity at La Beleña coincides with a period in southern Iberia experiencing intense social transformations resulting in significant demographic growth.

The burst of funerary activity

The chi-square tests and the medians of the modeled distributions indicate that the earliest collective tomb of the cemetery, Hypogeum 5, experienced a brief rise in funerary activity (0–110, 1σ; 0–145, 2σ), followed by a hiatus of about 280 years. A similar pattern can be observed in Hypogeum 6 where 17 individuals likely died within a very short timespan between cal BC 3350–3170 and 3320–3090 (2σ), with an interval of 0–55 years (1σ) or 0–140 years (2σ). A similar short-lived increase of burials has been observed among other European collective tombs, such as Apeldoorn–Wieselse Weg and Garderen-Bergsham in the Netherlands (Bourgeois and Fontijn 2015), Knowth in Ireland (Schulting et al 2017), and West Kennet (Bayliss et al 2007c) and Wayland’s Smithy I in England (Whittle et al 2007).

The sudden increase in burials within Hypogeum 6 may suggest that the corpses were deposited around the same time. However, due to the commingled and secondary nature of this burial, no stratigraphic relationship between them could be determined, leaving their synchronicity unresolved. Additionally, attempts to use ¹⁴C dates to establish their synchronicity are limited by the precision of the technique, including issues with measurement errors, calibration curves, and statistical tools. This challenge also pertains to the ¹⁴C dates of prehistoric mass graves in Spain (Alt et al 2020; Fernández-Crespo et al 2018), Germany (Meyer et al 2015) and Poland (Schroeder et al 2019) as their probability distributions did not reflect simultaneous interments due to the aforementioned issues. Indeed, the distributions of the 17 ¹⁴C dates of La Beleña in fact reveal more restricted ranges than those of the mass graves. The systematic osteological analysis of the human remains did not yield any evidence of inter-personal or palaeopathological disorders in of La Beleña. Nevertheless, the absence of infectious disease-associated pathologies does not rule out the possibility of such diseases being responsible for the sudden and intense burial activity within Hypogeum 6. It is noteworthy that palaeogenomic evidence points to high-mortality epidemics among European Late Prehistory populations (Andrades Valtueña et al 2022; Rascovan et al 2019; Swali et al 2023). Indeed, the impact of Yersinia pestis can explain the decline of population during the Late Neolithic in northern Europe between cal BC 3000 and 2900 (Blank et al 2020; Feeser et al 2019; Sjögren et al 2019). Therefore, one cannot rule out that infectious diseases were behind the brief and intense burial activity of Hypogeum 6.

Conclusions

The modeled chronology suggests La Beleña to be one of the oldest rock-cut tomb or hypogeum cemeteries of the Iberian Peninsula. In fact, the first human depositions of Hypogeum 5 coincide with the earliest Portuguese hypogea of Quinta da Abóboda, Barrancas I and Sobreira de Cima (Valera 2013, 2020a, 2020b). This type of funerary activity thus rapidly spread throughout southern Iberia. A similar rapid expansion of this tradition is observed in other regions of western Europe such as Provence where hypogea began to appear at around the same time (Sauzade 2021). The reasons and mechanisms behind this spread remain obscure. It does however coincide with more intense human mobility and long-distance exchanges observed from cal BC 3500 onwards (Díaz-del-Río 2023).

Burial activity at La Beleña then intensified around cal BC 3400–2900 (2σ). During this period, compelling evidence suggests a notable demographic growth stemming from agriculture intensification, population aggregation and the arrival of new groups from northwester Africa. The results of this study also suggest a brief surge of burials potentially related to catastrophic events such as an epidemic (Blank et al 2020). However, the limitations of the ¹⁴C dating methods prevent delving deeper into this hypothesis. Furthermore, the modeled chronology of La Beleña reinforces the existence of multi-stage funerary practices (primary deposition, anthropogenic manipulations, secondary depositions, etc.) not only observed in these hypogea but among other types of collective tombs throughout western Europe such as barrows, cairns, and natural caves.

While further research is needed to refine the details of these notions, the sequence of the cemetery of La Beleña currently provides evidence of a rapid expansion of hypogea throughout southern Iberia. Yet the connections between the different areas of Western Europe characterised by these hypogea remain unclear. The influence of long-distance exchange and human migration on this funerary tradition is unexplored yet. However, these burial features and practices undoubtedly played a role in the social transformations that occurred during the transition from the Late Neolithic to the Chalcolithic. This study hence reinforces the idea that the tradition of multi-stage burials such as those observed among the hypogea of La Beleña also affected other types of European megalithic collective tombs.