by Chris Fowler
Whilst the broad pattern of the shift away from marine resources in the Neolithic is striking and widespread, it is not unvarying across Europe. Even for the areas discussed above, sample sizes are often so small that the results can be viewed as provisional, though their consistency permits a fair degree of confidence. One of the more striking departures from this pattern comes from eastern Sweden, where a strong reliance on marine resources is documented at the Middle Neolithic B (c. 2800 BC) sites of Västerbjers and Ajvide on Gotland (Eriksson 2004; Lidén et al. 2004; Lindqvist and Possnert 1997), and at Korsnäs, at the time a small island on the east central coast (Fornander et al. 2008) (Fig. 19.1). Whilst domestic animals are present, human δ13C and δ15N values are both very elevated, suggesting a reliance on seals. The sites belong to the Pitted Ware culture (PWC), which shows a strong coastal orientation, with faunal assemblages also supporting the importance of sealing. Thus, there is little indication that these groups were committed to a farming way of life. This is also the case further north and east into the Baltic, where the uptake of farming was much more gradual and incomplete (see Bartosiewicz and Lillie, this volume). This begs the question of how we define the ‘Neolithic’: if based on a farming economy, then the PWC would not qualify, but of course this quickly becomes a pointless tautological exercise. What we label these societies is less important than understanding them, and in turn the variation that existed across Europe during the sixth to third millennia BC.
What makes the PWC in eastern Sweden particularly interesting is its coexistence with the distinct Neolithic Funnel Beaker (TRB) farming culture. Individuals from the TRB passage grave of Resmo on the island of Öland exhibit very different isotopic results, indicating far less use of marine resources (Lidén 1995), though still more than usually seen along the Atlantic façade. Further north on Öland, another Pitted Ware site, Köpingsvik, again shows elevated isotope values suggesting a focus on seals: human isotope values from the two sites are, with a single exception, completely non-overlapping (Eriksson et al. 2008) (Fig. 19.2). Thus, it is clear that the focus on sealing by Pitted Ware groups was not a matter of necessity, but of choice, though likely subject to rather different conditions than found further west in the Baltic, or along the Atlantic façade. Farming, whilst still possible, may have been less attractive given the shorter growing season at higher latitudes, the increased distance (in the Baltic) from the warming influence of the North Atlantic Drift (the Gulf Stream), and, crucially, the availability of viable alternatives. Sjögren’s (2003, 170–171) comment that Funnel Beaker farmers on the Swedish west coast (with terrestrial isotopic signatures from the passage grave of Hunnebostrand) chose this way of life despite the rather unfavourable environmental circumstances, as a marker of cultural identity, makes the point well. In addition to having different diets and lifeways, recent aDNA findings suggest that PWC and TRB groups in Sweden were also genetically distinct populations (Malmström et al. 2009).
FIG. 19.2. δ13C and δ15N values on human bone collagen from Neolithic sites in eastern Sweden. Västerbjers and Köpingsvik are Pitted Ware culture cemeteries, and Resmo is a Funnel Beaker (TRB) passage tomb. The single outlying datapoint from Köpingsvik is on a tooth (no bone measurement is available), clearly indicating that this individual originated in a group with a very different diet.
(data from Eriksson 2004 and Eriksson et al. 2008).
The Netherlands provide a rather different picture to that found elsewhere along the Atlantic façade, with a seemingly more gradual and piecemeal adoption of the Neolithic way of life. Human stable isotope data are currently only available from three sites, but these span the transition: Hardinxveld (Mesolithic, c. 5450–4500 BC), Swifterbant (transitional, c. 4200–4000 BC), and Schipluiden (middle Neolithic, c. 3600–3400 BC) (Smits et al. 2010). Whilst the earlier groups show no clear evidence in their δ13C values for any consumption of marine protein (Louwe-Kooijmans 2007; Smits and van der Plicht 2009), they do show significantly elevated δ15N values indicating the consumption of freshwater aquatic resources (fish and waterfowl). This is not particularly surprising, since the coastline around the low-lying Rhine–Meuse delta would not have been conducive to the exploitation of open-water marine species in either period. Moreover, Hardinxveld and Swifterbant are some 40–50km back from their contemporary coastlines. The extensive brackish and freshwater marshes behind the coastal barriers offered a rich habitat, but one with predominantly ‘terrestrial’ δ13C values. Domestic animal remains first appear in small numbers from c. 4500 BC, with evidence of cereals perhaps slightly later. The suitability of this low-lying region for growing cereals and keeping animals is debated (Bakels 2000; Bakels and Zeiler 2005; Cappers and Raemaekers 2008; Louwe-Kooijmans 2007); the highest land would be provided by clay levees and well-drained—but not particularly fertile—sand dunes that formed the focus of lowland occupation. Nevertheless, Cappers and Raemaekers (2008) argue that cereals were being grown here by the transitional Swifterbant culture. Another point of contention is the relationship between lowland and ‘upland’ sites, particularly once domesticates are attested in the latter: are these specialized seasonal camps of farmers based in the uplands, or are they the seasonal camps (none appear to be occupied year-round) of those following a hunting and gathering lifestyle, albeit with some use of domestic crops and animals?
Perhaps the most interesting stable isotope results are those from Schipluiden, a permanent settlement immediately behind the coast dune barriers, only some 3km from the contemporary coastline. The δ15N values are as high as those from late Mesolithic Hardinxveld, with highly variable but on average slightly elevated δ13C values (–18.8 ± 1.7‰), suggesting some component of marine protein, but, more importantly, the significant consumption of freshwater fish and waterfowl. One individual in particular demonstrates high marine consumption (Smits et al. 2010). It seems that the community at Schipluiden was practising a broad spectrum economy very different from that of more committed farmers further inland (Louwe-Kooijmans 2007), yet the large mammalian fauna is still dominated by domestic species (c. 60%), particularly cattle (Louwe-Kooijmans 2009, fig. 10). Located only c. 1km from Schipluiden, the contemporary site of Rijswijk presents a mixed farming economy with over 95% domestic fauna (Louwe-Kooijmans 2009, fig. 12). Unfortunately, no stable isotope data are available, but the close proximity of the two sites must raise questions concerning the relationship between them, as well as between these near-coastal communities, and the farming communities found further inland.
THE MEDITERRANEAN
Relatively few stable isotope studies have been undertaken in the Mediterranean region, though this is changing rapidly. Mesolithic populations here appear to have made considerably less use of marine resources than in Atlantic Europe (Salazar-García et al. 2014), whilst Neolithic populations do not differ substantially from those further north (it should be noted that the δ13C terrestrial endpoint tends to be 1–2‰ higher in southern Europe, so that direct comparisons between the two areas are misleading; van Klinken et al. 2000). The earliest site to be considered here is Khirokitia, Cyprus, dating to the seventh/sixth millennium BC and located 6km from the coast. Preservation of human bone collagen was poor, but no samples showed any evidence of marine protein (Lange-Badré and Le Mort 1998). Further west, the sixth/fifth millennium BC Neolithic sites of Pendimoun (Alpes-Maritimes, France) and Arene Candide (Liguria, Italy) near the coast again show no appreciable use of marine protein (Le Bras-Goude et al. 2006). A series of late Neolithic samples from Brochtorff Circle on Malta, spanning c. 4200–2700 BC, could be interpreted as showing some small contribution of marine foods, certainly more than the above sites, though not as much as could be expected given the small size of this island and its rocky soils (Richards et al. 2001). The inland Chasséen (c. 4300–4000 BC) site of Le Crès, Hérault, is located 24km from the present French Mediterranean coastline. Some marine molluscs are present, but isotopic analyses indicate a terrestrial diet; that this is based on domesticated
resources is confirmed by the exclusively domestic faunal assemblage and by the presence of wheat and barley (Le Bras-Goude et al. 2009). From southern Italy, three individuals from the Neolithic coastal site of Samari are slightly elevated in δ13C compared with five early/middle Neolithic individuals from four inland sites (–19.2‰ vs. –19.9‰), but interpretation of this as indicating some use of marine foods is complicated by the fact that δ15N values do not show a similar increase (data from Giorgi et al. 2005). Nevertheless, it is possible that low trophic level marine foods such as shellfish made a small contribution to the diet at Samari.
One of the most interesting Mediterranean studies comes from Greece, where Papathanasiou (2003) compares three coastal and three inland sites (Fig. 19.1). In contrast to Britain, the combined coastal sites exhibit slightly but significantly elevated values compared with the inland sites, suggesting a small contribution of marine protein at the former. However, the picture is more complicated, since one of the coastal sites, Aleoptrypa, actually groups with the inland sites and indeed has the lowest δ15N values of all sites (Fig. 19.3), pointing to significant regional variation. This may be the result of the greater consumption of legumes.
FIG. 19.3. δ13C and δ15N values on human bone collagen from Neolithic sites in Greece. Open symbols are coastal sites, solid are inland. Note the grouping of coastal Alepotrypa with the inland sites. Error bars show two standard errors.
(data from Papathanasiou 2003).
Stable nitrogen isotope values from most of the Mediterranean sites discussed here tend to be low compared with those observed in north-west Europe, suggesting that the Mediterranean diet may have relied less on animal protein. For example, there is a shift (Δ15N) of only 1.6‰ in average δ15N values between herbivores and humans at Le Crès (Le Bras-Goude 2009), which can be contrasted with an average difference of c. 5‰ at the inland site of Hazleton North (Hedges et al. 2008), a shift by no means atypical for Britain. The relatively high incidence of lactose intolerance in populations around much of the Mediterranean today is intriguing in this regard (McCracken 1971), and could suggest that milking was not as extensively practised in the Neolithic here as further north (Copley et al. 2005). Alternatively, as already mentioned, the consumption of legumes in the Mediterranean could lower human δ15N values even if equivalent amounts of animal protein were being consumed (Schulting and Hamilton 2012). This again demonstrates the limitations of the stable isotope technique, and the need for a multidisciplinary approach.
INLAND LAKES AND RIVERS
Away from the coasts, stable carbon and nitrogen isotopes have offered fewer dietary insights: most terrestrial food options, wild and domestic, exhibit variation that is at present too subtle to distinguish, though some intriguing possibilities are being explored (Lynch et al. 2008). As seen at Schipluiden, a notable exception is provided by communities living alongside and exploiting productive freshwater/estuarine ecosystems. These can be isotopically variable and complex, but elevated δ15N values can provide an indication of the consumption of fish and/or waterfowl (sometimes, but not always combined with ‘unusual’—whether elevated or depressed—δ13C values). The best documented examples are the Iron Gates region of the Danube, and the Dnieper Rapids of the Ukraine (see Bartosiewicz and Lillie, this volume, and references therein). In both cases there are only relatively small differences between Mesolithic and Neolithic isotopic results, though it must be borne in mind that ‘Neolithic’ contexts in the Ukraine are largely defined by the presence of pottery rather than by subsistence economy. Another notable example is Ostorf in northern Germany, a middle Neolithic cemetery in an area with abundant inland waterways that were clearly being exploited, given the high δ15N average of 13.7 ± 1.0‰ (Lübke et al. 2007).
In most other parts of Europe, the importance of freshwater resources is more difficult to demonstrate. Whilst δ15N values from Neolithic Britain, Ireland, and north-west France are relatively high (c. 9–11‰), they are not sufficiently elevated to suggest the significant consumption of freshwater fish, particularly given the other ways in which such values can be attained (e.g. manuring and/or use of wetland pastures) (Bogaard et al. 2007; Britton et al. 2008).
VARIATIONS ON A THEME
The above examples operate at a very broad scale, potentially glossing over local and regional variation in Neolithic diets. Given the problem of equifinality raised earlier, even isotopically identical diets can result from quite different combinations of domestic and wild plants and animals. This can be dealt with in part by measuring contemporary fauna from the same sites more systematically, and by considering other kinds of information—archaeological, archaeobotanical, zooarchaeological, palynological, osteological, etc.—that can shed further light on how people made their livings.
Less obviously, similar foodways can result in significantly different isotopic signatures, as would occur if animals grazed on plants with different isotopic values (some soils, for example, have higher δ15N values, and hence so do plants and the herbivores reliant on them). Thus human consumers in one region might have elevated δ15N values suggesting the consumption of a higher proportion of animal protein, whereas in reality the diet may be identical to that in a region with lower δ15N values. These are nuances that are only beginning to be appreciated; they come increasingly into play at the smaller scales of analysis that will begin to feature more strongly as the broad patterns are resolved.
A case study in southern England provides a useful illustration. Richards (2000, 2008) found that results from several approximately contemporary earlier Neolithic long barrows and chambered tombs exhibited slightly different isotopic values. This suggests that groups using different mortuary monuments, whilst all primarily farmers, consumed slightly different combinations of plants and animals. More variable results were noted for the large causewayed enclosure of Hambledon Hill, suggesting the presence of people drawn from surrounding communities, each with their variations on the farming theme. This is consistent with how these monuments are viewed, as built and used by a number of local communities. By contrast, individuals from the Hambledon Hill long barrow were suggested to show a narrower range of variation, suggesting they shared a more similar diet. This provides a very coherent and plausible account, and fits in with what we might expect of early farming communities. Similar local small-scale variation in cereal consumptions is also suggested by differences in dental caries rates at Pipton and Pen-y-wyrlod in south Wales (Wysocki and Whittle 2000).
But what has become appreciated recently is that faunal values from Neolithic sites in southern England also demonstrate a degree of inter-site variation, commensurate with that seen in humans, as demonstrated by the accumulating database of measurements on animals and humans from the same sites (Hamilton and Hedges 2011; Hedges et al. 2007, 2008) (Fig. 19.4). In other words, the isotopic differences between humans at different sites may relate to the consumption of the same proportions and types of domestic animals and plants, but with these themselves exhibiting slightly different isotope values (thus the humans from the Hambledon Hill enclosure may represent a coming together of groups from communities living in different environments but sharing similar subsistence patterns, though it should also be remarked here that the variability in the human isotopic data from Hambledon is actually no greater than that seen at a number of other sites). This is a difficult tangle to unravel, and demonstrates some of the difficulties of working at the small scale, though it should be possible to resolve some of these issues with adequate sample sizes and a well-designed research strategy.
A recent example of this approach comes from a study of three LBK sites in central Germany. Stable nitrogen isotope values for the adult humans differ only slightly between the sites: 8.4 ± 0.5‰ at Halberstadt; 8.8 ± 0.5‰ at Derenburg; and 9.0 ± 0.4‰ at Karsdorf. Taken at face value, this suggests a similar reliance on animal protein. However, a comparison with domestic herbivore values from the same sites shows that the mean spacing between humans and herbi
vores (Δ15N) is only 1.6‰ at Derenburg, whereas it is 2.3‰ at nearby Halberstadt, less than 10km distant (Oelze et al. 2011, tab. 3), indicating surprisingly fine-scaled geographic variability. At a wider scale, as Oelze et al. note (2011, 277), Δ15N values at all three sites are considerably lower than those seen in Neolithic Britain (e.g. c. 5‰ at Hazleton North), suggesting quite different farming practices with a greater emphasis on domestic animals in Britain. This does assume, probably legitimately, that the difference is not due to the significant consumption of legumes in Germany, where they are known to have been present (Bogaard 2004), versus their apparent absence in Britain (Jones and Rowley-Conwy 2007).
