Early Neolithic pastoral land use at Alsónyék-Bátaszék, Hungary (Starčevo culture): New insights from stable isotope ratios Authors: Magdalena Blanz, Marie Balasse, Delphine Frémondeau, Erika Gál, Anett Osztás, Anna Zs. Biller, Éva Á. Nyerges, Denis Fiorillo, Eszter Bánffy, Maria Ivanova Published: 2023 ABSTRACT The earliest introduction of livestock (cattle, goats, sheep, pigs) into the Carpathian Basin was an important step towards farming expansion into continental Europe. This spread beyond the environments of the southern Balkans was accompanied by a reduction in the spectrum of cultivated crops, changes in the relative representation of different domestic animals, and, most likely, adaptations of husbandry practices. How the earliest farmers in the Carpathian Basin kept their domestic stock is still understudied. We explored early animal management and land use strategies at the Starčevo settlement at Alsónyék-Bátaszék, Hungary (Early Neolithic, ca. 5800–5600 cal BC). Settled at the intersection of wide alluvial plains, waterlogged meadows and marshes to the east, and forested hills to the west, early farmers at Alsónyék had a wide variety of options for nourishing their livestock. We performed stable isotope ratio analysis of bone collagen (n = 99; δ13C, δ15N) and tooth enamel (nteeth = 28, sequentially sampled for δ13C and δ18O) from wild and domestic animals to locate them in the landscape and investigate herding practices on a seasonal scale. The bone collagen isotope ratios mostly indicate feeding in open environments. However, results from the sequential analysis of cattle and sheep enamel suggest diverse dietary strategies for winters, including consumption of forest resources, consumption of summer hay and grazing in an open environment. Most pigs appear to have had herbivorous diets, but several individuals likely supplemented their diet with animal protein. Stable isotope ratio results from the Lengyel phase at Alsónyék (ca. 4800–4300 cal BC) suggest more access to animal protein for pigs, and feeding in more open areas by wild boar, red deer and cattle compared to the Starčevo phase. This study’s results demonstrate considerable variability in early animal husbandry practices at Alsónyék. Introduction Domesticated animals and plants, initially originating from Southwest Asia and domesticated in the 10th-9th millennium BC [1], were introduced by Neolithic farmers to Thessaly and Aegean Macedonia in the early 7th millennium BC. The spread of agriculture to inland Europe took a decisive step at the turn of the 7th to the 6th millennium BC when settlers from the Starčevo-Çris-Körös cultural complex pioneered the northern Balkans and the Southern Carpathian Basin (modern northern Bulgaria, Serbia, Romania and Hungary; [2–4]). This northward diffusion of farming was accompanied by significant changes in the spectrum of plants cultivated and the species composition of animals kept [5–8]. In the sub-Mediterranean southern Balkans (modern-day Northern Greece and southern Bulgaria), a broad spectrum of crops was cultivated [8], which appears to have been the result of a strategy of diversification. Correlated with differences in environments and climatic conditions, expansion into the northern Balkans and the Great Hungarian Plain was generally accompanied by a reduction in the spectrum of crops cultivated, and the proportion of sheep, goats, and pigs diminished while cattle became dominant in many faunal assemblages (reviewed in [9]). Early Neolithic farmers in the Great Hungarian Plain settled preferentially on flat surfaces on hydromorphic meadow soils near freshwater [10]. Living in an environment of marshland islets and forested hills likely encouraged a large degree of adaptation by these early farmers [11]. Further investigations are needed to describe in more detail how these farming systems have adapted to the advantages and constraints of the landscape, including in terms of the respective–and not necessarily exclusive–spatial locations of crops and livestock at the site scale, as well as on a seasonal scale. This study focuses on animal husbandry at Alsónyék-Bátaszék in southern Hungary (Fig 1; Lat. 46° 12’ N, Long. 18° 42’ E), which is particularly suited to such investigations of the place of animals in the landscape due to being located at an intersection of different ecosystems. Using stable isotope ratios (δ13C, δ18O, δ15N) of animal teeth and bones, the aim of this study is to research in what environments cattle and sheep (the dominant species) were herded and how pigs–rare but present–were kept in the Early Neolithic, to understand how herders utilised Alsónyék’s wider surrounding environment. Analysis of sequential tooth enamel samples also allows investigations of diets on a seasonal scale, through which it is possible to examine the use of woodlands and wetlands as potential seasonal food sources, as well as the proximity of pigs to dwellings. To construct a framework for the range of possibilities in animal husbandry practices, references are also made to ethnographic and historical accounts from Hungary from the 18th to 20th centuries. This work contributes to more wide-ranging research on understanding the environmental, biological and socio-cultural factors in the initial dispersal and adaptation of farming systems across Europe. Map showing the location of Alsónyék and other sites mentioned in this article. Map data: Natural Earth II from www.NaturalEarthData.com (public domain). Alsónyék: Description of the site, environment, and subsistence practices Alsónyék subsite 5603/1 (see Fig 38 in [12]), where most of the Starčevo features at Alsónyék were found, was occupied by people from the Starčevo cultural complex in the first half of the 6th millennium cal BC (beginning ca. 5800–5730 cal BC, and ending 5575–5505 cal BC, both 95% probability [13]). It appears as a substantial, clustered settlement [14]. Later occupation of the site included Neolithic communities from the Linearbandkeramik (LBK), Sopot and Lengyel cultures (Table 1). The village lies in the Carpathian Basin in the southwestern part of Tolna Sárköz in Southeast Transdanubia (see Fig 1) at the transition of hilly margins to marshy lowlands [15]. 10.1371/journal.pone.0295769.t001Table 1Sites mentioned in the text with their associated cultural groups and phases as relevant to this study.Site nameCultural groupChronological phaseDating cal BCReferenceAlsónyékStarčevoEarly Neolithicca. 5800–5600this studyLBKMiddle Neolithicca. 5500–5000[25]SopotMiddle/Late Neolithicca. 5100–4800[25]LengyelLate Neolithicca. 4800–4300[25]BalatonszárszóEarly LBKMiddle Neolithicca. 5450–5250[26]Vinča-Belo BrdoVinča-Pločnik IILate Neolithicca. 4850–4600[27]Borduşani-PopinăGumelniţa A2Late Chalcolithicca. 4500–4250[28]EcsegfalvaKörösEarly Neolithicca. 5800–5600[29]Endrőd 119KörösEarly Neolithicca. 5800–5630[30]FüzesabonyALPMiddle Neolithicca. 5550–5000[26]Halai-Early Neolithicca. 6400–5800[31]Hârşova-tellGumelniţa A2Chalcolithicca. 4350–4000[28]Kouphovouno-Middle Neolithicca. 5800–5400[31]Lánycsók-ÉgettmalomStarčevoEarly Neolithicca. 5800–5600[32]Măgura—Boldul lui Moş IvčnusStarčevo-Criş IEarly Neolithicca. 6000–5800[33]Makriyalos II-Late Neolithic IIca. 4950–4500[31]StublineVinča-Pločnik IILate Neolithicca. 4850–4600[27]Vităneşti-MăguriceGumelniţa A2Chalcolithicca. 4450–4150[28]ZemunicaImpressed PotteryEarly Neolithicca. 6000–5800[34]References are to relevant studies on faunal stable isotope ratios (δ13C and δ15N). ALP = Alföld Linear Pottery, LBK = Linearbandkeramik. Sites mentioned in the text with their associated cultural groups and phases as relevant to this study. References are to relevant studies on faunal stable isotope ratios (δ13C and δ15N). ALP = Alföld Linear Pottery, LBK = Linearbandkeramik. The Danube is 16 km to the east of Alsónyék-Bátaszék (hereafter: Alsónyék) today, and, together with its tributary Sárvíz, it shaped the landscape considerably [15]. Former Danube channels left behind oxbow lakes and wide alluvial floodplains, dominated by (seasonally) waterlogged meadows and marshes, likely frequently inundated, and island-like plateaus above the floodplain [16, 17]. Gallery woods were likely present along rivers and oxbow lakes, and groves may have grown in the floodplains [18]. The settlement site of Alsónyék itself appears to have been outside the extensive flooding area [16], but compared to later groups, the Starčevo community at Alsónyék occupied the lower and wetter parts of the area [19]. Although crop cultivation to the east of the site was likely possible on the elevated islets dotted throughout the landscape, it would have been at high risk of harvest-loss due to flooding, likely making the cultivation of arable land to the west of the site more favourable [16]. To the west of Alsónyék, the Danubian floodplains are supplanted by skeletal soils and forested hills (Szekszárd Hills, 273 m; [14, 16]), which also provided arable land on the hilltops [16]. An open oak woodland would likely have been present which may have also included pine, elm, hazel, lime, beech and hornbeam [20]. Wild fauna during the Starčevo phase includes animal species consistent with both forested and open wetland environments, e.g. wildcats (Felis silvestris), beavers (Castor fiber) and multiple duck species (Anatinae) [21–23]. The faunal assemblage from the Starčevo phase at subsite 5603/1 in the southeastern part of Alsónyék is dominated by domestic animals (88%), with caprines (sheep and goats; for scientific names see Tables 2 and 3) and cattle being predominant, and only minor amounts of pigs (3%) and dogs (Table 2; data given as number of identifiable specimens (NISP) from [23]). Most cattle have been slaughtered as juveniles and adults, consistent with a focus on dairying, while pigs were mostly slaughtered as juveniles [23]. Hunting was not a dominant subsistence activity, with only ca. 12% of the identifiable mammal faunal assemblage being wild animals, mostly red deer and wild boar [23]. Freshwater fish remains were also present at the site (1.4% of NISPfaunal remains), including wels catfish, northern pike and carp [20, 21]. 10.1371/journal.pone.0295769.t002Table 2Mammal faunal spectrum at Alsónyék subsite 5603/1 from the Starčevo phase (identifiable mammals only).SpeciesCommon nameNISP% of NISPidentifiable mammal Ovis aries/Capra hircus * sheep/goat*468842.5 Bos taurus cattle463342.0 Sus domesticus pig3313.0 Canis familiaris dog460.4 Cervus elaphus red deer5655.1 Sus scrofa wild boar5254.8 Capreolus capreolus roe deer1601.5other (identifiable)780.7 Total 11026 100 *Around six sheep per goat; data from [23]. Mammal faunal spectrum at Alsónyék subsite 5603/1 from the Starčevo phase (identifiable mammals only). *Around six sheep per goat; data from [23]. 10.1371/journal.pone.0295769.t003Table 3Bone collagen δ13C and δ15N stable isotope ratios from the Starčevo phase at Alsónyék, subsite 5603/1.Speciescommon namenδ13C (‰)δ15N (‰)Min.MedianMax.Min.MedianMax. Bos taurus cattle10−21.55−20.71−19.444.965.417.07 Ovis aries sheep18−21.62−20.76−19.954.416.827.45 Sus domesticus pig13−22.36−20.43−19.136.056.899.16 Sus scrofa wild boar19−22.11−21.08−19.893.956.998.59 Canis familiaris dog6−22.54−21.27−20.726.058.829.43 Cervus elaphus red deer10−22.55−21.46−19.794.276.637.76 Capreolus capreolus roe deer10−23.72−21.20−20.145.476.367.14 Vulpes vulpes fox2−20.35−19.64−18.937.707.797.87 Abramis brama bream2−24.37−23.69−23.018.008.639.25 Cyprinus carpio carp3−27.18−27.15−25.976.868.378.64 Esox lucius northern pike2−22.11−21.41−20.709.979.989.98 Sander lucioperca zander3−21.77−20.15−19.139.479.8110.18 Silurus glanis wels catfish1−20.64−20.64−20.648.248.248.24For additional information see Table A in S1 File; sample numbers exclude samples that failed the quality criteria. Bone collagen δ13C and δ15N stable isotope ratios from the Starčevo phase at Alsónyék, subsite 5603/1. For additional information see Table A in S1 File; sample numbers exclude samples that failed the quality criteria. The recovered crops from the Starčevo phase at Alsónyék (einkorn, emmer and barley [24]) indicate that at this time a mixed farming economy was likely present, and people were somewhat locally bound due to the requirements of crop cultivation. Livestock herding could have taken place in the forested hills to the west of the site, in the wetlands and waterlogged humid meadows in the east along the river Sárvíz (pigs in particular; [16]), or on cropland after harvest. Stable isotope ratios for palaeodietary reconstruction Bone collagen Bone collagen stable carbon (δ13C) and nitrogen (δ15N) isotope ratios reflect those of consumed dietary protein (and other dietary components to a lesser extent) in the last years of an individual’s life with a systematic diet-collagen offset (ca. 5–6‰ for δ13C, and ca. 3–5‰ for δ15N; [35–40]). Therefore, bone collagen δ13C and δ15N values can be used to determine what food groups were commonly consumed. Consumer δ15N values are frequently used as indicators of trophic level and can be used to infer e.g. whether pig diets were likely supplemented with animal protein [28, 41–43], or consisted of fertilised crops [44], both elevating δ15N values. In younger individuals, milk consumption puts the offspring at a higher trophic level than the mother leading to higher δ15N values. However, when growth rates are high (e.g. in adolescence), δ15N values tend to be lower [45]. This introduces some uncertainties in interpretations of δ15N values of young animals. Hungary’s terrestrial vegetation is dominated by C3-plants, with only minor amounts of C4 plants being present in dry grasslands following recent colonisations [46], so that C4 plants were likely not contributing to Early Neolithic diets at Alsónyék to a significant extent. Forest grazing/browsing tends to lead to lower consumer δ13C values due to the canopy effect [47, 48]. Ground-level undergrowth was found to be depleted by 2–5‰ in dense forest compared to plants and trees in open environments [47, 49]. Bone collagen δ13C values from modern forest-dwelling red and roe deer from dense deciduous forests in Dourdan (France) and Białowieża (Poland) were on average −23.7‰, with all except one below −22.5‰ (n = 29; corrected for the fossil fuel effect; [48]). Values approaching these thresholds are therefore considered to likely reflect a significant dietary contribution of forest resources (particularly ground-level undergrowth), especially in the case of herbivores. However, not all animals occupying forests necessarily have low δ13C values, possibly in part due to feeding outside of the forest [50] and feeding on fruits (e.g. acorns) which are higher in δ13C, and can lead to higher δ13C values than expected for forest-dwelling animals, particularly in the case of suids [51]. Conversely, low δ13C values in suids may also occur due to consumption of freshwater fish [28]. Since water stress has been shown to elevate δ13C values [52], and heavy watering leads to around 1‰ lower δ13C values [53–55], it has been argued that freshwater wetlands may be expected to have lower δ13C values than C3 vegetation from drier areas [49, 56], lowering consumer δ13C values. However, compared to e.g. the canopy effect, this is likely only a minor effect. Since δ13C and δ15N values differ due to a large variety of different factors, it is important to establish site-specific baselines for different diets. To gain comparative references of the typical δ13C and δ15N values for the diets of the domesticated animals under study (cattle, pig, sheep), bones of wild fauna like wild boar, red and roe deer (primarily forest-dwelling), freshwater fish (potentially consumed by pigs), as well as dogs (consumers of higher trophic level foods) excavated from the same site and phase are also analysed. Sequential analysis of tooth enamel Sequential analysis of stable carbon and oxygen isotope ratios in tooth enamel gives access to dietary information on a seasonal scale. Bioapatite precipitates in oxygen isotopic equilibrium with body water [57], linked, through the ingestion of surface water, to local annual precipitation [58, 59], whose δ18O values vary seasonally with air temperature in continental Europe [60]. Additionally, animal behavior and physiology in response to seasonal changes in temperature and air humidity also affect body water oxygen isotope composition [61]. All factors combine to create a seasonal signal in tooth enamel δ18O values. Sequential sampling permits the retrieval of a one-year record from the third molar of sheep and cattle [62], or by combining measurements from the first and second incisors in suids [51], while the canines of male suids may provide a pluriannual record [51]. The stable carbon isotope ratios in bioapatite reflect those of the entire diet [35] as opposed to collagen, whose primary carbon source is proteins [39, 63]. A 14.1‰ isotope enrichment (ε) has been shown between diet and enamel bioapatite δ13C values in a variety of large ruminant mammals [64], while this enrichment is 13.3‰ in experimental pigs fed C3 diets [65]. Most C3 plants have δ13C values varying from −29‰ to −25‰ around an average value of −27‰ in open areas [66], or −25.5‰ in preindustrial times after correction for the fossil fuel effect [67]. This should lead to an average value of −11.8‰ in sheep and cattle enamel bioapatite, or −12.5‰ in pig enamel bioapatite–although a significant fruit component in suid diets [68, 69] would also elevate this value; the δ13C values measured in predominantly open areas and under continental climate are globally comprised between −13‰ and −9‰ in cattle and sheep tooth enamel [33, 70]. By contrast, animals dwelling in closed forests have lower bone collagen δ13C values (below −22.5‰, [48]), so that enamel bioapatite δ13C values should tend towards –13,8‰ in large ruminants (applying a 5% spacing between collagen and diet [39] and a 14.1‰ isotope enrichment between diet and enamel bioapatite [64]), or towards −14.6‰ in pigs (applying a 13.3‰ isotope enrichment between diet and enamel bioapatite [65]) although a significant contribution of forest fruits would elevate this value. We consider values approaching these thresholds as reflecting a significant contribution of forest resources to animal diet. Seasonal variations are expected in plant δ13C values, in response to seasonal changes in the environmental factors affecting the stomatal aperture and therefore the carbon isotope discrimination during photosynthesis, and also possibly due to changes in plant physiology during the growing season. The highest δ13C values are expected in the summer when the air temperature is the highest and air humidity and ground water are the lowest [71–73]. Consequently, δ18O and δ13C values are expected to follow the same pattern of seasonal variation. Indeed, in modern sheep grazing on the same pasture throughout the year, the δ18O and δ13C sequences in enamel were shown to vary in phase or very close to it [74] with amplitudes of variation of 1 to 3‰ in δ13C [75]. Deviations from this pattern may result from the contribution of different food sources during the year due to foddering or mobility between areas where environmental factors affect plants differently. Materials and methods Bone collagen stable isotope ratio analysis All bone and tooth samples in this study were excavated from Alsónyék subsite 5603/1, and date to the Early Neolithic, Starčevo phase (ca. 5800–5600 cal BC; [13]; stored at the Wosinsky Mór County Museum, Szekszárd). Around 2 g cortical bone samples of long bones were taken from domestic and wild terrestrial and aquatic fauna (n = 107; Table 3). Differentiation between wild boar and domestic pig was done according to size, following [76]. In the case of the mammals, the same part of the bone from the same side of the body (within each species group) was preferentially sampled, to prevent sampling the same individual twice. To enable this sampling strategy, in some cases, juvenile and subadult individuals were also sampled (see Table A in S1 File). This needs to be taken into account when interpreting the results as it introduces uncertainty when comparing between animals of differing ages, e.g. due to different nitrogen balances (see above). In the case of pigs, most samples were from younger individuals because pigs tend to be slaughtered as juveniles/subadults and few adult pig bones were available to sample. Previous studies have argued for including younger livestock in stable isotope ratio analyses (e.g. [77]) since this enables a more complete insight into husbandry practices without a bias toward adults. Bone samples were cleaned by abrading the outer surface using a tungsten carbide drill bit. Collagen was extracted from 200–230 mg ground bone using 1 M HCl, followed by 0.125 M NaOH, following a modified Longin (1971) method [78] described in [79], but with the gelatinisation step at 70 °C instead of 100 °C. In the case of fish bones, the NaOH step was shortened to initially 15 minutes. If the solution coloured during this time, it was exchanged for fresh 0.125 M NaOH solution for another 15-minute immersion (following [80]). Using an elemental analyser (EA; Thermo Flash 2000) interfaced with an isotope ratio mass spectrometer (IRMS, Thermo DeltaVAdvantage), coupled δ13C and δ15N measurements of 320-380 µg bone collagen were conducted. Within each run, multiple replicates of a secondary alanine standard were included (calibrated to primary standards IAEA-600 for δ13C, and IAEA-USGS25, IAEA-N-1 and IAEA-N-2 for δ15N). The alanine standards gave mean values of −21.47±0.09‰ for δ13C (mean±σ; expected value: −22.16‰ with reference to VPDB) and 0.70±0.13% for δ15N (expected value: +0.59‰ with reference to AIR), and 40.07±1.46% for C content (expected value: 40.44%) and 15.56±0.09% for N content (expected value: 15.72%) over the course of all measurements (n = 23), and were used to correct the measured data. Quality criteria to ensure data robustness were chosen such that results were rejected if the bone collagen yield was less than 1%, or if the collagen had a C/N (molar) ratio outside of 2.9–3.45, or if the C content was less than 13%, or the N content less than 4.8% (following suggestions in [81–83]). Isotopic niche spaces were modelled as Bayesian ellipses with the R package SIBER [84]. Tooth enamel sequential stable isotope ratio analysis The study material includes teeth from cattle (11 upper third molars-M3), sheep (9 lower M3), pig (1 lower M3), red deer (2 lower M3) and wild boar (1 upper canine, 1 upper M3, 1 lower M3, 2 lower incisors from the same individual; Table C in S2 File). The enamel surface was cleaned with a tungsten carbide drill bit. Teeth were sequentially sampled using a diamond-coated drill bit on the lingual side of the anterior lobe for the cattle M3; on the vestibular side of the middle lobe for the sheep and red deer M3; on the lingual side of the anterior cusp for the suid molars, and on the labial side for the suid incisors and the maxillary canine. The sampling procedure for the suid teeth was illustrated in [51]. These enamel powders were pre-treated to eliminate diagenetic carbonates (0.1 M acetic acid for 4 h at room temperature, 0.1 ml/mg). Pre-treated enamel samples weighing ~600 µg were reacted with 100% phosphoric acid at 70°C in individual vessels in an automated cryogenic distillation system (Kiel IV device), interfaced with a DeltaVAdvantage isotope ratio mass spectrometer. The analytical precision for each run, estimated from 5 to 8 analyses of our laboratory carbonate standard (Marbre LM, calibrated to the NBS-19 international standard) was always less than or equal to 0.05‰ for δ13C values and 0.04‰ for δ18O values (both with reference to VPDB). For each run, the Marbre LM gave a mean δ13C value comprised between 2.09‰ and 2.20‰ (expected value 2.13‰) and a mean δ18O value comprised between −2.02‰ and −1.96‰ (expected value −1.83‰). The δ18O values for sheep and cattle tooth enamel were previously published in [85, 86] respectively. In the sheep and cattle molars and in the wild boar canine, showing full annual cycles, the phase shift between the δ18O and δ13C sequences was determined using a sinusoidal model approach after [74]; see also Table H in S2 File. Results Bone collagen stable isotope ratios During collagen extraction, four fish bone samples dissolved completely and could not be analysed, and the measured results from one sheep and four fish were excluded because they did not meet the quality criteria (see Table A in S1 File, where the complete set of data is also listed). Table 3 and Figs 2 and 3 show the δ13C and δ15N results for the samples that passed the quality criteria. Stable carbon (δ13C) and nitrogen (δ15N) isotope ratio results for bone collagen from the Starčevo phase at Alsónyék, subsite 5603/1. The 90% prediction ellipses were modelled excluding an outlier dog and two pig datapoints (see S3 File). Stable carbon (δ13C) and nitrogen (δ15N) isotope ratio results for mammal bone collagen from the Starčevo phase at Alsónyék, subsite 5603/1. This figure shows the same data as in Fig 2, but visualised as boxplots. The line in the boxplot indicates the median, and the cross the mean. Whisker boundaries were chosen as 1.5 × interquartile range. Fish and foxes were excluded due to low sample numbers. The results for most species overlap to a large extent (Figs 2 and 3). A one-way analysis of variance (ANOVA) was conducted to compare δ13C values between pigs, wild boar, dogs, red deer, roe deer, cattle and sheep (F(6,79) = 2.53, p = 0.027). It revealed differences in the mean δ13C values between species, but a post-hoc Tukey test showed no significant differences. A second one-way ANOVA (F(6,79) = 5.81, p = 0.00005) followed by a Tukey test indicated differences in mean δ15N values between pigs and cattle, and between dogs and wild boar, red deer, roe deer, cattle and sheep. Domestic pigs had a smaller range of δ15N values than wild boar (3.1‰ compared to 4.6‰; Fig 3, right). There was no statistically significant difference in δ15N or δ13C between juvenile (n = 6) and subadult pigs (n = 6), and no other age-based differences were identified. Red deer and roe deer yielded on average the lowest δ13C values among terrestrial mammals (Fig 3), the lowest collagen δ13C value being measured in a roe deer (−23.7‰, ALSCap9), followed by three red deer with δ13C values between −23.0‰ and −22.0‰, which are close to and below the threshold value suggesting a significant contribution of forest resources to the animals’ diets. The sampled freshwater fish exhibited a wide breadth of δ15N and δ13C values, with differences visible between dietary groups: benthic omnivores like carp and bream had the lowest δ13C values (−27.2‰ to −23.0‰) of all measured samples, whereas piscivorous fish (northern pike, wels catfish and zander) had the highest δ15N values (8.2‰ to 10.2‰). Tooth enamel sequential stable isotope ratios The results from the sequential analysis of tooth enamel are shown in Figs 4 and 5, Table 4, and Tables D-H in the S2 File (including phase shift modeling). Sheep enamel bioapatite δ13C values vary overall between −13.8‰ and −10.0‰. All sheep have recorded seasonal changes in their diet δ13C values (Fig 4). The amplitude of intra-tooth variation varies between 1.5‰ (ALS Ovis4) and 3.5‰ (ALS Ovis6). Different patterns of variations are observed. Most sheep show a sinusoidal variation in δ13C values in phase with the seasonal changes in δ18O values (phase shift comprised between 331 and 368°, Table H in S2 File; Patterns A and B in Fig 4). Among those, three sheep (ALS Ovis2, Ovis6 and Ovis9) have lower winter δ13C values (−13.8‰ to −13.6‰) tending towards the threshold indicating significant consumption of forest resources (Figs 4 and 5, Pattern B). ALS Ovis3 and ALS Ovis5 deviate from the sinusoidal pattern of variation and have a reduced amplitude of variation between δ13C values recorded in winter and the consecutive summer (0.1‰ and 0.5‰ respectively; Pattern C, Fig 4). Results from the sequential analysis of stable carbon (δ13C) and oxygen (δ18O) isotope ratios in cattle (Bos) and sheep (Ovis) third molars from the Starčevo phase at Alsónyék (ALS). The figure shows some typical results for the different observed patterns. Patterns A, B and C: see main text. W = winter; S = summer. Results from the sequential analysis of stable carbon (δ13C) and oxygen (δ18O) isotope ratios in tooth enamel. Samples were wild boar (Sus18, 19, 20 and 21), pig (Sus22) and red deer (Cervus1 and 2) from the Starčevo phase at Alsónyék. 10.1371/journal.pone.0295769.t004Table 4Tooth enamel carbon and oxygen stable isotope ratios from the Starčevo phase at Alsónyék.Specimencommon nameSpeciesδ13C (‰)δ18O(‰)MinMaxMΔMinMaxMΔALS Cervus1 M3red deer Cervus elaphus -12.9-12.5-12.70.4-8.8-4.6-6.74.2ALS Cervus2 M3red deer Cervus elaphus -14.3-13.1-13.71.2-8.7-4.9-6.83.9ALS Sus20 I1wild boar Sus scrofa -15.0-13.1-14.01.9-8.8-5.8-7.33.0ALS Sus20 I2wild boar Sus scrofa -15.0-13.8-14.41.1-8.4-6.4-7.42.0ALS Sus18 Cwild boar Sus scrofa -13.9-11.6-12.72.3-8.6-5.6-7.13.0ALS Sus19 M3wild boar Sus scrofa -14.2-13.6-13.90.6-9.3-7.6-8.41.7ALS Sus21 M3wild boar Sus scrofa -13.6-12.4-13.01.2-7.4-6.2-6.81.2ALS Sus22 M3dom. pig Sus domesticus -11.5-11.0-11.20.5-6.7-5.7-6.21.0ALS Ovis1 M3sheep Ovis aries -12.4-10.7-11.51.6-7.8-2.9-5.34.9ALS Ovis2 M3sheep Ovis aries -13.8-11.4-12.62.5-9.6-2.4-6.07.3ALS Ovis3 M3sheep Ovis aries -12.6-10.1-11.42.6-8.4-0.6-4.57.8ALS Ovis4 M3sheep Ovis aries -12.2-10.7-11.51.5-6.8-2.1-4.44.7ALS Ovis5 M3sheep Ovis aries -11.5-9.7-10.61.8-7.5-0.8-4.16.7ALS Ovis6 M3sheep Ovis aries -13.7-10.6-12.13.0-7.6-1.4-4.56.2ALS Ovis7M3sheep Ovis aries -13.2-10.8-12.02.4-9.6-3.7-6.75.9ALS Ovis8 M3sheep Ovis aries -12.9-11.5-12.21.4-8.9-3.5-6.25.4ALS Ovis9 M3sheep Ovis aries -13.6-11.6-12.62.0-7.7-1.7-4.75.9ALS Bos1 M3cattle Bos taurus -11.0-9.7-10.31.4-7.0-2.9-4.94.1ALS Bos2 M3cattle Bos taurus -14.0-11.8-12.92.2-7.3-2.8-5.04.6ALS Bos3 M3cattle Bos taurus -12.6-10.4-11.52.2-8.2-4.1-6.14.1ALS Bos4 M3cattle Bos taurus -11.2-10.1-10.61.2-9.1-1.5-5.37.6ALS Bos5 M3cattle Bos taurus -11.8-10.2-11.01.6-8.7-5.4-7.13.2ALS Bos6 M3cattle Bos taurus -12.0-10.6-11.31.4-8.5-5.6-7.02.9ALS Bos7 M3cattle Bos taurus -11.9-10.9-11.41.0-7.8-3.8-5.84.1ALS Bos8 M3cattle Bos taurus -12.0-10.0-11.02.0-6.8-4.0-5.42.9ALS Bos9 M3cattle Bos taurus -12.6-10.8-11.71.8-6.2-3.7-5.02.5ALS Bos10 M3cattle Bos taurus -13.3-11.8-12.51.5-7.0-3.2-5.13.8ALS Bos11 M3cattle Bos taurus -12.0-9.9-11.02.1-7.6-3.8-5.73.8For additional information see S2 File. Intra-tooth minimum (Min) and maximum (Max) value;M = (Min+Max)/2; Δ = Max−Min. Sheep and cattle δ18O data were published in [85, 86], respectively. Tooth enamel carbon and oxygen stable isotope ratios from the Starčevo phase at Alsónyék. For additional information see S2 File. Intra-tooth minimum (Min) and maximum (Max) value; M = (Min+Max)/2; Δ = Max−Min. Sheep and cattle δ18O data were published in [85, 86], respectively. Cattle enamel bioapatite δ13C values vary between −14.0‰ and −9.7‰. All cattle have recorded seasonal variations in their diet δ13C values (Fig 4) with amplitudes of intra-tooth variation of 1.0‰ to 2.2‰. As with sheep, most cattle follow Pattern A (ALS Bos4, Bos5, Bos8, Bos9, Bos10 and Bos11) or Pattern B (ALS Bos2 with a winter δ13C value of −14.0‰) with phase shifts between the δ18O and δ13C sequences varying from 310° to 358°; Table H in S2 File). ALS Bos1, Bos6 and Bos7 do not show a sinusoidal pattern of variation in δ13C values (Pattern C) but rather have stable values over the summer and the preceding or consecutive winter recorded in the M3 (Fig 4). The suid teeth have recorded seasonal variations in enamel bioapatite δ18O values (Fig 5). Over a year is recorded in the wild boar canine (ALS Sus18 C) and a complete year may be reconstructed when combining the wild boar first and second incisors (ALS Sus20 I1 and I2) whose formations overlap in time [51]. The sequences recorded in the wild boar and pig M3s do not reflect complete annual cycles. Overall, the δ13C values recorded in the suids teeth vary between −15.0‰ and −11.1‰. In the wild boar canine (ALS Sus18 C), the δ13C values vary in phase with the δ18O sequence (phase shift = 355°) with an amplitude of variation of 2.3‰. The pattern of variation in ALS Sus20 I1&I2 also shows a trend for lower δ13C values in winter (−15‰) and higher in the summer (−13.1‰). The short δ18O sequences recorded in the three suid third molars are centered on the summer; in these teeth, the δ13C values vary little around −14‰ and −13‰ in the wild boar M3s (ALS Sus19 and Sus21), although a steep gradient towards higher values (−11.1‰) is measured in the part of the tooth that was formed last, and corresponding to late summer, in ALS Sus21M3 (Fig 5). In the only pig’s molar (ALS Sus22 M3), δ13C values vary between −11.5‰ and −11‰. The δ18O sequences measured in the two red deer third molars show a steep decreasing trend suggesting a record over a summer-autumn-winter sequence. Over this period, these deer have recorded decreasing δ13C values comprised between −12.5‰ and −12.9‰ in ALS Cervus1 and between −13.4‰ and −14.3‰ in ALS Cervus2. Fig 6 compares the range of variation in δ13C values in all teeth. The wild fauna (red deer and wild boar) shows the lowest δ13C values, in contrast to higher δ13C values in domestic animals (pig, sheep, and cattle), although in some of the cattle and sheep teeth, lower δ13C values are observed on a seasonal basis. Range of variation of tooth enamel δ13C values in the red deer (Cervus), wild boar (Sus18, 19, 20 and 21), cattle (Bos), sheep (Ovis) and pig (Sus22) from the Starčevo phase at Alsónyék. Patterns A, B and C: see main text. Discussion Wild fauna: Setting Alsónyék’s surrounding landscape The bone collagen δ13C values of benthic fish were lower and the bone collagen δ15N values of the piscivorous fish were more elevated compared to nearly all terrestrial bone collagen samples (Fig 2), revealing a clear separation in stable isotope ratios between animals occupying the terrestrial and aquatic domains. Freshwater fish consumption would therefore lead to higher bone collagen δ15N values, as well as lower δ13C values, if benthic fish were consumed. Feeding in forested areas likely explains the lower δ13C values in some of the terrestrial animals, with e.g. roe deer ALS Cap9 having the lowest measured δ13Ccollagen value of −23.7‰. Some of the δ13Cenamel values are similarly low (around −13.1‰ in ALS Cervus1 and as low as −14.3‰ in Cervus2). These low δ13C values are consistent with year-round grazing/browsing in a forested area with plants subject to the canopy effect, possibly located on the hills to the west of the site. However, around two thirds of the 19 deer had δ13Ccollagen values between −21.5‰ and −19.8‰, likely reflecting feeding in predominantly open areas. Measured wild boar collagen δ13C values between −21.1‰ and−19.9‰ suggest none of the wild boar sampled for bone collagen fed exclusively on undergrowth of dense forests, and may have moved frequently between forests and more open areas, possibly occupying similar habitats to wild and domestic herbivores. The sequential enamel samples give a more detailed picture: The low δ13Cenamel values from the four sampled wild boar (Fig 6) indicate feeding in predominantly closed areas, with higher δ13Cenamel values recorded in late summer/autumn in ALS Sus18C and ALS Sus21M3 (Figs 5 and 6) likely reflecting a greater contribution of forest fruits (e.g. acorns) at this time of the year for these individuals [51, 87]. The wild boar collagen δ15N values showed very wide variability, from 4.0‰ to 8.6‰ (cf. a biological variability of ca. 1.5‰ for domestic pigs of differing ages and sex when consuming the identical diets; [45]). This suggests large differences in trophic levels between different wild boar individuals. Two individuals had similar δ15N values to dogs (> 8‰: ALSSuss6, ALSSuss12), suggesting a large degree of animal protein consumption, possibly partly by feeding on wetland fish naturally trapped after flooding [88]. In the case of wild boars with low δ13C and high δ15N values, mushroom consumption may also have elevated the δ15N values [89]. This large variability in the wild boar baseline δ15N values complicates the interpretation of domestic pig diets. Cattle and sheep husbandry The cattle and sheep collagen δ13C values suggest grazing in predominantly open areas, possibly also including wetlands. Ethnographic evidence from the Carpathian basin describes wetlands being used as pastures [90–92], although mainly for cattle and pigs. Historical accounts from Hungary from the 18th to 20th centuries describe wetlands being seen as dangerous for sheep because they were liable to catch parasites (liver fluke), from which they could not be cured [93]. Floodplains and marsh edges were also described as having a predominance of unappealing, less nutritious grasses, which were still used as reserve pastures, particularly in dry years [93, 94]. Keeping sheep in the flat wetlands despite the disadvantageous environmental conditions has been suggested from lesions found on sheep remains from Körös settlements in eastern Hungary [95–97]. Nevertheless, caprine remains rather than pig remains are dominant in the faunal assemblages [96]. Cultivated fields could also have been used for grazing on a seasonal basis. In five of the twenty analysed sheep and cattle molars, the sequential enamel δ13C values deviate from the expected sinusoidal pattern of variation with a reduced amplitude of variation between δ13C values recorded in winter and the consecutive or preceding summer (Pattern C, Fig 4 and S2 File). In sheep (ALS Ovis3, Ovis5), this reduced amplitude appears to be due to a rise in the winter δ13C values, possibly caused by summer hay being provided in winter. Such hay could have consisted of crop wastes and grass (including wetland grasses; [93, 94]). In some cattle, the reduced amplitude of variation might rather be due to a lowering in summer δ13C (ALS Bos 1 and Bos6), possibly caused by forest leaves provided in the summer. Ethnographic accounts report that tree-fodder can be particularly important for cattle and sheep in dry summers [93, 98, 99]. Forest resources also seem to have been used to supplement cattle and sheep diets during winter, as evidenced in four of the twenty analysed sheep and cattle molars showing significantly lower δ13C values in winter (Pattern B, Figs 4 and 6). The phase shift between δ13C and δ18O sequences does not deviate from what is observed in other specimens, meaning that this contribution occurs mainly when temperatures are the lowest in winter. In the Alsónyék area of Hungary in the 20th century, snow cover in winter lasted an average of 35–40 days [100], reaching up to 25 cm [18]. Similar conditions appear to have prevailed in the past (within 1 °C of modern values; [101]), limiting the animals’ foraging activities in winter. Ethnographic evidence from Central Europe in the 18th to 20th century indicates that domesticated animals were frequently kept in deciduous forests from spring until heavy snowfall in winter. Historical accounts from Hungary describe winter fodder being prepared by collecting and drying leaf-bearing twigs and branches, mistletoe, as well as acorns and beechnuts [93]. Most other sheep and four cattle had winter δ13Cenamel values between −12‰ and −13‰, indicating that these individuals could also have consumed forest resources with low δ13C values, but to a lower extent (or possibly in combination with larger amounts of acorns, which are higher in δ13C), or may have relied on freshwater wetland grasses. There is little indication of forest resource use in the bone collagen δ13C values, apparently contrasting with the enamel results. This discrepancy is likely due to the averaging effects in bone, whereby the effects of short-term consumption of foods with different δ13C values are attenuated [102]. Pig husbandry Historic sources from the 18th to 20th centuries show that pigs frequently had very varied diets, for example feeding in wetlands, where they were able to eat the vegetation, bird eggs and fish [93], and were frequently freely ranging in oak forests in autumn until they could not find acorns any more in winter and returned to where they were fed by humans [93]. The wide range of measured bone collagen δ15N and δ13C values shows large variability in pig diets. Eleven of the sampled pigs from Alsónyék had low δ15N values, between 6.1‰ and 7.7‰, similar to most of the sampled wild boar, with δ13C values mostly above −21.5‰ suggesting primarily herbivorous feeding in a predominantly open environment. However, the consumption of acorns (elevated δ13C values) may mask feeding on forest undergrowth [51]. The high enamel δ13C values (−11.5‰ to −11.0‰) for the sampled pig molar suggest feeding in an open environment for this individual. Since current ecological studies have shown wild boar diets to generally consist of around 86–96% plant matter (reviewed in [68]), and because the δ15N values of wild and domesticated herbivores overlap with those of pigs, a highly plant-based diet can be inferred from these results for most of these domestic pigs. Suids likely thrived in the environment surrounding Alsónyék, where acorn-bearing forests met with wet, marshy areas (cf. [23, 93]), so it is notable that pigs only made up 3% (NISPidentifiable mammal) of the mammal assemblage recovered from the Starčevo phase at Alsónyék [23]. This is comparable to the Starčevo site Lánycsók-Égettmalom and the Körös/Criş sites Ecsegfalva 23 and Endrőd 119, all in Hungary (Fig 1), where similarly wet conditions likely conducive to pig-keeping were present, but only low numbers of pigs were recovered (0.6% to 3% of animal remains by NISPtotal; [95, 103–105]). This scarcity of pigs is also visible in the scarcity of non-ruminant adipose fats revealed by residue analysis of pottery sherds from both Alsónyék and Ecsegfalva 23 [106]. The near absence of pigs in favour of caprines and cattle has been suggested to be due to culturally driven motivations, possibly related to emotional reasons or taste preferences at other early Neolithic sites in eastern Hungary [94, 95]. Two of the 13 analysed pigs (ALS Susd5 and Susd8) had elevated δ15N values of 9.0‰ and 9.2‰, similar to dogs and the lower end of the spectrum of human δ15N values (Fig 7A; [25]), indicating consumption of fertilised crops or higher trophic level foods. Historical evidence from Hungary suggests these could have been, for example, meat, freshwater fish (stable isotope ratios suggest pike, catfish and zander in the case of dogs and ALS Susd5), whey, excess milk and leftover human food [93]. Stable carbon (δ13C) and nitrogen (δ15N) isotope ratio results for bone collagen from Alsónyék. (A) Starčevo phase, data from this study and [25]. (B): Lengyel phase, data from [25]. Ethnographic evidence from Greece shows that household pigs (i.e. 1–2 pigs stall-fed by one or more households) tended to be given dairying by-products, kitchen scraps and surplus products, whereas larger herds were taken to forage in fields and woods, and fed cereals in dry summers, or to encourage their return [107]. The bone collagen data from the Starčevo phase Alsónyék therefore suggest that a small number of pigs may have been kept as household pigs, intensely fed, whereas most pigs were kept extensively, grazing and foraging for themselves, with occasional dietary supplementation by humans. Comparison to other phases at Alsónyék As part of earlier work, faunal and human remains from Starčevo, Linearbandkeramik (LBK), Sopot and Lengyel phases (see Table 1; [14]) at Alsónyék were analysed for bone collagen δ13C and δ15N [25]. Comparison to this study indicates cattle, red deer and wild boar δ13C values becoming higher in later phases at Alsónyék (Fig 8). This suggests feeding in more open areas and less reliance on forest resources after the Starčevo phase. As this trend appears not only for cattle (domesticated animals), but also for wild boar and red deer (wild animals), this suggests deforestation from the Starčevo phase towards the Lengyel phase, likely as a result of increased grazing pressure, cutting down trees for wood, and/or intentional forest clearing. Stable carbon (δ13C) isotope ratio results for red deer, wild boar and cattle bone collagen from the Starčevo, LBK, Sopot and Lengyel phases at Alsónyék, shown as boxplots. The line in the boxplot indicates the median, and the cross the mean. Whisker boundaries were chosen as 1.5 × interquartile range. Data from this study and [25]. Pigs from the Lengyel phase at Alsónyék had ca. 2–3 ‰ higher δ15N values than wild boar, cattle and sheep (Fig 7B), suggesting better access to animal protein (or fertilised crops). Pig remains were around three times as common (as a proportion of domestic animals) at Alsónyék during the Lengyel period compared to the Starčevo period [22], which, coupled with the difference in pig diets, could indicate a higher importance placed on pig keeping by the Lengyel than by Starčevo people at Alsónyék. Comparison to other sites To place the above results into a broader context, the literature was reviewed to identify numerically published faunal bone collagen δ13C and δ15N datasets with nfaunal≥25 and npigs≥5 from Hungary, Greece, Croatia, Serbia and Romania from the Early Neolithic to the Chalcolithic (Fig 1). Where data from multiple Neolithic phases was available, the earliest phase was compared to (cf. Table 1). Analysis of remains from LBK phases at Füzesabony, Hungary, and from the Vinča phase at Stubline, Serbia, showed a large overlap between herbivore and pig δ13C and δ15N values [26, 27], suggesting pigs had little or no access to animal protein (e.g. whey, slaughter waste). (This trend is also consistent with data from Măgura—Boldul lui Moş Ivčnus, but only two domestic pigs have been analysed from this site so far [33]). By comparison, Early Neolithic remains from Zemunica, Croatia, show several pigs had low δ13C and δ15N values compared to cattle and sheep/goats [34], suggesting these pigs were kept primarily on forest resources, while cattle and sheep were less reliant on forest resources. At Zemunica, one analysed outlier individual had higher δ15N values, and may have been a stall-fed pig [34]. Remains from Vinča-Belo Brdo (see Table 1 and Fig 1; [27]), Halai [31], Kouphovouno [108], Makriyalos [31, 109] and Balatonszárszó-Kis-erdei-dűlő [26] all showed an overlap between pig, cattle and sheep δ13C and δ15N values. However, at these sites, there was a tendency for pig δ15N values to be on average slightly elevated compared to sheep and cattle, while pigs with δ15N values toward the lower end of the spectrum displayed by cattle and sheep are missing. This suggests some (limited) availability of animal protein to pigs. Additionally, several outlier pigs at these sites had elevated δ15N values, similar to the results seen for the Starčevo phase at Alsónyék, which appear to have had more access to animal protein (or fertilised crops), possibly as a result of being more intensely kept household pigs. From these data, a picture seems to be emerging, whereby pig husbandry strategies employed in Southeast Europe (cf. Fig 1) during the Early and Middle Neolithic mostly involved extensive pig herding (with occasional dietary supplementation) for the majority of individuals, while a few pigs were more intensively kept, possibly stall-fed in villages for at least part of their lives. Future data will confirm or deny this apparent trend. In contrast, results from the late Chalcolithic (second half of 5th millennium BC) sites Vităneşti-Măgurice, Hârşova-tell and Borduşani-Popină, all in southeastern Romania, show only little overlap in δ15N values between pigs and herbivores, suggesting pig diets at these sites to have comprised significantly larger amounts of animal protein [28]. Therefore, there appear to be clear spatial and temporal trends in pig keeping, the underlying affecting factors for which (whether environmental, cultural or other) remain to be discovered. Conclusion The results of this study showed a large extent of overlap in bone collagen δ13C and δ15N values for pigs, wild boar, domesticated and wild herbivores from the Starčevo phase at Alsónyék, suggesting that these animals may well have shared the same grazing/browsing areas for at least parts of the year. The spread of the data demonstrates variability between individuals of the same species, indicating that on an individual basis, different foods and habitats were used to different extents. Dense woodland or open environment “specialists” were the exception, and most of the studied animals from the Starčevo phase appear to have made use of the variability of diverse resources provided by the environment surrounding Alsónyék. The differences between individual cattle, sheep and pig diets could be due to metabolic variation, interannual variability in weather conditions, differing animal husbandry practices between households, and specific individuals being chosen to receive special treatment. The collagen and enamel δ13C data from the Starčevo phase at Alsónyék indicate that sheep and cattle tend to have fed in more open areas dominated by C3 plants (possibly including grazing stubbles on croplands) compared to some of the roe deer and red deer. Sequential enamel analyses showed seasonal variability in the diets of sheep and cattle, whereby winter diets consisted of either grazing in open environments, consumption of forest resources, provision with summer hay, or a combination thereof. This variability in winter diets of domesticated herbivores suggests multiple strategies for coping with the challenges of wintery conditions, made feasible by the diverse environment around Alsónyék. Most pigs appear to have mainly consumed plant matter, which, combined with the low numbers of pig remains at Alsónyék during the Starčevo phase, suggests pig husbandry at only very low intensity. Only a few pigs appear to have received greater amounts of dietary supplementation with animal protein or fertilised crops, emphasising the non-intensive character of most pig husbandry at Alsónyék during the Early Neolithic Starčevo phase. Comparisons of the data in this study with published datasets from Neolithic sites in Greece, Serbia, Croatia and Hungary suggests that the large extent of overlap in bone collagen δ13C and δ15N values for pigs, wild boar, domesticated and wild herbivores observed in this study was not uncommon in this period. Sharing of the same grazing/browsing areas by different wild and domestic species may have been widely practiced during the 6th and 5th millennium in Southeast Europe. The comparison of our data with those from later assemblages at the same site indicates a trend to higher δ13C values for cattle, red deer and wild boar during the later phases at Alsónyék (Fig 8). This suggests a shift to less reliance on forest resources after the Starčevo phase, likely due to increased deforestation. These observations reinforce the importance of using animals from the same period as baselines for studies of human diets. Thus, in addition to elucidating past dietary management of domestic stock at Alsónyék, this study also provides baseline δ13C and δ15N values for future animal and human isotope ratio studies of Early Neolithic people in the Carpathian Basin. Supporting information Bone collagen δ13C and δ15N results. Contains Tables A and B. (XLSX) Click here for additional data file. Tooth enamel δ13C and δ18O results. Contains Tables C-H. (XLSX) Click here for additional data file. Figures of bone collagen δ13C and δ15N results. Includes brief explanation of Bayesian ellipse modelling. (PDF) Click here for additional data file.