Human-animal interactions at Swifterbant

The presence of morphologically domestic cattle and suids could point to animal management in Swifterbant. The fact that the Bos are larger than the cattle at Schipluiden is consistent with findings from central Europe where domestic cattle continuously decrease in size from the Early Neolithic to the Late Neolithic (Manning et al. Reference Manning, Timpson, Shennan and Crema2015b). The predominance of adult cattle at S3 is mirrored in Neolithic assemblages at LBK sites and at Schipluiden, but S3 lacks the young calves found at these sites where cattle were exploited for dairy production (Gillis et al. Reference Gillis2017; Kamjan et al. Reference Kamjan, Gillis, Çakırlar and Raemaekers2020). Cattle could have been kept to adult age for a multitude of reasons, including meat, breeding and feasting. The find of a cattle foetal bone supports a scenario of live cattle being kept at Swifterbant. It is otherwise difficult to draw conclusions from the S3 age profiles due to the small sample size. At S4, due to fragmentation, only one adult and one calf (< 3 months) could be identified (Kranenburg & Prummel Reference Kranenburg, Prummel, Raemaekers and de Roever2020). These sites were, however, likely part of a wider system in this landscape and their seasonal occupation may mean that animals were also kept and slaughtered in other areas, or in other seasons.

If the assumption of local animal husbandry is made, the clustering of cattle and sheep dietary isotopes requires further explanation: two different dietary management regimes or the import of livestock from elsewhere are both possibilities. Phytosociological analysis of the local area demonstrates that the difference in mean δ¹⁵N values between the clusters could not have been caused by natural variation in nutrient availability in the environment and therefore must be due to human intervention (Figure 8). The fact that the two clusters are also represented in the Bos from S4 demonstrates that the two dietary regimes were already in place at the onset of settlement in the Swifterbant region, from 4240–4160 BC onwards.

Figure 8.

Variation in the total amount of nitrogen in the soil (in mg per kg) for the most frequently identified plant communities at Swifterbant arranged from highest to lowest (figure by authors).

One possible explanation for the substantially elevated δ¹⁵N values of the cluster 1 Bos and the three sheep is salt-marsh grazing. Elevated δ¹⁵N values are observed in cattle and caprines grazing in salt-marsh regions in the prehistoric and historic Netherlands (Prummel et al. see item in References) and in Bronze Age Britain (Britton et al. Reference Britton, Müldner and Bell2008). Cattle and red deer from the coastal site of Schipluiden also have elevated δ¹⁵N values (Kamjan et al. Reference Kamjan, Gillis, Çakırlar and Raemaekers2020). Salt marshes were not available in the immediate Swifterbant area. The coast was 50–80km away but represents a terra incognita for this period: due to erosion, there are no preserved coastal sites from the fifth millennium BC. Between the coast and Swifterbant was a brackish, marshy environment that may explain the elevated δ¹⁵N values if transhumance were practised (Vos Reference Vos2015). Salt-marsh grazing would therefore indicate seasonal mobility.

Another explanation is manured pasture or feed (Bogaard et al. Reference Bogaard, Heaton, Poulton and Merbach2007). However, if the arable fields at Swifterbant were manured, it is questionable whether livestock would have grazed on or been fed with the stubble consistently enough to produce such elevated δ¹⁵N values. Current data indicate that the arable fields at Swifterbant were small and therefore probably not suitable for this purpose. Alternatively, the values could be a result of ‘self-manuring’, whereby plants on (sufficiently small) pasture grounds repeatedly used for grazing will obtain elevated δ¹⁵N values due to the accumulation of animal dung, leading to elevated values in the livestock themselves (Shahack-Gross et al. Reference Shahack-Gross, Simons and Ambrose2008; Makarewicz Reference Makarewicz2014). If livestock grazed on the small open plots on the Swifterbant riverbank (Figure 1), this may have occurred. This would also explain the fact that the phytosociological vegetation analyses resulted in the identification of the open environment grassland community Triglochino-Agrostietum nasturtietosum, which persists only under extensive grazing (Sýkora et al. Reference Sýkora, Schaminée, Weeda, Schaminée, Stortelder and Weeda1996; Schepers Reference Schepers2014). Finally, nutritional stress is known to raise δ¹⁵N values but not to the degree established in the Swifterbant animals (Doi et al. Reference Doi, Akamatsu and González2017).

The cluster 2 cattle, in contrast, have notably depleted δ¹³C values. Bone collagen δ¹³C values of between –22.5‰ and –18.5‰ are typical of herbivores foraging on C₃ plants in an open environment (Berthon et al. Reference Berthon, Kovačiková, Tresset and Balasse2018). The δ¹³C values of cluster 2 are instead most consistent with a diet composed significantly of closed-canopy vegetation, which typically produces more negative values (<–22.5‰, Drucker et al. Reference Drucker, Bocherens, Bridault and Billiou2003). Cluster 2 δ¹³C values are comparable with aurochs grazing in the dense forests of Atlantic Denmark, whereas the domestic cattle from the Scandinavian Neolithic generally grazed in open environments (Noe-Nygaard et al. Reference Noe-Nygaard, Price and Hede2005; Gron & Rowley-Conwy Reference Gron and Rowley-Conwy2017). The morphological aurochs from S3 also falls within cluster 2; its diet thus corresponds with the Danish aurochs.

Seasonal foddering with forest vegetation would not deplete livestock collagen δ¹³C values to the extent witnessed in cluster 2; for this, a year-round diet is necessary (Berthon et al. Reference Berthon, Kovačiková, Tresset and Balasse2018). While forest-grazing cattle are not consistent with findings from the Scandinavian Neolithic, there is no reason to assume that this could not have been a management strategy in the Dutch wetlands. Forest-grazing and forest fodder were, after all, cattle husbandry practices used by earlier LBK communities in Europe (Gillis & Zanon Reference Gillis, Zanon, Nicoud, Balasse, Desclaux and Théry-Parisot2021). The Swifterbant area was characterised by forests on the riverbanks, and major parts of the area—in the basins behind the riverbanks—would have been too wet year-round for livestock grazing (Figure 1). Keeping livestock in this region may have necessitated utilising these woods. Alternatively, the cluster 2 cattle may have been acquired from elsewhere (see below).

The presence of very small suids (all adults) at S3 with elevated δ¹⁵N values suggests that there were domestic pigs in the assemblage, particularly given the individual with enamel hypoplasia, while the very large suids probably represent wild boar. Suids with isotopic values similar to those of humans and/or dogs at a site might indicate pig husbandry as loosely managed pigs often forage on human refuse (Rosvold et al. Reference Rosvold, Halley, Hufthammer, Minagawa and Andersen2010; Balasse et al. Reference Balasse, Cucchi, Evin, Bălăşescu, Frémondeau, Horard-Herbin, Stépanoff and Vigne2018). It is also possible that the Swifterbant pigs were kept in the same manner as the cattle and sheep, leading to similarly elevated nitrogen values.

The intermediate size of the suids overall could reflect a pig population with genetic admixture—in other words, hybridisation between wild boar and pigs of Near Eastern descent, a documented phenomenon across Europe by this time (Frantz et al. Reference Frantz2019). It could also represent a mixed assemblage made up of managed domestic pigs, hunted wild boar and hybrids (wild or managed). Such a mix would explain the diversity in isotopic values of suids at S3, which are comparable with the Hardinxveld wild boar overall, and the age profile that is similar to Hardinxveld but contains more piglets.

An alternative scenario is that the morphologically domestic animals were not in fact managed by Swifterbant communities. It has been suggested that early incidences of livestock in northern Europe result either from foragers exchanging with neighbouring farming communities, the raiding of those communities or the hunting of feral herds that escaped from farming settlements (Gron & Sørensen Reference Gron and Sørensen2018). If the cluster 2 cattle were hunted feral cattle, this would explain why they have δ¹³C values comparable to aurochs—although it does not explain why the caprine, which even feral are not typically forest-grazing, also has such low values. The cluster 2 cattle could also have been forest-grazed livestock obtained from other farming communities, though if so it is not currently possible to establish their source. The LBK settlements in the southern Netherlands and Germany date to earlier than Swifterbant. Contemporary farming settlements may have existed downriver to the east, in modern Germany, but current data are too limited to explore this.

A scenario in which livestock are incidentally acquired but not husbanded at Swifterbant would fit with the dates of the emergence of farming in northern Germany and southern Scandinavia, which first occurs after 4000 BC. However, several lines of evidence argue against this. First, the tightly clustered Bos metrics, which suggests a fairly homogeneous assemblage in terms of size, and the scarcity of aurochs fit better with a scenario of cattle management than hunting. The presence of all body parts of Bos and Sus and of a Bos foetal bone also suggests the management of animals on location rather than hunting or just the exchange of butchered remains. This is further implied by the fact that isotopic results from both cranial and postcranial elements fit in the two clusters. Thus, cranial parts found on site—which are less likely to occur through hunting or gift exchange of butchered parts—also derive from animals with distinct diets. Local livestock grazing would also explain the presence of a grassland community that persists only under extensive grazing. The discovery at S3 of several suid coprolites, including two containing cereal remains and one with sheep wool in it, again suggests the presence of live livestock (Kubiak-Martens & van der Linden Reference Kubiak-Martens and van der Linden2022). Finally, the evidence for arable farming testifies to agricultural practices at Swifterbant.

Rethinking the transition

We argue that these lines of evidence make a scenario of local animal management in the period 4240–4050 BC more likely than a scenario of solely exchanging/raiding/hunting livestock. While this would be the earliest evidence for such a practice in northern Europe, and thus inconsistent with the transition narrative in other regions, there is no reason to assume that early farming appeared in the same way at the same time everywhere in northern Europe. Indeed, the Swifterbant Culture already provides evidence for early arable farming unlike anything known in northern Europe for this period (Raemaekers et al. Reference Raemaekers2021). This demonstrates that these communities had already adapted their lifeways to the needs of other species in their care. While the keeping of livestock would have had very specific requirements, there is nothing to suggest these communities did not have the knowledge for such an undertaking, nor that the local environment could not have supported it. In fact, that such dynamic practices should be present in the Swifterbant region is unsurprising considering the richness of resources in this wetland environment (cf. Zeder Reference Zeder2016: 336). The enigmatic dietary regimes of the animals may therefore be the result of management strategies suited to this landscape and its seasonal changes.

The results of this study do, however, leave room for multiple forms of human-animal interactions at Swifterbant. It is plausible, for example, that some cattle and sheep were kept in this region, while some were brought from elsewhere, through transhumance, exchange, raiding or the hunting of feral animals—which could explain the isotopic evidence for two dietary systems. Yet the contrast between earlier Swifterbant Culture sites, where few Bos remains are found, and the presence at Swifterbant of morphologically domestic cattle with managed diets indicates that a change in human-Bos interactions had by this time occurred. The human-suid relationship appears to remain more fluid with an enduring focus on wild boar alongside possible pig management. These results suggest that earlier Swifterbant Culture sites, such as Hardinxveld (5450–4250 BC), could fit with a model where foragers incidentally acquire livestock (Gron & Sørensen Reference Gron and Sørensen2018), but that by 4240–4160 BC in the Swifterbant region communities had assumed control over the diet of domesticated animals. These communities managed their domestic animals and crops while also exploiting natural resources through hunting, gathering, fishing and fowling.

This perspective on the Neolithisation process in the Netherlands has far-reaching implications for our understanding of the Neolithic in northern Europe as a whole. It suggests that sweeping narratives should be reassessed to focus more on local trajectories and regional processes. These results thereby support arguments that domestication should be ‘decentred’ and studied as a multitudinous practice involving complex multispecies relations (cf. Lien et al. Reference Lien, Swanson, Ween, Swanson, Lien and Ween2018). This perspective also has implications for how scholars and the general public view the emergence of agricultural and pastoral societies. There is still the tendency to perceive this as a process whereby farming replaces hunting and gathering, in an evolutionary, linear trajectory (Lavi et al. Reference Lavi, Rudge and Warren2023). The Swifterbant Culture provides a case study that forces us to rethink these narratives and instead investigate the flexibility of prehistoric societies.