Ancient lipids reveal continuity in culinary practices across the transition to agriculture in Northern Europe

Abstract

Farming transformed societies globally. Yet, despite more than a century of research, there is little consensus on the speed or completeness of this fundamental change and, consequently, on its principal drivers. For Northern Europe, the debate has often centered on the rich archaeological record of the Western Baltic, but even here it is unclear how quickly or completely people abandoned wild terrestrial and marine resources after the introduction of domesticated plants and animals at ∼4000 calibrated years B.C. Ceramic containers are found ubiquitously on these sites and contain remarkably well-preserved lipids derived from the original use of the vessel. Reconstructing culinary practices from this ceramic record can contribute to longstanding debates concerning the origins of farming. Here we present data on the molecular and isotopic characteristics of lipids extracted from 133 ceramic vessels and 100 carbonized surface residues dating to immediately before and after the first evidence of domesticated animals and plants in the Western Baltic. The presence of specific lipid biomarkers, notably ω-(o-alkylphenyl)alkanoic acids, and the isotopic composition of individual n-alkanoic acids clearly show that a significant proportion (∼20%) of ceramic vessels with lipids preserved continued to be used for processing marine and freshwater resources across the transition to agriculture in this region. Although changes in pottery use are immediately evident, our data challenge the popular notions that economies were completely transformed with the arrival of farming and that Neolithic pottery was exclusively associated with produce from domesticated animals and plants.

The transition from fishing, hunting, and gathering to farming was one of the most profound transitions in human history, with far-reaching consequences for biodiversity, human health, and cultural development—effects that can still be felt today. Farming drove rapid global demographic expansion of human populations (1), although reasons for its success are harder to decipher from archaeological evidence. For decades, archaeologists have sought to characterize the shift to food production by investigating regional sequences around the globe (2–5). Despite their efforts, little consensus has been reached on central issues, such as the speed and completeness of the change or whether indigenous people or colonizers had a more dominant role. It has been suggested from stable isotope evidence of human bone that in some regions, notably Northern and Western Europe, the transition to agriculture led to a rapid and complete dietary change (e.g., refs. 6 and 7). This evidence supports a traditional view that new economic practices—based on the cultivation of cereals and the rearing of livestock—and cultural innovations—notably, pottery and new forms of stone tools—rapidly spread from centers of domestication as a “package,” completely transforming society in their wake; these elements are the major indices of the so-called “Neolithic revolution” (8). Ceramic technology has been viewed as a central component of this package. The use of pottery vessels for processing starchy grain, manipulating liquids such as milk or beer, and storing agricultural surplus, although largely untested, conforms with the notion of sedentary lifeways—so much so that the mere presence of pottery, in many contexts, is taken to be representative of a Neolithic culture focused on farming.

The universal applicability of this model, however, is fundamentally challenged by many examples of pottery-using forager societies (9) and zooarchaeological evidence (10–12) that hunting, gathering, and fishing may have persisted well after farming was introduced. Even where the spread of pottery and farming are thought to have followed similar trajectories, such as in Central and Eastern Europe, it now appears from high-resolution radiocarbon dating that they may have had quite different origins (9, 13). Therefore, contrary to commonly held perceptions, there is currently very little evidence to directly link pottery with agriculture or pastoralism. A simple approach for directly testing this link is to consider for what purpose Early Neolithic ceramic vessels were used, which is now possible through the application of robust chemical and isotopic criteria for distinguishing the origin of lipids that become bound and stabilized within the ceramic matrix or in carbonized surface deposits during culinary practices (14–17). If it can be demonstrated that Neolithic ceramics were exclusively used for processing domesticates, then the concept of a transforming package would seem more likely, which in turn would support the interpretation of gross dietary change with the arrival of agriculture from the bone isotope data.

Already, lipid residue analysis has shown that some of the earliest ceramic vessels in Central and Northern Europe and Northwestern Anatolia were used for processing dairy products (18–20). Although these findings seemingly reinforce the link between pottery and farming, we note that the majority of Early Neolithic pots analyzed so far cannot be securely associated with either wild or domesticated products. Additionally, wild plant and aquatic food remains have been observed microscopically in a small number of charred residues on pottery of Mesolithic and Neolithic date (21, 22), indicating that there may be more complexity in pottery use than implied by a universal change to the processing of reared and cultivated foods in the Neolithic. Clearly, the persistence of culinary practices that originated in preagricultural societies cannot yet be ruled out and requires systematic investigation. Here we compare culinary traditions across the transition to agriculture, by examining one of the most well-known examples of pottery use by terminal Mesolithic hunter-gatherer-fishers with their early Neolithic successors.

In the Western Baltic region of Northern Europe, the appearance of cereals and domestic animals demonstrates that farming was practiced from at least ∼4000 calibrated years (cal) B.C. Here, a model of rapid and complete change from foraging to farming, driven by the arrival of migrant farmers, has found support from two sources. First, human bone collagen stable isotope data has been interpreted to suggest that marine foods were rapidly abandoned as soon as domestic terrestrial foods were introduced (7, 23). Second, initial analysis of human ancient mitochondrial DNA from a range of skeletons from adjacent regions of Germany and Sweden has revealed genetic discontinuity between hunter-gatherers and early farmers (24) and between hunter-gatherer and modern populations (24, 25), which is consistent with a high degree of population replacement (26), at least concerning the maternal lineage.

In the Western Baltic, there is also a distinct change in many aspects of material culture at this time, which some scholars have seen as support for a synchronous economic, cultural, and demographic transition. In this region, ceramics associated with the Ertebølle culture (EBK) predate the first evidence for agriculture and pastoralism by at least 600 y (5, 27, 28). However, even here there is a clear change in pottery designs and manufacturing techniques with the arrival of farming. Pointed-bottomed EBK vessels were rapidly and totally replaced by flat- or rounded-based Funnel Beaker (TRB) vessels (27–30), although some forms, such as oval-shaped “blubber lamps” continued to be produced. Impressions of domesticated cereal grains are first encountered on TRB vessels (22, 27), and accelerator mass spectrometry (AMS) radiocarbon dates made on the earliest bones of domesticated cattle from Denmark are contemporary with dates on organic inclusions from the earliest TRB vessels in the same region (ref. 31; Table S1). Moreover, human skeletons from this region dating before 4000 cal B.C. have isotope values consistent with heavy marine consumption, whereas nearly all prehistoric skeletons dating to after 4000 cal B.C. are interpreted to have had terrestrial diets (7, 23). Therefore, even here, the evidence for a synchronous economic shift in food procurement and a cultural shift in pottery technology appears to be compelling.

To test this hypothesis directly, we determined the stable isotopic composition of 100 carbonized surface deposits and the isotopic and structural characteristics of absorbed lipids preserved in 133 out of 220 ceramic vessels examined, from late forager (EBK), transitional, and early farming (TRB) sites (Fig. 1 and Table 1).

Fig. 1.

Location of sites from where Late Mesolithic (EBK) and Early Neolithic (TRB) vessels were obtained. Inset shows the geographical extent of the EBK (A) and TRB (B) cultures and their typical vessel forms.

Table 1.

Sample information and summary of lipid residue data

Period and site/region	Site description/context	Date, ka cal B.C.*	Vessels, n	>5 ug⋅g−1 lipid, n	Aquatic biomarkers, %†	Surface deposits, n
Neolithic Funnel beaker (TRB)
Åkonge	Inland settlement at lake edge	4.0–3.8	20	20	20	13
Åmose	Pots deposited in inland lake	3.9–2.7	8	8	50	8
Salpetermosen	Pot deposited in inland lake	3.3	1	1	0	1
Roskilde Fjord	Isolated underwater find	3.3–3.1	1	1	0	1
Transitional
Neustadt (EBK)	Submerged coastal settlement	4.6–4.1	34	25	24	17
Neustadt (TRB)	Submerged coastal settlement	4.1–3.7	30	26	19	16
Wangels (TRB)	Submerged coastal settlement	4.1–3.7	16	13	23	4
Bjørnsholm (TRB)	Coastal shell midden	3.9–3.5	14	9	0	0
Norsminde (TRB)	Coastal shell midden	3.9–3.5	12	2	0	0
Stenø	Inland settlement at lake edge					13
Late Mesolithic Ertebølle (EBK)
Åle	Coastal shell midden	4.6–4.0	1	1	0	0
Tybrind Vig	Submerged coastal settlement	4.6–4.0	46	19	21	20
Timmendorf-Nordmole	Submerged coastal site	4.4–4.1				2
Teglgård-Helligkilde	Submerged coastal settlement	4.6–4.0	1	1	100	1
Förstermoor	Inland settlement	5.2–4.3				1
Ringkloster	Inland settlement	4.8–4.0	36	7	14	3

The data are summarized by vessels, taking into account when multiple samples were analyzed per vessel.

*Dates of the pottery sampled.

^†Defined by the presence of at least one of the three isoprenoid alkanoic acids (phytanic, pristanic or 4,8,12-TMTD) and ω-(o-alkylphenyl)alkanoic acids of carbon lengths C18, C20, and C22.

Results and Discussion

Carbonized surface residues from coastal locations (n = 61) were significantly enriched in ¹³C (t test: t = 8.19, P < 0.01) compared with those from inland locations (n = 39; Fig. 2). The isotope data were consistent with a substantial contribution of ¹³C-enriched carbon from marine organisms to vessels found in coastal locations, regardless of their typology and date. This result is not the pattern predicted by a change from foraging/hunting/fishing to farming, although other sources of carbon enrichment must also be considered, including postdepositional alteration (32). However, ∼20% of the pots from coastal sites that yielded a lipid residue also contained aquatic biomarkers, including isoprenoid fatty acids and long-chain (>C18) ω-(o-alkylphenyl)alkanoic acids (Table 1 and Fig. 3), which meet the established criteria for aquatic lipid identification in archeology (14). Significantly, the relatively long chain length of the ω-(o-alkylphenyl)alkanoic acids indicates an aquatic origin, because these compounds are formed from long-chain polyunsaturated fatty acids (C20 and C22) that are absent in terrestrial animal fats (17), whereas the isoprenoid acids, which are also at high abundance in aquatic oils, are absent in plants. These residues are highly unlikely to derive from the depositional environment because the ω-(o-alkylphenyl)alkanoic acids are only formed at high temperatures (>270 °C; ref. 17). Interestingly, a high proportion of the pots that contained aquatic biomarkers (40%) were Early Neolithic TRB vessels conventionally associated with the new agro-pastoral economy. Finally, medium chain-length C₁₆ and C₁₈ n-alkanoic acids from 30 of the 97 vessels from coastal sites that yielded a lipid residue, including 13 Early Neolithic vessels, were noticeably enriched in ¹³C (e.g., δ¹³C_16:0 or δ ¹³C_18:0 > −24‰; Fig. 4), within the range of modern marine oils from this region (Fig. 4), and confirming that the original vessel contents were marine in origin. The combination of long-chain (>C18) ω-(o-alkylphenyl)alkanoic acids, isoprenoid fatty acids, and single-compound carbon isotope determinations provides the strongest evidence to date for the identification of aquatic resources in archaeological pottery.

Fig. 2.

Bulk stable isotope analysis of charred surface deposits. (A) Bulk δ¹³C and δ¹⁵N data of surface residues removed from the inside of Late Mesolithic EBK vessels (open circles) and Early Neolithic TRB vessels (filled circles) from coastal (blue) and inland (green) sites. These data are compared with charred animal and plant products created experimentally by repeated use of replica pottery vessels. (B) A typical surface residue adhered to the rim of an EBK vessel.

Fig. 3.

Partial gas chromatograms of a typical absorbed residue from a Neolithic funnel beaker (N2804). (A) The major fatty acids (Cx:y) with carbon number (x) and number of unsaturations (y). IS, internal standard (tetratriacontane); P, a plasticiser contaminant. (B) Expanded to show more detail. TMTD, 4,8,12-trimethyltridecanoic acid, an isoprenoid acid found at high concentration in marine organisms (33). (C) The m/z 105 mass chromatogram of this sample, showing the range of ω-(o-alkylphenyl)alkanoic acids, produced from protracted heating of tri, di-, or monounsaturated fatty acids from the oils of marine organisms (14, 17), with carbon lengths 16 (open circles), 18 (filled circles), 20 (open stars), and 22 (filled stars). The structure of the most abundant isomer of ω-(ο-alkylphenyl)octadecanoic acid methyl ester is also shown. Methyl ester derivatives are shown, which were prepared from the total lipid extracts by using BF₃/MeOH complex (14% wt/vol).

Fig. 4.

Compound-specific stable isotope analysis of ceramic residues. δ¹³C values of individual C₁₆ and C₁₈ alkanoic acids extracted from authentic reference fats (A) and from Late Mesolithic EBK vessels (open circles) and Early Neolithic TRB vessels (filled circles) (B) from coastal (blue) and inland (green) sites. The marine and freshwater reference fats (A) were obtained from Danish coastal waters, rivers, and lakes; the terrestrial data, obtained from ref. 15, is complemented with wild boar and cows’ milk from Northern Germany. These data are plotted with 95% confidence ellipses (Systat; Version 13). The low amounts of saturated fats in marine oils make this analysis insensitive to marine products when they are mixed with more saturated terrestrial fats. However, the relative high C_16:0/C_18:0 ratio in marine fats and oils means that the C_16:0 is a more sensitive indicator of marine lipids in mixtures, as shown by a hypothetical mixing line between the mean values for ruminant milk fat and marine mammal blubber, with the percent contribution by weight of marine mammal (MM) marked. The δ¹³C_16:0 values of oils from freshwater species (n = 5), including a freshwater eel, could not be separated from terrestrially derived lipids (n = 42), but marine oils (n = 17) were significantly different from terrestrial and freshwater (one-way ANOVA; F = 3.15, P < 0.01).

Further analysis of the data showed that the lipids obtained from eight Early Neolithic vessels must have been very heavily derived from marine products because the δ¹³C values of n-octadecanoic acid are significantly enriched (>−22‰) and beyond the range measured in the reference terrestrial animals. Any significant contribution estimated to be >5% by weight of ruminant fat or >20% of porcine fat would be expected to shift the δ¹³C_18:0 beyond this value because of their higher concentrations of n-octadecanoic relative to marine fish oils and marine mammal fats (e.g., refs. 35 and 36).

In contrast, Neolithic vessels from inland sites had surface deposits (Fig. 2) and medium-chain n-alkanoic acids, which were depleted in ¹³C (Fig. 4), indicative of more terrestrial input. However, even at these inland locations, 28% of the TRB pots also contained aquatic biomarkers (Table 1). In fact, several of these pots from the Åmose region of central Zealand (Table 1) dated to at least 300 y after domestic cattle were first introduced to this island (22, 31). Further isotopic analyses of n-alkanoic acids from these samples indicated that they are more consistent with a freshwater rather than a marine source (32). ¹⁵N-enriched surface deposits (δ¹⁵N > 8‰; Fig. 2) and freshwater fish bone embedded in several samples (22) support this interpretation.

The ceramic evidence for a continuation in marine exploitation is supported by spectacular finds of huge coastal fish weirs and marine shell middens along the Danish coast that date to the Neolithic period (29, 37, 38). Similarly, along the Baltic coast of Northern Germany, the remains of marine mammals, gadids, and eels are frequently found at coastal sites, but have been difficult to date precisely because they are often found in unstratified submerged deposits and contain “old” carbon derived from the Baltic Sea. Here, the ceramic residue evidence from sites that spanned the transition to agriculture, such as Neustadt and Wangels (∼4600 to 3800 cal B.C.), showed continuous use of marine resources, despite the arrival of domestic animals, as established from ovicaprine and cattle bone that securely dated to 4150 and 4050 cal B.C., respectively (39).

However, there was also clear evidence that foods from domesticated animals were processed in Early Neolithic ceramics. Approximately one-third of the TRB foodcrusts, including pots from coastal sites, had relatively depleted ¹³C and ¹⁵N values (Fig. 2), indicative of low-trophic-level terrestrial foodstuffs, rather than aquatic resources. Although these values could be derived from wild plant or terrestrial animal products—or indeed could have altered in the burial environment—the isotopic values for n-alkanoic acids extracted from just more than half of the Neolithic vessels yielding detectable lipid residues met the established criteria for ruminant dairy products (Fig. 4; ref. 15). Remarkably, it seems that dairying was practiced at coastal and inland sites as soon as domestic animals appeared in the sequence. The mixed character of the early TRB economy was well illustrated by Neolithic ceramics containing both aquatic biomarkers and n-alkanoic acids principally derived from ruminant sources (Table S2). Overall, the evidence discredits the notion that the appearance of Neolithic ceramics was closely associated with an economic package that rapidly replaced all elements of the forager lifestyle.

Conclusions

From our data, we suggest that the introduction of farming in the Western Baltic was not as dramatic or sudden as has been inferred by some scholars from stable isotope analysis of human remains from this region (7, 23). Because relatively few human samples underpin the bone isotope analyses, it may not be possible to track dietary change using this method at a sufficient geographic and chronological resolution to completely discount continuity within individual regions or across generations. Furthermore, the shift from marine to terrestrial diets may have been more equivocal than implied from these data (11, 40). At a minimum, the processing of marine and freshwater foods continued well after the introduction of food production, and the culinary practices identified here reflect a more complex process of acculturation. The population historical background for this process could have been by farmers—notably, pastoralists—who reached the coast and began to exploit the abundant marine resources available to them, or, alternatively, it could have been coastal foragers who, through increased contact with nearby farmers, began to exploit domesticated animals and plants in addition to fishing, hunting, and foraging. Either way, our food residue data give reason to question the assumption that the apparently synchronous appearance of Neolithic (TRB) ceramics and of domestic plants and animals in the Western Baltic region represents a radical departure from hunter-gatherer lifeways. Rather, the very success of farming as an economic practice and the rapid demographic expansion following its introduction (1) may be attributable to the ability of incipient farmers to rapidly adapt to new ecological settings by combining food production with exploitation of local wild resources.

Materials and Methods

A total of 220 potsherds were selected from 15 different sites (Table 1 and Fig. 1). Each site was AMS radiocarbon dated on charcoal and/or terrestrial animal remains. In most cases, EBK and TRB pots were found in stratigraphic sequence. For the mixed Mesolithic and Neolithic deposits at Neustadt and Wangels, only vessels that could be clearly typologically identified were chosen—confirmed, in several cases, by direct AMS dates on “surface deposits” and/or inclusions in the pot matrix (Table S1). However, in cases where the pot contained marine or freshwater lipids, the AMS date was deemed unreliable because of the incorporation of old carbon from aquatic reservoirs (31). Such a reservoir effect was confirmed in several cases where AMS produced an older date on the charred “surface deposits” than organic inclusions, soot, or charcoal associated with the same vessel (Table S1).

The lipid analysis followed established protocols (15, 32, 34). Briefly, ceramic powder (∼1 g), drilled from the interior surface of each potsherd, or crushed surface residue (∼15 mg) was weighed and extracted by ultrasonication with 3 aliquots of dichloromethane:methanol (2:1 vol/vol; 5 mL). The solvent extract was separated from the powder and, where necessary, treated with activated copper turnings to remove elemental sulfur. The solvent was removed by evaporation to dryness under N₂ (40 °C) to obtain a total lipid extract (TLE). An aliquot of each TLE was silylated and analyzed by gas chromatography-mass spectrometry (GC-MS). Another aliquot of the TLE was methylated for the analysis of fatty acid methyl esters (FAMEs) which were extracted with hexane (3 × 1 mL) and analyzed by GC-MS analysis and by GC-combustion-isotope ratio MS (GC-C-IRMS). For GC-C-IRMS, instrument precision on repeated measurements was <0.3‰, and the accuracy determined from FAME and n-alkane isotope standards was <0.8‰. In addition, crushed surface residues (∼1 mg) were analyzed by IRMS as reported (32). All reported δ¹³C values were relative to Vienna Pee Dee Belemnite international standard, and all modern samples, including marine samples from the Baltic Sea, were adjusted for the addition of the effects of postindustrial carbon (41).