The development of farming was a catalyst for the evolution of the human diet from the varied subsistence practices of hunter-gatherers to the more globalised food economy we depend upon today. Although there has been considerable research into the dietary changes associated with the initial spread of farming, less attention has been given to how dietary choices continued to develop during subsequent millennia. A paleogenomic time transect for 5 millennia of human occupation in the Great Hungarian Plain spanning from the advent of the Neolithic to the Iron Age, showed major genomic turnovers. Here we assess where these genetic turnovers are associated with corresponding dietary shifts, by examining the carbon and nitrogen stable isotope ratios of 52 individuals. Results provide evidence that early Neolithic individuals, which were genetically characterised as Mesolithic hunter-gatherers, relied on wild resources to a greater extent than those whose genomic attributes were of typical Neolithic European farmers. Other Neolithic individuals and those from the Copper Age to Bronze Age periods relied mostly on terrestrial C3 plant resources. We also report a carbon isotopic ratio typical of C4 plants, which may indicate millet consumption in the Late Bronze Age, despite suggestions of the crop’s earlier arrival in Europe during the Neolithic.
Citation: Gamarra B, Howcroft R, McCall A, Dani J, Hajdú Z, Nagy EG, et al. (2018) 5000 years of dietary variations of prehistoric farmers in the Great Hungarian Plain. PLoS ONE 13(5): e0197214. https://doi.org/10.1371/journal.pone.0197214
Editor: Luca Bondioli, Museo delle Civiltà, ITALY
Received: January 18, 2018; Accepted: April 27, 2018; Published: May 10, 2018
Copyright: © 2018 Gamarra et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: The isotopic analyses were funded by the Irish Research Council Advanced Research Project Grant (RPG2013-2) (http://research.ie/), European Research Council starting grant ADNABIOARC (263441) (http://cordis.europa.eu/project/rcn/97532_en.html) held by RP and H2020-MSCA-IF-2015 (703373) (http://cordis.europa.eu/project/rcn/201137_en.html) held by BG.
Competing interests: The authors have declared that no competing interests exist.
The transition to agriculture was one of the most significant events in human history, driving major biological and cultural changes globally. A key element in most agricultural transitions is a sedentary lifestyle, which is typically accompanied by changes in subsistence from hunting and gathering to farming . The nature, duration, and timing of the onset of agriculture varied across Europe (see [2–4]) as did the technological, cultural, and biological changes associated with this transition . Agriculture spread into Europe from northwestern Anatolia and the Near East during the first half of the 7th millennium BC  and reached the Carpathian Basin ~6,000 BC . The Great Hungarian Plain (GHP), also known as the Nagy-Alföld, is a lowland situated centrally in the Carpathian Basin, east of the Danube, and connected with the Mediterranean, the Pontic steppe and Central Europe [7,8] (Fig 1). The GHP was a key area involved in the spread and development of farming across Europe. This area was the meeting point of eastern and western European cultures, and as such was a major cultural and technological transitional region throughout prehistory [9,10].
Gamba et al.  analysed the genomes of nine Neolithic, one Copper Age, two Bronze and one Iron Age (5,800–830 cal BC) burial(s) spanning a 5,000-year temporal transect from the Early Neolithic to Early Iron Age. They evaluated the interface between genetic changes brought on by migrations and interactions during these key techno-cultural transitions in the GHP. Interestingly, although eight of the nine Neolithic and Copper Age individuals were genetically affiliated with modern-day Sardinians (a pattern which has been observed for numerous other European Neolithic farmers [14–17]), one of the Early Neolithic Körös individuals analysed has a genomic profile characteristic of an non-admixed European Mesolithic hunter-gatherer. This unexpected result suggests that this Early Neolithic individual may not have had shared the same subsistence practices as other early and later Neolithic individuals. Their study also showed genomic turnovers coinciding with the advent of the Bronze and Iron Ages, which contrasted with ~2,800-years of genetic continuity during the preceding Neolithic and Copper Age periods.
Archaeological context and subsistence practice in the Great Hungarian Plain
Agriculture arrived in the GHP with the appearance of the Early Neolithic Körös culture [2,18] (see Table 1 for more details). This was the northernmost expression of the Balkan Early Neolithic complex (Starčevo-Körös-Criş cultures) that had spread farming throughout south-east Europe from its origins in the Near East . Körös subsistence strategies were predominantly based on grain cultivation (wheat, barley, einkorn and legumes) and animal husbandry [2,19]. Most of the Körös faunal assemblages suggest that these early farmers had a characteristically southeastern European subsistence that was heavily reliant on sheep and goat husbandry, followed by cattle contribution (although in some Körös sites, the number of cattle bones was higher than ovicaprids ), and a very limited use of pigs and wild resources [2,21]. The Neolithic Linearbandkeramik culture (LBK) appeared in Transdanubia (West Hungary) and spread across the loess plains of Central Europe westward to the Paris Basin and eastward to the Ukraine, and eventually being responsible for establishing farming across much of the northern regions of Europe [22,23]. The Alföld Linear Pottery culture (ALP) was a LBK variant culture in the GHP. The subsistence practice was characterized by a major reliance on cattle [2,21,23], as well as cereals (mainly: emmer, barley, einkorn), legumes (pea and lentils), and flax cultivation [2,23–25]. The ALP culture was succeeded by the Late Neolithic Tisza–Herpály–Csőszhalom complex . The range of subsistence practices was similar to the previous ALP, with reliance on grain cultivation and an emphasis on domesticated cattle at the expense of ovicaprids (sheep and goats). Additionally, it included an increase of wild resources consumption .
The transition from the Late Neolithic to the Copper Age in the GHP was characterised by an overall cultural continuity, but with some changes in settlement patterns [27,28] and subsistence, namely from a reliance on agricultural products to a focus on animal husbandry (mainly cattle). This shift became predominant especially during the Middle Copper Age [27,29].
Bronze metallurgy arrived to the Carpathian Basin in the middle of the 3rd millennium BC, as a result of trade links with the northern Pontic and Balkan Peninsula [30,31]. During the Bronze Age, several new regional cultures appeared on the GHP in association with local technological developments, the emergence of extensive trade networks for metallurgy, and the arrival of various groups from the steppe [30,32–35]. The subsistence practices of the Early and Middle Bronze Age tell cultures (named for the new type of settlements tells or nucleated villages; e.g. Nagyrév, Hatvan, Ottomány and Perjámos in eastern Hungary) were characterised by intensive crop cultivation and animal husbandry [32,34,36]. This also includes domestic horses that arrived from steppe cultures, and the exploitation of freshwater resources [30,32]. The Late Bronze Age was a period of major social and economic change associated with pan-regional connections with cultures from the western part of Central Europe (e.g. Tumulus, Urnfield) [35–37]. The earliest archaeobotanical evidence of broomcorn millet is reported in China, dating 8,000 BC . It was used as a staple crop in Neolithic northern China by 6,000 BC  and spread westward to Central Asia and Eastern Europe over time [40,41]. While some evidence suggests millet consumption by Early Neolithic individuals in the GHP [42,43], it was not until the middle of the 2nd millennium BC that a shift was documented in cultivation preferences and millet was cultivated as a staple crop in the Carpathian Basin, as well as in continental Europe [36,44,45].
Iron metallurgy arrived in Central Europe during the first millennium BC . During the Early and Middle Iron Ages, pre-Scythian (also referred as Mezőcsát communities) and Scythian cultures from the Eastern Steppe inhabited the GHP and the adjacent northern mountainous region . These populations practiced a form of nomadic stockbreeding [47,48], leaving limited traces of material culture behind in the archaeological record. Mezőcsát groups who lived predominantly in the northern part of the GHP are suggested by some to be descendants of these nomadic groups , while others argue local continuity from the Late Bronze Age to the Middle Iron Ages, and the adoption of a pastoralist lifestyle as a result of contacts with eastern populations .
Carbon and nitrogen stable isotope ratios as paleodietary indicators
The use of carbon and nitrogen stable isotope ratios to study paleodiet is based on the principle that the isotope values of food consumed by animals and humans are reflected in the individual’s tissues (see [49–51]). Moreover, as bone is constantly being turned over by remodelling, the stable isotope ratios of bone collagen are indicative of the average diet over a period of time prior to death that may extend many years or even decades in adult individuals . In brief, carbon stable isotope ratios (δ13C; see S2 Appendix) are primarily used to distinguish between the consumption of C4 and C3 photosynthesizing plants, and the animals raised on them . They also allow differentiation between the consumption of terrestrial and marine foods [54,55]. In terrestrial ecosystems, the range of δ13C in C3 plants, which include most trees, shrubs, temperate grasses, and domesticated cereals such as wheat, barley, oats, and rice, varies widely from −24‰ to −36‰ (mean −26.5‰) depending on environmental factors [56,57]. C4 plants are typically arid adapted species and include some domesticated plants like maize, sorghum, and millet. C4 plants tend to have less variable and higher δ13C values (mean −12.5‰) than C3 plants, and these values create a bimodal distribution making them distinguishable from each other. Although variation in δ13C values exists, freshwater fish consumption results in lower δ13C values than in terrestrial C3 environments, with a range of −23‰ to −21‰ δ13C values in Europe [58–60].
Nitrogen stable isotope values (δ15N) in tissues increase with trophic level, with a stepwise increase of 3–5‰ between diet and consumers. This trophic level increase is used to analyse the relative consumption of plant versus animal proteins in the diet [61,62]. Aquatic resources, from both marine and freshwater ecosystems, can be markedly 15N-enriched compared to terrestrial foods, due to longer and more complex food webs [58,63]. It is of relevance to note that various environmental factors including water stress (e.g. [64,65]), soil conditions (e.g. [66,67]), climate (e.g. [68,69]), health of the individual , or breastfeeding practices  have also shown to influence nitrogen isotope ratios. The use of animal manure additionally increases δ15N values in soils resulting in higher nitrogen isotope ratios in the tissues of animals and humans consuming fertilised crops [72–74].
Previous stable isotope research in Hungary has focused on Neolithic and Copper Age populations [22,25,42,75,76]. These studies indicate that during the Late Neolithic and Copper Age on the GHP, animal protein featured more prominently in peoples’ diet than in previous Early and Middle Neolithic periods [25,42,76]. Additionally, terrestrial C3 plants made most of the contribution to individual’s diets, except for some sites where some individuals’ δ13C values suggest an increase in C4 plant consumption (e.g. millet) . Likewise, freshwater resources probably did not contribute significantly to Neolithic and Copper Age diets . However, no palaeodietary isotopic studies included samples from Bronze and Iron Ages from the eastern GHP and have analysed dietary changes associated with the arrival of new material culture and migration events.
This study includes isotopic data from a 5,000 years transect of the GHP, including Bronze and Iron Age periods, and analyses dietary changes along prehistoric time in the GHP. Specifically, the aim of this study is to investigate if the genetic changes and cultural transitions occurred in the GHP from Early Neolithic to Iron Age were associated with dietary shifts. The analysis focuses on carbon and nitrogen stable isotope ratios of bone collagen of 13 individuals reported in , together with 39 additional individuals from the same sites. We found that changes in dietary patterns were not always accompanied by cultural and genetic shifts. We also report higher consumption of C4 plant resources in the Late Bronze and Iron Age periods.
Material and methods
Carbon and nitrogen stable isotope analysis was carried out on collagen from bone samples of 52 human individuals from both sexes and various ages, along with 17 faunal bone samples representing 6 species from three settlements (Tables 2 and 3; see details in S1 and S2 Tables). All necessary permits were obtained for the described study, which complied with all relevant regulations. Only data from adults and subadults >4 years old were used for statistical purposes in order to avoid individuals that might have still been breastfeeding or in the weaning process by the time of death . The samples span from the early Neolithic Körös culture to the Early Iron Age Mezőcsát culture. We combined previous isotopic data for humans and fauna from the same region with the new data from this study in order to increase sample sizes and to analyse diachronic patterns of dietary change (detailed in Tables 2 and 3). Published data without associated ages of individuals were not used. The variability of isotopic data from sites of the same time period is likely to be due to localvariations in the isotope composition of the food web, rather than a difference in subsistence practice as shown by . Nevertheless, human samples from the Körös Early Neolithic site of Tiszaszőlős-Domaháza were compared separately from the rest of Early Neolithic individuals to test if their diet was significantly different, as their hunter-gatherer genetic affinity suggests.
Collagen extraction followed a modified Longin method  (described in detail in S2 Appendix). Stable isotope analysis was carried out following the routine procedures of Light Stable Isotope Mass Spectrometry Laboratory of the Department of Geological Science at University of Florida (Gainesville, USA) and the University of Bradford Stable Light Isotope Laboratory (Bradford, UK) (described in S2 Appendix). The atomic C:N ratio range of 2.9−3.6 was used for quality control of the bone collagen, as it corresponds to well-preserved samples . The distribution of isotopic values obtained was tested for normality using Shapiro-Wilk test, using p<0.05 at the statistical significance level (S3 Table). Statistical analyses (ANOVA and HSD Tukey post-hoc tests) were performed to assess differences between samples belonging to different periods. For those periods that do not follow a normal distribution, non-parametric Kruskal-Wallis and Mann Whitney U tests (for mean comparisons) were used. Statistical data were generated using SPSS version 24 and PAST version 3.16 . All data generated or analysed during this study are included in this published article (and in Supporting Information files).
Bone collagen was obtained for 48 of the 52 human samples and for all faunal samples. Only one human sample did not have an atomic C:N ratio within the acceptable quality range of 2.9–3.6  and its isotopic data were not considered for further statistical analyses and discussion. The information on faunal and human data obtained for this study are summarised in Table 4, and detailed in S1 and S2 Tables.
Isotope values of local fauna
Both domestic fauna (including ovicaprids, cattle and pig) and wild fauna (red deer, aurochs and wild boar) were used for statistical purposes, and are reported in S1 Table. The results are similar to those reported previously for other fauna from the same region and time periods. The stable isotope ratios of the local fauna analysed, with δ13C ranging from −21.8‰ to −19.4‰ and a mean value of −20.6‰ ±0.6‰ (1σ), are consistent with expected ranges for a terrestrial C3 environment . The nitrogen isotopic values range from 5.7‰ to 9.5‰, with a mean value of 7.0‰ ±1.2‰ (1σ) (Table 4 and S1 Table).
When comparing domesticates, no significant differences were found in carbon isotopic mean values over time (ANOVA: F = 2.631, p = 0.076) (Fig 2A). However, it appears that Middle Neolithic and Copper Age domesticated animals show higher δ15N values than those sampled from the Early Neolithic (Mann-Whitney U tests, p< 0.001) (Fig 2B). By contrast, wild fauna do not show significant isotopic changes over time (Kruskal-Wallis δ13C: H = 5.635, p = 0.060; ANOVA δ15N: F = 0.639, p = 0.542).
Boxplot showing the (A) δ13C and (B) δ15N values of fauna samples from Early Neolithic to Copper Age. Faunal data in this study were supplemented by published data of Early Neolithic [42,76], Middle Neolithic [22,25] and Copper Age  samples of the GHP. Domesticated fauna results on the left in yellow; wild fauna on the right in green. The dots within the boxes represent individual values of the samples; the horizontal line within the boxrepresents the median value; the vertical lines represent the range of data; and the asterisks are the possible outliers.
Isotope values of the human samples
The human δ13C values reported here range from −22.6‰ to −14.4‰ (mean value of −19.4‰ ±1.5‰ (1σ)) and the δ15N values range between 8.8‰ and 13.1‰ (mean value 10.7‰ ±0.8‰ (1σ)) (Table 4 and S2 Table).
When combining the current and published data, most of the individuals from Early Neolithic to Early Bronze Age have similar δ13C values consistent with a diet based on terrestrial C3 resources (Figs 3A and 4; S2 Table). However, Körös Early Neolithic individuals (n = 2) from the Tiszaszőlős-Domaháza site have significant lower δ13C values (p<0.05) than other Körös Early Neolithic and later period individuals (Table 5), similar to the Mesolithic individual from Maroslele-Pana  (Fig 5). By contrast, individuals from the Late Bronze Age and Iron Age have significantly (p<0.05) higher δ13C values than those reported for previous periods (Fig 3A, Table 5).
Boxplot showing the (A) δ13C and (B) δ15N values of human samples from Early Neolithic to Early Iron Age. Human isotopic values were combined with previous published data on the GHP from Early , Middle [22,25] and Late Neolithic , together with Early and Middle Copper Ages . Genetic affinities are based on . The dots within the boxes represent individual values of the samples; the horizontal line within the box represents the median value; the vertical lines represent the range of data.
Both human and fauna isotopic data are represented by mean isotopic values (standard deviation ±1σ indicated by bars). Domesticated and wild fauna belongs to Early/Middle Neolithic and Copper Age periods, and combined with published data [22,25,42,76]. Human results from this study were also combined with previous published data in the GHP [22,25,76]. E. Neolithic* represent mean values of Körös Tiszaszőlős-Domaháza site.
Individual human results from this study (represented by highlighted symbols) were also combined with previously published data in the GHP [22,25,76]. E. Neolithic* represent mean values of the Körös Tiszaszőlős-Domaháza site.
When comparing samples from Early Neolithic to Middle Copper Age, individuals have a range of δ15N values suggesting varying amounts of animal protein intake, especially in the case of the Early Neolithic (Fig 3B). As indicated above, the Körös Tiszaszőlős-Domaháza individuals have the highest δ15N values, with significant differences than other Körös Early Neolithic and later period individuals, with the exception of Late Copper and Early Bronze Age individuals (Table 6). The mean δ15N values of Early Neolithic individuals (without including the Tiszaszőlős-Domaháza samples) have significantly lower values (p<0.05) than individuals from the Middle Neolithic to Middle Copper Ages (Table 6). At the same time, Late Neolithic samples, together with the Middle Copper Age ones, have significantly higher δ15N mean values (p<0.05) than the values for Middle Neolithic individuals (Fig 3B and Table 6). Individuals from the Late Copper Age have significantly lower δ15N values (p<0.05) than Late Neolithic and Middle Copper Ages samples (Table 6). Bronze and Iron Age δ15N values tend to be more restricted and less variable than previous periods. Late Bronze Age individuals have significantly higher δ15N values (p<0.05) than those of the Early Neolithic (Table 6).
Variations in the carbon and nitrogen isotope ratios suggest that there were three main dietary patterns characteristic of Great Hungarian Plain populations spanning from the Early Neolithic to the Early Iron Age (Figs 4 and 5). Additionally, variation in the δ15N values may be indicative of changes in agricultural and farming activities during Hungarian Prehistory.
Mesolithic dietary pattern during the Körös Early Neolithic
The two Körös individuals from the site of Tiszaszőlős-Domaháza have lower δ13C and higher δ15N values than both the other Körös and the later period individuals. These δ13C values are too low for a diet based entirely on the local terrestrial resources, and suggest the incorporation of freshwater fish resources in the diet. This is supported by the faunal assemblage from the Körös levels at this site . The site was dominated by wild resources including a substantial number of fish remains and mussel shells. Although fish remains are also found to some degree at other Körös sites, the isotope ratios from the other Körös individuals analysed here and in other studies  suggest that they did not form a substantial part of the diet elsewhere, despite their proximity to freshwater resources access. Likewise, these δ13C values are very similar to those found in a Mesolithic individual from a southern site (Maroslele-Pana) interpreted as freshwater resource consumer . The δ15N values of Körös Tiszaszőlős-Domaháza individuals, nevertheless, are even higher (~1.5‰) than the Maroslele-Pana Mesolithic individual. This suggests that freshwater resources either made a greater contribution to the diet than at Maroslele-Pana, or that higher trophic level freshwater resources were consumed. Thus, the levels of fish consumption at Tiszaszőlős-Domaháza would appear to be a continuation of a Mesolithic dietary pattern, in accordance with the non-admixed Western hunter-gatherer genetic pattern found previously . However, although these results are statistically significant, more samples are needed to fully understand the dietary pattern at this particular site.
Dietary continuity from the Early Neolithic to Early Bronze Age
The remainder of the Neolithic, Copper and Early Bronze Age human isotopic data suggest a broadly similar diet based on C3 terrestrial resources. Nevertheless, there appears to be some temporal and regional variation. The Körös Early Neolithic samples have lower δ13C mean values than those of the subsequent Middle Neolithic, Copper and Early Bronze Age. Although small, the consistency of this difference suggests that it is a genuine reflection of either environmental influences on plant or animal δ13C values, such as increased irrigation , or dietary change, such as the increase in consumption of carbon enriched C4 plant resources (either consumed directly or by animals who ingested them). Although archaeobotanical evidence for the presence of common millet  and isotopic indications of C4 plant consumption exist in some Hungarian sites from the Early Neolithic , millet was probably not used as an important dietary and foddering source until later periods, as discussed below. The less negative δ13C values from Middle Neolithic to Early Bronze Age might indicate the inclusion of some C4 plant source in the diet, explaining its difference with previous Early Neolithic samples. However, the consumption was not as important as other C3 crops as their δ13C values still indicate that diet is overwhelmingly based on terrestrial C3 plants.
Millet consumption from the Late Bronze Age in the Great Hungarian Plain
Although two of the Bronze Age individuals analysed in  have overall similar genetic affinities to modern-day Central Europeans, the isotope data presented here suggest that their subsistence was based on different resources. The isotope ratios of the Early Bronze Age individuals are similar to those observed throughout the Neolithic and Copper Age periods with a diet mostly based on terrestrial C3 plant resources. In contrast, the Late Bronze Age and Early Iron Age individuals have significantly higher δ13C‰ than the values reported for earlier periods. Marine consumption would seem extremely unlikely given the landlocked location of the GHP, but the consumption of millet is entirely possible. As mentioned above, there is some scattered evidence that broomcorn millet may have been present in Europe (including Hungary) from the Early Neolithic [40,42,43]. However, recent direct radiocarbon dating of millet grains in some Eastern and Central Europe sites has cast doubt on evidence for millet predating the Middle Bronze Age . Stable isotope analysis on human bone collagen, together with the higher frequency and richness concentrations of millet grains in the archaeobotanical record, suggest that millet was consumed on a large-scale and used as an important crop in Europe from the second millennium BC. This consumption was especially noticeable from the Late Bronze Age onwards [44,45,82–84], although there is also evidence that it was present during the Middle Bronze Age in northern Italy . The regular consumption of millet as a staple crop by the Late Bronze Age is thus in accordance with the archaeobotanical evidence reported in Hungary . This may suggest that the two Bronze Age individuals (reported in ) represent separate migrations into the GHP by groups with similar genetic ancestries. Taken together, these individuals’ consumption of millet may reflect the exogenous dietary practices of migrant populations. Establishing whether millet cultivation was adopted by indigenous people through trade with people consuming it (e.g. as part of the network package from other areas, such as northern Italy ), or migrants introducing their crop to the local population in the GHP, requires more genetic data along with other isotope approaches (strontium and oxygen isotope analyses). The inclusion of more isotopic data from Early and Middle Bronze Age individuals will also be needed to address the question if millet was consumed as a staple crop in earlier periods, as in other areas of continental Europe . Alternatively, it may also be possible that cultivation of the C4 crop gradually intensified from the Bronze to Iron Ages. The C4 plant input in the diet of Late Bronze and Early Iron Age people may originate from the direct ingestion of cultivated millet or consumption of fauna foddered with this C4 plant. However, without associated faunal values it is not possible to discern between these two possibilities.
Insights into changing agricultural and farming practices
The nitrogen data exhibit variability from Early Neolithic to Early Iron Age. The results from humans (excluding the Körös Tiszaszőlős-Domaháza individuals) and domesticated animals, show significantly higher and more homogenous δ15N values from the Middle Neolithic onwards. Nitrogen isotopic ratios have been shown to be affected by several environmental factors (see  for a review). Among them, changes in the landscape due to human activities can also impact the δ15N values in the soil and the plants growing in it. Forest clearance resulting from slash-and-burn agriculture and animal husbandry practices have been shown to increase the nitrogen content in soils, although this pattern is not consistent in all cases . A recent study on human impact on the landscape in North-East Hungary (in the Middle Tisza floodplain)  shows periods of woodland clearance for agricultural and farming activities in the area during Neolithic times. Especially important was the AVK Middle Neolithic period when there was an unequivocal woodland clearance and burning impact on the landscape. According to this work, this favored the spread of plant species indicative of woodland recovery following clearance. This might explain why not only the human and domesticated data from Middle Neolithic present higher δ15N values, so too do the wild fauna from Middle Neolithic (Figs 2B and 3B). Additionally, this study  suggests that clearance continued during the Early Copper Age period, which might have also increased δ15N values in wild fauna for the same reason. However, our data suggest that, although there is a slight increase in δ15N values for the wild fauna during Middle Neolithic, this difference is not statistically significant (as it is for humans and domesticates). Therefore, although the homogeneity of δ15N values may be explained by changes to agricultural practices, other factors may also have caused an increase of the values themselves in humans and domesticated fauna.
Previous studies in the same area show that high δ15N values from Late Neolithic to Copper Age might be due to both an increase of animal protein consumption and to the effect of fertilising crops with animal secondary products (e.g. manure) [25,29]. Nitrogen isotope ratios have shown to significantly increase in modern crops fertilised with animal dung in long-term experimental studies [72–74]. These results showed the manuring effect in nitrogen isotopic composition in bone collagen results, and probably the overestimation of animal protein intake in Neolithic human palaeodietary reconstructions . Thus, a 9‰ to 11‰ δ15N range in values of human bone collagen was suggested for a combination of manured crops and animal protein intake in early Neolithic European farmers [72,73,88]. Although the sample size is small, the lack of significant variation in δ15N values in the wild fauna used here suggest that probably the enriched 15N isotopic ratios from Middle Neolithic domesticated fauna are due to the use of fodder from manured crops [86,88]. Thus, the isotopic results for human data, with values >9‰, might be indicative of both an increase of animal protein intake and consumption of agricultural products fertilised with manure from the Middle Neolithic onwards. This suggests a cultivation of manured crops earlier than previously suggested [25,29].
The decrease of the mean δ15N values during the Early Copper Age could be ascribed to important changes that occurred during the Neolithic and Copper Age transition. During this transition, there were crucial changes in the use of the landscape, with settlements more dispersed across the landscape and fewer nucleated sites . Giblin demonstrated that these changes did not necessarily lead to abrupt changes in dietary patterns between these two periods, as originally proposed . In fact, the observed decrease in δ15N mean values is not statistically significant, and it is the variability of individual values that increase from previous Late Neolithic ones (Fig 3B). Giblin argued that the variability during the Early Copper Age might represent more regional/local site differences in terms of protein consumption and nitrogen enrichment patterns, rather than different subsistence practices. Additional samples from more sites of north and south east Hungary will shed more light on this issue.
As well as this, Hoekman-Sites and Giblin  argued for an increase in animal protein consumption in the Middle and Late Copper Age, based on an increased presence of animal fat in pottery residues. The fall in δ15N values in our data for Late Copper Age individuals would not appear to support this interpretation. However our sample size for this time period is small (n = 3) and results must be interpreted with caution.
In summary, the results of the isotopic data analysed here suggest changes in dietary patterns from Early Neolithic to Early Iron Age that are not in all cases associated with cultural and genetic shifts. The dietary practices of the individuals from the Early Neolithic Körös site of Tiszaszőlős-Domaháza is in accord with the genetic data. Their dietary isotopic profiles resemble the continuity of a Mesolithic hunter-gatherer pattern with an important reliance on freshwater resources consumption. The Mesolithic genomic signature of one of the individuals, together with similar dietary patterns, might indicate some evidence of contact between Early Neolithic communities and indigenous hunter-gatherers, as previously suggested . A diet based on terrestrial C3 plant resources seems to be common from Early Neolithic to Copper Age periods, in accordance with the genomic stasis . Although a genomic shift was shown with the advent of Bronze Age technology, isotopic results from Early Bronze Age individuals suggest that they were not accompanied by important changes in dietary pattern and that they continued with the same subsistence practices as in previous Neolithic and Copper Age times. However, the Late Bronze Age, together with Early Iron Age individuals, have significantly higher δ13C ratios than previous times suggesting a dietary change with a higher consumption of millet, which is in accordance with its wide cultivation in Europe from the second millennium BC. Nitrogen isotopic ratios from both domesticated fauna and humans are suggestive of consumption of fertilised crops from Middle Neolithic, which may be probable due to the use of secondary agricultural products (manure) .
S1 Table. Stable isotope data and sample information for faunal samples analysed.
All samples represent a single individual. Samples were analyzed in duplicate except for those marked with an asterisk. The δ13C‰ and δ15N‰ values represent average values of duplicate runs for each sample. (EN = Early Neolithic; MN = Middle Neolithic). Samples collected from 1Faunal Collection of the Archaeological Department of the Déri Museum (Debrecen, Hungary), 2Hungarian Natural History Museum (Budapest, Hungary).
S2 Table. Stable isotope data and sample information for human samples analysed.
All samples were analyzed in duplicate except for those marked with an asterisk. The δ13C‰ and δ15N‰ values represent average values of duplicate runs for each sample. Samples that do not meet the quality range (C:N = 2.9–3.6) are in bold. The upper/lower M1 samples employed here correspond to a root dentine subsample belonging to the very last stage of formation of the tooth, following (AlQahtani et al. 2010) and the subsampling method in Beaumont and Montgomery, 2015. Legend: no col. = no collagen; EN = Early Neolithic; MN = Middle Neolithic; LN = Late Neolithic; ECA = Early Copper Age; MCA = Middle Copper Age; LCA = Late Copper Age; EBA = Early Bronze Age; LBA = Late Bronze Age; EIA = Early Iron Age. Infant = 0−2 years.
S3 Table. Results of normality tests (Shapiro-Wilk W statistic) for human and faunal isotopic values.
Human and faunal number of samples (N) includes the new samples reported in this study, together with the published data (references in Tables 2 and 3). *Early Neolithic samples from Tiszaszőlős-Domaháza site were treated separately for comparison In bold: p<0.05.
The authors would like to dedicate this article to Zsuzsanna K. Zoffmann who passed away during the preparation of this study. The authors would also like to thank Alexandra Anders for comments in the revision of the manuscript.
- 1. Munro N. Epipaleolithic subsistence intensification in the southern Levant: the faunal evidence. In: Hublin J-J, Richards MP, editors. The evolution of hominin diets: integration approaches to the study of Paleolithic subsistence. Netherlands: Springer; 2009. pp. 141–155.
- 2. Whittle AWR. Europe in the Neolithic: the creation of new worlds. Cambridge: Cambridge University Press; 1996.
- 3. Bellwood P. First farmers:the origins of agricultural societies. Oxford: Blackwell Publishing; 2005.
- 4. Whittle AWR, Bickle P. Early farmers: the view from archaeology and science. Oxford: Oxford University Press; 2014.
- 5. Pinhasi R, Stock JT. Human bioarchaeology of the transition to agriculture. London: John Wiley & Sons; 2011.
- 6. Oross K, Siklósi Z. Relative and absolute chronology of the Early Neolithic in the Great Hungarian Plain. In: Anders A, Siklósi Zs, editors. The first Neolithic sites Central/South-East European transect. Volume III. The Körös culture in eastern Hungary. Oxford: BAR International Series 2334; 2012. Pp. 129–159.
- 7. Pécsi M, Sárfalvi B. The geography of Hungary. Budapest: Corvina; 1964.
- 8. Sherratt A. Economy and society in prehistoric Europe: changing perspectives. Edinburgh: Edinburgh University Press; 1997.
- 9. Zs Visy. Hungarian archaeology at the turn of the millenium. Budapest: Ministry of National Cultural Heritage & Teleki László Foundation; 2003.
- 10. Milisauskas S. European prehistory: a survey. New York: Springer Science & Business Media; 2011.
- 11. Wessel P, Smith WHF. New, improved version of Generic Mapping Tools released. Eos, Trans Am Geophys Union. 1998; 79: 579.
- 12. Amante C, Eakins BW. ETOPO1 1 Arc-Minute global relief model: procedures, data sources and analysis. NOAA Technical Memorandum NESDIS NGDC-24. National Geophysical Data Center, NOAA. 2009. https://doi.org/10.7289/V5C8276M
- 13. Gamba C, Jones ER, Teasdale MD, McLaughlin RL, Gonzalez-Fortes G, Mattiangeli V, et al. Genome flux and stasis in a five millennium transect of European prehistory. Nat Commun. 2014;5: 5257. pmid:25334030
- 14. Haak W, Lazaridis I, Patterson N, Rohland N, Mallick S, Llamas B, et al. Massive migration from the steppe was a source for Indo-European languages in Europe. Nature. 2015; 522: 207–211. pmid:25731166
- 15. Mathieson I, Lazaridis I, Rohland N, Mallick S, Patterson N, Roodenberg SA, et al. Genome-wide patterns of selection in 230 ancient Eurasians. Nature. 2015;528: 499–503. pmid:26595274
- 16. Mathieson I, Roodenberg SA, Posth C, Szécsényi-Nagy A, Rohland N, Mallick S, et al. The genomic history of southeastern Europe. Nature. 2018; 555:197–203. pmid:29466330
- 17. Lazaridis I, Mittnik A, Patterson N, Mallick S, Rohland N, Pfrengle S, et al. Genetic origins of the Minoans and Mycenaeans. Nature. 2017;548: 214–218. pmid:28783727
- 18. Anders A, Siklósi Zs. The first Neolithic sites in Central/South-East European transect. Volume III. The Körös culture in Eastern Hungary. Oxford: BAR International Series 2334; 2012.
- 19. Whittle A. The first farmers. In: Cunliffe B, editor. Oxford: Oxford University Press; 1998. pp. 136–166.
- 20. Domboróczki L. Report on the excavation at Tiszaszőlős-Domaháza-puszta and a new model for the spread of the Körös Culture. In: Kozłowski JK, Raczky P, editors. Neolithisation of the Carpathian Basin: northernmost distribution of the Starčevo culture. Kraków: Polish Academy of Arts and Sciences; 2010. pp. 137–176.
- 21. Bartosiewicz L. Plain talk: animals, environment and culture in the Neolithic of the Carpathian Basin and adjacent areas. In: Bailey DW, Whittle A, Cummlings V, editors. (Un)Settling the Neolithic. Oxford: Oxbow; 2005. pp. 51–63.
- 22. Whittle A. Hungary. In: Bickle P, Whittle A, editors. The first farmers of central Europe: diversity in LBK lifeways. Oxford: Oxbow Books; 2013. pp. 49–97.
- 23. Bickle P, Whittle A. The first farmers of central Europe: diversity in LBK lifeways. Oxford: Oxbow; 2013.
- 24. Bogaard A. Neolithic farming in central Europe: an archaeobotanical study of crop husbandry practices. Oxford: Routledge; 2004.
- 25. Giblin JI. Isotope analysis on the Great Hungarian Plain: an exploration of mobility and subsistence strategies from the Neolithic to the Copper Age. Ph.D. Thesis, The Ohio State University. 2011.
- 26. Biró K. The Neolithic in the Northern part of the Great Hungarian Plain and the Northern mountain range. In: Zs Visy, editor. Hungarian archaeology at the turn of the millenium. Budapest: Ministry of National Cultural Heritage & Teleki László Foundation; 2003. p. 101.
- 27. Horváth LA, Virág ZsM. History of the Copper Age. In: Zs Visy, editor. Hungarian Archaeology in the turn of the millennium. Budapest: Ministry of National Cultural Heritage & Teleki László Foundation; 2003. pp. 125–127.
- 28. Parkinson WA. Integration, interaction, and tribal “cycling”: the transition to the Copper Age on the Great Hungarian Plain. In: Parkinson WA, editor. The Archaeology of tribal Societies. Ann Arbor, Michigan: International Monographs in Prehistory; 2002; pp. 391–438.
- 29. Hoekman-Sites HA, Giblin JI. Prehistoric animal use on the Great Hungarian Plain: a synthesis of isotope and residue analyses from the Neolithic and Copper Age. J Anthropol Archaeol. 2012;31: 515–527.
- 30. Sherratt A. The emergence of élites: earlier Bronze Age Europe, 2500–1300 BC. In: Cunliffe B, editor. Prehistoric Europe: an illustrated history. Oxford: Oxford University Press; 1998. pp. 244–276.
- 31. Kulcsár G. The Early Bronze Age. In: Zs Visy, editor. Hungarian Archaeology in the turn of the millennium. Budapest: Ministry of National Cultural Heritage & Teleki László Foundation; 2003. pp. 141–142.
- 32. Poroszlai I. Tell cultures of the early and Middle Bronze Age. In: Zs Visy, editor. Hungarian Archaeology in the turn of the millennium. Budapest: Ministry of National Cultural Heritage & Teleki László Foundation; 2003. pp. 142–143.
- 33. Vaday A. Chronological charts. In: Zs Visy, editor. Hungarian archaeology at the turn of the millenium. Budapest: Ministry of National Cultural Heritage & Teleki László Foundation; 2003. pp. 483–488.
- 34. Dani J, Horváth T. Őskori kurgánok a magyar Alföldön. A Gödörsíros (Jamnaja) entitás magyarországi kutatása az elmúlt 30 év során. Áttekintés és revízió. Archaeolingua Alapítvány; 2012.
- 35. Fischl KP, Kiss V, Kulcsár G, Szeverényi V. Transformations in the Carpathian Basin around 1600 BC. In: Meller H, editor. 1600 BC—Cultural change in the shadow of the Thera-Eruption? Halle: Landesamt für Denkmalpflege und Archäologie Sachsen-Anhalt-Landesmuseum für Vorgeschichte; 2013. pp. 355–372.
- 36. Harding A. Reformation in Barbarian Europe, 1300–600 BC. In: Cunliffe B, editor. Prehistoric Europe: an illustrated history. Oxford: Oxford University Press; 1998. pp. 304–335.
- 37. Csányi M. The Tummulus culture: Invaders from the west. In: Zs Visy, editor. Hungarian archaeology at the turn of the millenium. Budapest: Ministry of National Cultural Heritage & Teleki László Foundation; 2003. pp. 161–163.
- 38. Lu H, Zhang J, Liu K, Wu N, Li Y, Zhou K, et al. Earliest domestication of common millet (Panicum miliaceum) in East Asia extended to 10,000 years ago. Proc Natl Acad Sci. 2009;106: 7367–7372. pmid:19383791
- 39. Liu X, Jones MK, Zhao Z, Liu G, O’Connell TC. The earliest evidence of millet as a staple crop: new light on Neolithic foodways in North China. Am J Phys Anthropol. 2012; 149: 283–290. pmid:22961368
- 40. Hunt H V, Vander Linden M, Liu X, Motuzaite-Matuzeviciute G, Colledge S, Jones MK. Millets across Eurasia: chronology and context of early records of the genera Panicum and Setaria from archaeological sites in the Old World. Veg Hist Archaeobot. 2008; 17: 5. pmid:19657476
- 41. Motuzaite-Matuzeviciute G, Staff RA, Hunt H V, Liu X, Jones MK. The early chronology of broomcorn millet (Panicum miliaceum) in Europe. Antiquity. 2013;87: 1073–1085.
- 42. Pearson JA, Hedges REM. Stable carbon and nitrogen analysis and the evidence for diet at Ecsegfalva and beyond. In: Whittle A, editor. The Early Neolithic on the Great Hungarian Plain: investigations of the Körös culture site of Ecsegfalva 23, County Bekes. Archaeological Institute of the Hungarian Academy of Sciences & School of History and Archaeology, University of Cardiff, Budapest & Cardiff; 2007. pp. 413–419.
- 43. Gyulai F. The archaeobotanical characterization of the Körös Culture. In: Anders A, Siklósi Zs, editors. The first Neolithic sites in Central/South-East European transect. Volume III. The Körös culture in Eastern Hungary. Oxford: BAR International Series 2334; 2012. pp. 223–230.
- 44. Stika H-P, Heiss A. Plant cultivation in the Bronze Age. In: Fokkens H, Harding A, editors. The Oxford handbook of the European Bronze Age. Oxford: Oxford University Press; 2013. pp. 348–369.
- 45. Lightfoot E, Liu X, Jones MK. Why move starchy cereals? A review of the isotopic evidence for prehistoric millet consumption across Eurasia. World Archaeol. 2013;45: 574–623.
- 46. Taylor T. Thracians, Scythians and Dacians, 800 BC−AD 300. In: Cunliffe B, editor. Prehistoric Europe: an illustrated history.Oxford: Oxford University Press; 1998. pp. 373–410.
- 47. Kemenczei T. The beginning of the Iron Age: The Pre-Scythians (8th century B.C.). In: Zs Visy, editor. Hungarian Archaeology in the turn of the millennium. Budapest: Ministry of National Cultural Heritage & Teleki László Foundation; 2003. pp. 177–183.
- 48. Metzner-Nebelsick C. Early Iron Age pastoral nomadism in the Great Hungarian Plain–migration or assimilation? The Thraco-Cimmerian problem revisited. In: Davis-Kimball J, Murphy ME, Koryakova L, Yablonski TL, editors. Kurgans, ritual sites, settlements Eurasian Bronze and Iron Age. Oxford: BAR–International Series 890; 2000. pp. 160–184.
- 49. Katzenberg MA. Stable isotope analysis: a tool for studying past diet, demography, and life history. In: Katzenberg MA, Saunders S, editors. Biological anthropology of the human skeleton. New Jersey: Wiley Online Library; 2007. pp. 411–441.
- 50. Lee‐Thorp JA. On isotopes and old bones. Archaeometry. 2008; 50: 925–950.
- 51. Schwarcz HP, White CD, Longstaffe FJ. Stable and radiogenic isotopes in biological archaeology: some applications. In: West JB, Bowen GJ, Dawson TE, Tu KP, editors. Isoscapes: understanding movement, pattern, and process on Earth through stable isotope mapping. Heidelberg: Springer; 2010. pp. 335–356.
- 52. Hedges REM, Clement JG, Thomas CDL, O’Connell TC. Collagen turnover in the adult femoral mid‐shaft: modeled from anthropogenic radiocarbon tracer measurements. Am J Phys Anthropol. 2007;133: 808–816. pmid:17405135
- 53. Van der Merwe NJ, Vogel JC. 13C content of human collagen as a measure of prehistoric diet in woodland North America. Nature. 1978;276: 815–816. pmid:364321
- 54. Tauber H. 13C evidence for dietary habits of prehistoric man in Denmark. Nature. 1981; 292: 332–333. pmid:7019718
- 55. Chisholm BS, Nelson DE, Schwarcz HP. Stable-carbon isotope ratios as a measure of marine versus terrestrial protein in ancient diets. Science. 1982; 216: 1131–1132. pmid:17808502
- 56. O’Leary MH. Carbon isotope fractionation in plants. Phytochemistry. 1981; 20: 553–567.
- 57. Van der Merwe NJ, Medina E. The canopy effect, carbon isotope ratios and foodwebs in Amazonia. J Archaeol Sci. 1991;18: 249–259.
- 58. Katzenberg MA, Weber A. Stable isotope ecology and palaeodiet in the Lake Baikal region of Siberia. J Archaeol Sci. 1999; 26: 651–659.
- 59. Dufour E, Bocherens H, Mariotti A. Palaeodietary implications of isotopic variability in Eurasian lacustrine fish. J Archaeol Sci. 1999; 26: 617–627.
- 60. Lillie M, Richards MP, Jacobs K. Stable isotope analysis of 21 individuals from the Epipalaeolithic cemetery of Vasilyevka III, Dnieper Rapids region, Ukraine. J Archaeol Sci. 2003; 30: 743–752.
- 61. Schoeninger MJ, DeNiro MJ. Nitrogen and carbon isotopic composition of bone collagen from marine and terrestrial animals. Geochim Cosmochim Acta. 1984;48: 625–639.
- 62. Bocherens H, Drucker D. Trophic level isotopic enrichment of carbon and nitrogen in bone collagen: case studies from recent and ancient terrestrial ecosystems. Int J Osteoarchaeol. 2003;13: 46–53.
- 63. Minagawa M, Wada E. Stepwise enrichment of 15N along food chains: further evidence and the relation between δ15N and animal age. Geochim Cosmochim Acta. 1984;48: 1135–1140.
- 64. Ambrose SH, DeNiro MJ. Reconstruction of African human diet using bone collagen carbon and nitrogen isotope ratios. Nature. 1986;319: 321–324.
- 65. Heaton THE. The 15N/14N ratios of plants in South Africa and Namibia: relationship to climate and coastal/saline environments. Oecologia. 1987;74: 236–246. pmid:28311996
- 66. Handley LL, Raven JA. The use of natural abundance of nitrogen isotopes in plant physiology and ecology. Plant Cell Environ. 1992;15: 965–985.
- 67. Marshall JD, Brooks JR, Lajtha K. Sources of variation in the stable isotopic composition of plants. In: Michener R, Lajtha K, editors. Stable isotopes in ecology and environmental science. Malden: Wiley-Blackwell; 2007. pp. 22–60
- 68. Ambrose SH. Effects of diet, climate and physiology on nitrogen isotope abundances in terrestrial foodwebs. J Archaeol Sci. 1991; 18: 293–317.
- 69. Craine JM, Elmore AJ, Aidar MPM, Bustamante M, Dawson TE, Hobbie EA, et al. Global patterns of foliar nitrogen isotopes and their relationships with climate, mycorrhizal fungi, foliar nutrient concentrations, and nitrogen availability. New Phytol. 2009;183: 980–992. pmid:19563444
- 70. Reitsema LJ. Beyond diet reconstruction: stable isotope applications to human physiology, health, and nutrition. Am J Hum Biol. 2013; 25: 445–456. pmid:23784719
- 71. Fuller BT, Fuller JL, Harris DA, Hedges REM. Detection of breastfeeding and weaning in modern human infants with carbon and nitrogen stable isotope ratios. Am J Phys Anthropol. 2006;129: 279–293. pmid:16261548
- 72. Bogaard A, Heaton THE, Poulton P, Merbach I. The impact of manuring on nitrogen isotope ratios in cereals: archaeological implications for reconstruction of diet and crop management practices. J Archaeol Sci. 2007; 34: 335–343.
- 73. Fraser RA, Bogaard A, Heaton T, Charles M, Jones G, Christensen BT, et al. Manuring and stable nitrogen isotope ratios in cereals and pulses: towards a new archaeobotanical approach to the inference of land use and dietary practices. J Archaeol Sci. 2011; 38: 2790–2804.
- 74. Szpak P, Longstaffe FJ, Millaire J-F, White CD. Stable isotope biogeochemistry of seabird guano fertilization: results from growth chamber studies with maize (Zea mays). PLoS One. 2012; 7: e33741. pmid:22479435
- 75. Cramp L. AMS Dating and stable carbon and nitrogen isotope analysis of the Middle Neolithic Linearbandkeramic site of Balatonszárszó, Hungary. M. Sc. Thesis, University of Oxford. 2004.
- 76. Whittle A, Bartosiewicz L, Boric D, Pettitt P, Richards M. In the beginning: new radiocarbon dates for the Early Neolithic in northern Serbia and south-east Hungary. Antaeus. 2002; 25: 63–117.
- 77. Longin R. New method of collagen extraction for radiocarbon dating. Nature. 1971; 230: 241–242. pmid:4926713
- 78. DeNiro MJ. Postmortem preservation and alteration of in vivo bone collagen isotope ratios in relation to palaeodietary reconstruction. Nature. 1985; 317: 806–809.
- 79. Hammer Ø, Harper D, Ryan P. PAST-Palaeontological statistics package for education and data analysis. Paleontol Electronica. 2001;4: 1–9.
- 80. Richards MP, Hedges REM. A Neolithic revolution? New evidence of diet in the British Neolithic. Antiquity. 1999;73: 891–897.
- 81. Wallace M, Jones G, Charles M, Fraser R, Halstead P, Heaton THE, et al. Stable carbon isotope analysis as a direct means of inferring crop water status and water management practices. World Archaeol. 2013; 45: 388–409.
- 82. Lightfoot E, Šlaus M, Šikanjić PR, O’Connell TC. Metals and millets: Bronze and Iron Age diet in inland and coastal Croatia seen through stable isotope analysis. Archaeol Anthropol Sci. 2015;7: 375–386.
- 83. Laffranchi Z, Huertas AD, Brobeil SAJ, Torres AG, Cantal JAR. Stable C & N isotopes in 2100 Year-BP human bone collagen indicate rare dietary dominance of C4 plants in NE-Italy. Sci Rep. 2016;6: 38817. pmid:27934943
- 84. Valamoti SM. Millet, the late comer: on the tracks of Panicum miliaceum in prehistoric Greece. Archaeol Anthropol Sci. 2016;8: 51–63.
- 85. Tafuri MA, Craig OE, Canci A. Stable isotope evidence for the consumption of millet and other plants in Bronze Age Italy. Am J Phys Anthropol. 2009;139: 146–153. pmid:19051259
- 86. Szpak P. Complexities of nitrogen isotope biogeochemistry in plant-soil systems: implications for the study of ancient agricultural and animal management practices. Front Plant Sci. 2014;5: 288. pmid:25002865
- 87. Magyari EK, Chapman J, Fairbairn AS, Francis M, de Guzman M. Neolithic human impact on the landscapes of North-East Hungary inferred from pollen and settlement records. Veg Hist Archaeobot. 2012;21: 279–302.
- 88. Bogaard A, Fraser R, Heaton THE, Wallace M, Vaiglova P, Charles M, et al. Crop manuring and intensive land management by Europe’s first farmers. Proc Natl Acad Sci. National Acad Sciences; 2013;110: 12589–12594. pmid:23858458