Introduction

The 4th-3rd centuries BCE in Central Europe are marked by distinct developments in the archaeological evidence of the La Tène culture. This period of the so called ‘Celtic migrations’ (e.g. Bouzek 2009) was investigated for isotopic evidence of the historically postulated intensive mobility of people (Scheeres 2013, 2014). Even in situations where strontium and oxygen isotopic analysis delivered very ambiguous evidence for population mobility during the flat cemeteries period (4th-3rd centuries BCE), some influx of foreign groups to Bohemian territory can be documented in other ways, such as through the mobility of the personal objects at the beginning of the 4th century BCE (Danielisová et al. 2021b). This study aims to provide a general characterisation of the La Tène population of Prosmyky in the Lovosice area (northwestern Bohemia), with a focus on diet through analysis of stable carbon and nitrogen isotopes. Of particular interest are possible socio-economic and temporal diet-related differences that could be linked with the presumed developments in La Tène society and might provide more insight into the processes forming human societies during this transitory period.

The most notable change is the overall alteration of the burial rite. Richly furnished elite burials and flat cremation cemeteries, common in the preceding 6th-5th centuries BCE, disappear, and flat inhumation cemeteries become prevalent. They usually number around 10–30 graves and served only local communities. Larger cemeteries of over a hundred burials also appear, commonly denominated as ‘community’ sites shared by several settlements. The graves typically comprise either richly furnished female graves or those with only one or two pieces of bronze jewellery. Among the most distinctive are the burials of males with weapons (iron swords, spears and shields), in traditionalistic views usually considered as the ‘leading’ social group (Venclová et al. 2013; Danielisová et al. 2019). Male burials without weapons are generally more numerous than warrior graves, and apart from that, graves without any personal items or other grave goods occur as well.

Not only the burial rite, but also the settlement structure changed with the onset of the 4th century BCE, the latter becoming more decentralised (i.e. without evidence of fortified hillforts or agglomerations) and typically consisting of unenclosed sites of family to community size (Danielisová and Venclová 2019). In the course of the 3rd century BCE, more nucleated settlements start to appear in the ‘traditional settlement zone’ along the major watercourses and historical long-distance lines of communication, such as the Middle Danube area and the Amber Road corridors (Venclová and Militký 2014; Čižmář and Danielisová 2021). These open agglomerations became more distinctive during the LT B2-C phases (from the middle of the 3rd century to the beginning of the 2nd century BCE) and their emergence is accompanied by progress in technological approaches and socio-economic structure as well (Danielisová et al. 2021a). The increased homogenisation of the materials used for the production of personal jewellery indicates the stabilisation of local socio-economic conditions and in the 3rd century new elements in the material culture, technology, and overall social structure start to emerge. The trend appears to have originated from the territories southeast of Bohemia (Sankot 2014; Danielisová et al. 2021a). The most prominent examples of the new settlement category are the agglomerations along the eastern branch of the Amber Road. Even though in Bohemia there is no certain representative of this category known as yet, these developments may be expected here too, principally along the Labe (Elbe) river and in the areas adjacent to the foothills of the Krušné hory (Ore Mountains, Erzgebirge; Danielisová et al. 2021b). Indications, such as the distinctive local coin production, suggest that some sites had to exist somewhere within this densely populated settlement zone in the most fertile region of Bohemia. The Lovosice area has been hypothesised to contain an agglomeration (Salač 1990, 2011).

Stable isotope analysis is a common tool used in bioarchaeology for the examination of the lifestyles of past populations. It works on the principle that human tissues are formed from nutrients obtained in the diet and as such carry the “isotopic imprint” of dietary sources, modified by certain biochemical processes, i.e. the reactions involved in the nitrogen balance of the organism. The basic principles of the dietary analysis have been explained in great detail elsewhere (see e.g. Peterson and Fry 1987; Tykot 2004 or Katzenberg 2008; Lee-Thorp 2008 and the references therein). In short, carbon-13 is used to estimate the proportion of the C4 plants or, in combination with nitrogen, marine sources in the diet, as variations in carbon fixation and associated discrimination against 13C exist in the biosphere. A minor increase of about 1‰ in δ13C values can also be seen with each trophic level, but it is more pronounced (3–6‰) in δ15N (Bocherens and Drucker 2003; Hedges and Reynard 2007; O’Connell et al. 2012). Nitrogen is therefore mainly used for estimating the proportion of the animal protein intake in humans, although other factors might influence it as well. The bone collagen is commonly used for dietary analysis in bioarchaeology. Due to its turnover with bone remodelling, collagen provides isotopic information about the protein fraction of the diet over several years preceding the individual’s death, depending on the sampled bone (Fahy et al. 2017; Hedges et al. 2007).

The Prosmyky site

The site of Prosmyky is located by the Labe (Elbe) river in the town of Lovosice in the north-western part of the Czech Republic (GPS: 50°30’54” N, 14°5’36” E), as shown in Fig. 1. Archaeological excavations, which took place in 2005 (M. Půlpán, J. Blažek; unpublished) and in 2008–2009 (J. Blažek, E. Weisnerová; unpublished), revealed part of a large burial ground, containing 97 skeletal inhumations and 3 cremation burials, although the total number of burials is likely higher, as the neighbouring land plots have not been excavated. Prosmyky was in use during the LT B1-C1 phases, from approximately 380 to 200 BCE. It is a typical case of a La Tène period flat cemetery with skeletons deposited in supine positions and a north-south orientation. Inhumations were commonly accompanied by grave goods consisting mostly of personal objects (bronze jewellery) and iron weaponry, that can be used to ascertain chronological and to a certain extent also regional differences, using typological and relative-chronological methods. Some graves in Prosmyky revealed specific furnishings or were furnished with ceramic vessels that are not typical for Bohemia, and can be considered as special equipment of potentially foreign origin. The relative chronological assessment of the graves was carried out using the archaeological typology of the personal objects such as brooches, rings/jewellery, belts, and other items, along with an absolute chronological analysis using radiocarbon dating; this was performed on 27 samples that confirmed the LT B1-C1 dating of the burials. The cemetery included 7 warrior burials with swords, spears and shields in the grave; male burials without weaponry were also documented, even though the exact number is not known due to the poor preservation of the anthropological material. The richest female burials contained complete costume attire consisting of torcs, sets of brooches, bracelets, belts and anklets. A similar social structure could be observed for both sexes for the whole duration of the site. Female burials with fewer jewellery items or with no personal objects at all were documented as well, and the cemetery also contained numerous burials of infants and juvenile individuals.

Fig. 1
figure 1

Location of Prosmyky and other La Tène sites mentioned in text. ‘Swiss plateau’ and ‘Alps’ show approximate locations of several different sites

Full size image

Materials and methods

From graves with preserved skeletal remains, all the available adult individuals buried in Prosmyky were selected for isotope analysis, along with 3 subadults, numbering 55 samples in total. Due to the poor bone preservation a probable sex estimation was available only for 24 individuals, with 11 individuals designated as males or probable males and 13 as females or probable females; for the 31 remaining, morphological sex estimation was not possible. The sex of individuals was evaluated based on sex-diagnostic features of the pelvic bones (e.g., Brůžek 2002) and skull (Ferembach et al. 1979; Buikstra and Ubelaker 1994). The age at death was evaluated based on several methods and osteological indicators - the character of the articular surface of the sacro-iliac junction, the character of the pubic symphysis, attrition of the occlusal surfaces of the teeth, obliteration of the cranial sutures and the state of general wear of the skeleton (e.g., Buikstra and Ubelaker 1994; Stloukal et al. 1999; Schmitt et al. 2002; Schmitt 2005). An age at death estimation was not feasible for 5 individuals. The numbers of individuals with successful isotope analysis for each of the demographic groups are show in Table 1.

Table 1 Number of individuals in demographic categories
Full size table

Faunal reference was established by analysing bone collagen of 4 Bos taurus and 3 Ovis/Capra specimens, excavated in Prosmyky.

Due to the generally unsatisfactory state of preservation of human bones the isotope analysis of the bone hydroxyapatite was not carried out. Collagen was extracted as described in Longin (1971) as modified by Bocherens (1992), from approximately 250 mg of ground up bone, ultrasonically cleaned in deionised water before powdering. The bulk compact bone samples were taken from the diaphyses of the long bones, preferentially femora, that were presumed to have a higher chance of containing preserved collagen due to their robustness. Where these were unavailable, the mid-section of the costae was used instead. To minimise the risk of diagenetic contamination, the surface layer of the bone was mechanically abraded. In short, extraction followed subsequent rinses of the material in 1 M HCl (40 ml, 20 min), 0.125 M NaOH (40 ml, 20 h) and solubilizing at 100 °C in 0.01 M HCl (17 h), concluded by freeze-drying. The δ13C and δ15N ratios were measured using EA-IR mass spectrometry at Iso-Analytical Ltd. in Crewe, UK. The VPDB and AIR scales using IAEA-CH-6 and IAEA-N-1 inter-laboratory comparison standards were used to calibrate the stable carbon and nitrogen isotopic compositions. In-house standards IA-R068 (soy protein, δ13CV−PDB = -25.22‰, δ15NAIR = 0.99‰), IA-R038 (L-alanine, δ13CV−PDB = -24.99‰, δ15NAIR = -0.65‰), IA-R069 (tuna protein, δ13CV−PDB = -18.88‰, δ15NAIR = 11.60‰) and a mixture of IAEA-C7 (oxalic acid, δ13CV−PDB = -14.48‰) and IA-R046 (ammonium sulphate, δ15NAIR = 22.04‰) were used to monitor measurement uncertainty, which was less than 0.1‰ (1SD) for carbon and 0.2‰ for nitrogen. The routine analytical error for %C and %N was 2% in RSD (relative standard deviation) terms.

Results

Collagen proved to be generally well-preserved within the sample set, with only four samples failing to yield sufficiently preserved collagen (more than 1% of initial bone sample weight). These samples, including one warrior individual, were discarded and are not presented here. Overall, the collagen weight% of the analysed bones averaged 9.14 ± 3.91%. The remaining samples passed the collagen quality criteria (Ambrose 1990; van Klinken 1999) with an average nitrogen content of 15.84 ± 0.73% and carbon content of 42.95 ± 1.97%, and an average C: N ratio of 3.15 ± 0.04, ranging from 3.04 to 3.27.

The mean (± 1SD) faunal isotope ratios were -21.06 ± 0.41‰ and 6.60 ± 0.93‰ for δ13C and δ15N, respectively. As can be seen on Fig. 2, Bos and Ovis/Capra isotopic values did not differ noticeably but show a relatively wide spread of nitrogen ratios, which can be attributed to variability in the local environment (i.e. the pasture plots). Measured (mean ± 1SD) carbon isotope ratios for Bos and Ovis/Capra specimens were -21.05 ± 0.28‰ and -21.07 ± 0.54‰ and nitrogen ratios 6.43 ± 0.94‰ and 6.84 ± 0.86‰ respectively.

Fig. 2
figure 2

Isotope ratios from Prosmyky. ‘Warriors’ denotes individuals buried with items of iron weaponry

Full size image

The isotope ratios of the 51 measured human bone samples, as shown in Fig. 2, ranged in δ13C from -19.76 to -13.44‰, with an average of -18.67 ± 1.03‰ (1SD). Similarly, the δ15N ratios varied from 8.58 to 11.53‰, averaging 10.1 ± 0.65‰ (1SD). The spacing between the human and faunal averages was 2.39‰ for δ13C and 3.5‰ for δ15N. Samples PRO40 and PRO43 represent outliers in carbon, with notably less negative δ13C (-15.28 and -13.44‰). Demographic parameters are limited in their cases, with only age at death (30–50) being known for PRO40. Significant differences were observed in δ15N depending on the age of the individuals, as presented in Fig. 3. Individuals with an estimated skeletal age lower than 30 years (subadults excluded) have lower δ15N than those older than 30 years, as well as “intermediate” individuals, pooling around 30 years of age (ANOVA, p = 0.0053). No such difference is present in δ13C (Kruskal-Wallis, p = 0.6133). The intermediate group consists of individuals with age estimations that did not allow reliable assortment into other groups, e.g. aged 20–40 years. No statistically significant sex-based differences were noted in either δ13C (Welch t-test, p = 0.811) or δ15N (Welch t-test, p = 0.0517), even though males exhibit increased nitrogen values, with the statistical test close to the significance limit of p < 0.05. However, this is most likely caused by the limited number of sexed individuals, and the weapon burials comprising 1/3 of the males. These warrior individuals appear to have elevated δ15N (10.54 ± 0.56‰, mean ± 1SD) in comparison to adults buried without weapons (10.10 ± 0.63‰, mean ± 1SD), although no statistically significant difference was found (Welch t-test, p = 0.1239). The same applies to carbon isotopes (Kruskal-Wallis, p = 0.8733), even when the outlying samples are discarded (Welch t-test, p = 0.8616). On the other hand, a statistically significant (ANOVA, p < 0.0001) steady increase is present in relation to chronology, with the carbon outlier from grave 186 excluded (PRO43), as younger graves exhibit higher δ13C values, as shown in Fig. 3. The mean difference reaches 0.71‰ between the graves dated to the LT B1 and LT C1 periods. The δ15N appears unchanged (ANOVA, p = 0.164) throughout the duration of cemetery usage.

Fig. 3
figure 3

Boxplot age-based comparison of δ15N (A) and the chronological increase of δ13C (B) measured at Prosmyky

Full size image

Discussion

The results from the human bone collagen indicate a diet based mainly on C3 plants, with a variable input of C4 plants. In historical European contexts the expected C4 plant would be broomcorn millet, Panicum milliaceum. It is unlikely that millet was used as animal fodder, and it appears to have been exclusively for human consumption, as the animal carbon values do not display the enrichment seen in humans and the average spacing of the mean δ13C values of humans and fauna exceeds the trophic step increase of ca. 1–2‰. The average spacing in δ15N values between the herbivorous fauna and humans fits within the trophic step increase range (3–6‰, Bocherens and Drucker 2003; Hedges and Reynard 2007; O‘Connell et al. 2012) and suggests that meat and/or dairy consumption was important, albeit the degree of animal protein intake varied in humans. The difference from the animal δ15N mean ranges from 1.98‰ in the lowest human value to 4.93‰ in the highest. As shown in Fig. 4, the overall structure of the diet does not differ noticeably from other contemporary sites in the Czech Republic (Le Huray and Schutkowski 2005; Le Huray et al. 2006), but the δ15N values are generally higher than those reported from La Tène period Switzerland (Moghaddam et al. 2016, 2018). This difference may most likely be a result of distinct environmental baselines; the animal data are, however, limited in the Swiss region to allow for proper comparison.

Fig. 4
figure 4

Comparison of isotope data (mean ± 1SD) from Prosmyky and other La Tène sites from the Czech Republic (Le Huray and Schutkowski 2005; Le Huray et al. 2006) and Switzerland (Moghaddam et al. 2016, 2018). ‘Swiss plateau’ and ‘Alps’ show summary data from several different sites

Full size image

At Prosmyky, the chronological differences between the isotope data signify the gradual transformation of the population’s dietary structure, with δ13C steadily increasing from La Tène B1 to La Tène C1 by 0.71‰. As this change does not affect the nitrogen ratios, the overall intake of animal proteins most likely remained the same. The increase of δ13C values implies that the population changed some agricultural practices related to its subsistence. The simplest explanation would be the increased reliance on millet, as this C4 plant would shift the δ13C towards more positive numbers. Millet, however, did not become a staple food item at Prosmyky, as even the highest average δ13C of -18.4‰ from the LT C1 phase is around the commonly used borderline value of -18‰, which is considered to signify prominent C4 plant consumption in European contexts (Le Huray and Schutkowski 2005). Higher reliance on C4 plants has been observed, e.g. at the 3rd century BCE burial site at Verona (northern Italy, cf. Laffranchi et al. 2016), where the mean value for adults was reported as -15.3‰. A case of diachronic increase of δ13C, similar to Prosmyky, has been documented in the contemporary site of Münsingen in Switzerland (Moghaddam et al. 2016); the authors argue that climate change played a role in this isotopic shift. The same explanation could be applied here, as millet is a plant suited to a dryer and warmer environment (Moudrý 2005; Konvalina et al. 2007; Miller et al. 2016; Filipović et al. 2020; Lightfoot et al. 2015), but the environmental data to support this hypothesis are lacking. Nevertheless, as millet cultivation is common on historical sites from the late Bronze Age (Filipović et al. 2020; Hunt et al. 2008; Lightfoot et al. 2013), several individuals to whom millet must have been a distinctive part of the diet are present at Prosmyky, as can be seen in the two outliers with notably less negative δ13C. Unfortunately, the archaeological record (both were buried without any grave goods) and osteological material is in their cases severely limited, and offers no clue to the reasons behind their peculiar isotopic results.

In Central European burials, individuals with weapons have been repeatedly reported as a subgroup with a distinct dietary structure when compared to the rest of the population, displaying increased animal protein intake, although usually still within the observed population variabilities (Le Huray and Schutkowski 2005; Le Huray et al. 2006; Moghaddam et al. 2016, 2018). The same trend is followed at Prosmyky. The warriors fall well within the isotopic values of the population, but display a tendency to include higher nitrogen ratios, even when compared to the richly furnished burials of tentative females (based on the character of the burial equipment). This tendency is however not statistically significant, and a dietary distinction cannot be clearly made. The overall diet of the warrior subgroup seems rather to be slightly above average as concerns animal protein consumption.

Significant differences in δ15N were, however, detected in relation to age; the population of Prosmyky exhibits a notable increase of δ15N values in older individuals. This isotopic shift is apparent in individuals older than approximately 30 years of age, although the limitations of the estimations of the skeletal age at death as well as bone turnover speed present a problem for precise determination of this threshold. The presence of dietary differences related to age in adults is rather unusual in isotopic studies; even in different time periods the evidence is limited, and usually concerns subadults (Turner et al. 2007; Pearson et al. 2015; Tafuri et al. 2018; Martyn et al. 2018), with the exception of studies trying to assess the duration of breastfeeding and weaning (e.g., Fuller et al. 2006; Bourbou et al. 2013). Still, some cases have been reported in Iron Age Europe, e.g. in burials from Yarnton, Britain (Lightfoot et al. 2009). The character of those differences, however, does not correspond to Prosmyky, as mature individuals appear to be more variable in δ15N rather than exhibiting homogeneously increased values as reported for Bohemia. Furthermore, the subadults of Yarnton, except for infants, are also depleted in δ13C, which is not the case at Prosmyky (Lightfoot et al. 2009). Surprisingly, a similar tendency of older individuals to have higher δ15N has been, once again, also reported at Münsingen (Moghaddam et al. 2016), although it is statistically significant only with the non-parametric test and involves senile individuals in contrast to the ca. 30 years of age and older reported for Prosmyky. Determining the reasons behind the dietary differences is challenging. The prominent social status of the “elderly” might account for the traditional view of their potentially higher animal protein intake. However, the elevated nitrogen ratios do not appear to be linked with rich grave good assemblages, and the “elder status” by itself is usually not the only factor determining prominent social standing (Appleby 2010). Moreover, such circumstances are unfortunately practically undetectable by bioarchaeological methods alone. One possible explanation that could fit with the isotopic data may lie in the exclusion of subadults from the “extensive” meat consumption at e.g. feasts, which were likely an important aspect of Iron Age social culture, as reported in the historical accounts (Arnold 1999; Enayat 2014; Maini and Curci 2013). If meat consumption was limited for the subadults, the bones of individuals aged around 20–30 years could still contain a substantial part of the collagen synthetised prior to the supposed dietary change even with the generally expected faster bone turnover rates in younger individuals (Hedges et al. 2007). The prominent position of meat and its possible cultural significance could be reflected by food offerings in graves that are documented especially in the south-eastern part of the Central European contexts of the La Tène culture (e.g., Müller et al. 2008; Tankó and Tankó 2012). At the northern Italian Monterenzio Vecchio necropolis, the distribution of animal bones in graves showed that meat offerings were included in all graves regardless of sex or status; still, they were more plentiful in the cases of warrior males and generally lacking - with exceptions - in the subadult burials (Maini and Curci 2013). Even though the links between food offerings and actual diet are vague, the results would agree with the diet being “meatier” for warriors and restricted for subadults. On the other hand, the degree of restriction would be questionable, since subadults with abundant offerings were present. It is also debatable who was actually considered an “adult” in La Tène society. Furthermore, in Bohemia animal bone depositions in graves are only a rare occurrence (e.g., Mangel et al. 2019) and none were excavated in Prosmyky. It is therefore unclear to what extent the contemporary Northern Italian population is comparable to the Bohemian one, even though some historians presume speculative links between the two (Bouzek 2013). The manifestation in the case of the Italian burials could be a result of synergy with the Etruscans or other Italic tribes (Maini and Curci 2013; Sorrentino et al. 2018).

Manuring, as mentioned by Moghaddam et al. (2016), increases δ15N and could very well be the reason behind the isotopic data, especially as it was without doubt present in Iron Age agricultural production (Danielisová 2019). But to explain the age-based differences in diet, it would have to be assumed that there were at least two different agricultural regimes, whose products would be separated in their availability based on age. That appears unlikely, as La Tène settlements of this period were based mostly on smaller family or community-based sites even though in an otherwise intensively populated agricultural landscape (Danielisová and Venclová 2019).

It has been shown that famine or starvation can lead to an increase of δ15N values, as the starving individual catabolises their own tissues (Reitsema 2013; Walter et al. 2020). In human bioarchaeology, such an increase in δ15N can be seen especially when using analysis of incremental dentine (Beaumont and Montgomery 2016; Crowder et al. 2019; Walter et al. 2020). Bone, however, due to its slow turnover rate, lacks the temporal resolution seen in dentine and therefore gives an average dietary signal of several years, even decades (Hedges et al. 2007). Nutritional stress would need to be long-standing to be reflected in its isotopic values and even then the catabolic effect might not be manifested in the bulk bone nitrogen (Walter et al. 2020). The deficient skeletal preservation at Prosmyky does not allow for a reliable assessment of skeletal signs of malnourishment. Nevertheless, the elevated δ15N in older individuals could still be linked to limited food availability in the population. In the case of repeated crop failures or suboptimal harvest yields, individuals relying on plants or with limited access to animal food sources could be at higher risk of starvation and subsequent demise at a younger age. An increased intake of animal protein in the diet can be associated with a higher chance of survival to an older age (Prowse et al. 2005; Reitsema et al. 2016). As argued by Prowse et al. 2005; individuals able to obtain a more nutritious diet would survive to older ages either thanks to their better health status or their overall “better” living conditions preventing various life-threatening or -shortening stressors. Therefore, the noted higher animal protein intake in older individuals might not necessarily be a sign of prominent socio-dietary status itself, but rather a reason for their later age of death. Such food shortages and the subsequent threat of starvation could also be a fitting explanation for the gradual shift in agricultural production and dietary composition towards the more resilient millet, which also provides faster harvests, greater yields under unfavourable conditions and lasts longer when stored than common grains (Rachie 1975; Spurr 1983; Miller et al. 2016). It is worth mentioning that millet was considered famine food by the Romans (Spurr 1983). However, this narrative faces several problems. Prosmyky is situated in one of the most fertile regions of today’s Czech Republic, which was continuously inhabited from the Neolithic (Salač 2011), and no data are known that would support long-standing environmental influence as a driving factor of the agricultural shift. On top of that, the continuity of the local communities can further be seen lasting into the late La Tène and subsequently the Roman period. It is therefore unlikely that a long-term dietary crisis would be behind the change in the dietary habits of the local populations. On the contrary, increasing technological advances and new materials are visible in the geochemical analysis of the bronze jewellery of the latest graves from the site (Danielisová et al. 2021a), indicating more complex changes in the socio-economic sphere than just environmentally driven ones.

It must be stressed here that the temporal shift in δ13C and the age differences of δ15N might not have been connected at all. Just as the burial rite changed at the turn of the 4th century BCE, typical flat cemeteries with inhumations were gradually abandoned in favour of cremation burials in the Late La Tène period, tentatively taken as evidence of the more prominent influence of south-eastern territories. During the 2nd-1st centuries BCE, no archaeological evidence of regular cemeteries is known in most of Central Europe (Chytráček et al. 2021). Some archaeologists believe this transformation of the spiritual world of the La Tène population spread in a top-down manner, meaning from the elites to the rest of the society (Chytráček et al. 2021). Furthermore, in the following Late La Tène period new types of settlement appear in the environs of today’s Czech Republic, such as oppida, fortified hillforts, and large unenclosed lowland settlements, ranging in size up to several hundred inhabitants (Danielisová and Venclová 2019). Millet production can be used as a kind of “risk-management” tool, providing a stable food source in case of – amongst other things – increased food demand with growing settlements and populations (Zavodny et al. 2017). The increasing role of millet in the diet might therefore be a result of agricultural intensification (Miller et al. 2016); age-based differences in meat consumption, on the other hand, would represent another, separate cultural phenomenon. Food production can easily be a controlling social element in complex societies where, however, it implies large numbers of people completely removed from food production. In a situation where La Tène society remained primarily an agricultural one until its very end, and where there is no evidence of such large non-producing bodies (such as an army, clergy, craftsmen, etc.) resulting from a developed division of labour system, it would be premature to straightforwardly connect the (presumed) increase of millet cultivation solely with the appearance of agglomerations that are deeply rooted in the agricultural countryside. However, there is no doubt that La Tène society developed constantly from the beginning, and the noted temporal shift in δ13C could have been an accompanying feature of this development, even when settlement nucleation might not have been fully archaeologically evidenced in the Lovosice/Prosmyky area. Once again, drawing definite conclusions is hampered by the lack of sufficient data.

To further investigate the obtained results, isotopic data published by Le Huray and Schutkowski (2005) were revisited and statistically analysed for age-based differences. An analogous differentiation as at Prosmyky was applied to the populations of Radovesice I and II (due to the small sample number analysed together, with subadults over 7 years of age included) and Kutná Hora. Only individuals with known skeletal age estimations were used, and were separated into age categories as shown in Fig. 5. Surprisingly, a similar trend of increased δ15N of individuals older than 26 years was noticed at the Radovesice cemeteries (Kruskal-Wallis, p = 0.0037) but was absent at Kutná Hora (ANOVA, p = 0.547). Both sample sets are contemporary to Prosmyky with Radovesice being located in relatively close proximity (ca. 20 km). Kutná Hora, on the other hand, is situated some 110 km away in eastern Bohemia. Similarly, as visualised in Fig. 5, the trend was indicated but not statistically significant in the data published by Le Huray et al. (2006) from the sites of Tišice (Welch t-test, p = 0.4135), Ruzyně (ANOVA, p = 0.34) and Jinonice (Kruskal-Wallis, p = 0.4287). Those sites are located in what is now Prague or by the Labe (Elbe) north of Prague (Tišice), approximately half the distance from Lovosice to Kutná Hora. Unfortunately, it was not possible to explore the temporal changes in millet consumption in these datasets and the comparison with Prosmyky is not ideal, but the results still suggest that the factors influencing the noted dietary differences could have been at least to some degree based on or linked to individual regional developments.

Fig. 5
figure 5

Age-based comparison of δ15N applied to isotopic data from Le Huray and Schutkowski 2005 and Le Huray et al. 2006

Full size image

Conclusion

The isotope study applied to the population from Prosmyky showed a diet not notably different from other burial sites of the La Tène period Bohemia. It was based on C3 plants with a limited input of millets; this input, however, gradually increased during the chronological span of the cemetery, which provides valuable insight into the dynamics of millet consumption in Europe. It could also represent a factor accompanying various other developments of La Tène society either of local or exogenous origin. The higher animal protein intake of the warrior individuals reported from other Bohemian sites is indicated at Prosmyky as well, but significant differences in δ15N are present between the age-classes. It appears that individuals older than ca. 30 years had access to a diet with a higher proportion of these proteins, which is a rather unusual result in dietary studies. It is clear that the available data is too limited to allow definite conclusions and explanations for the isotopic results presented in this study, and much more work is needed to properly investigate the processes behind the developments in Central Europe during this period, but the population of Prosmyky has allowed an unusual viewpoint of the issue and could possibly offer a direction for future studies.