Introduction

Switzerland in the transition between Roman times and the Early Middle Ages

Most of modern-day Switzerland was part of the Roman Empire during the first five centuries AD. From the time of Augustus, up until the middle of the 3rd century AD, this was a prosperous region, thanks to the protection of well-defended and distant Imperial borders and a peaceful Romanization of the local population (Flutsch et al. 2002). The Romans established numerous civilian settlements, e.g. Iulia Equestris (modern-day Nyon), Aventicum (Avenches), and Augusta Raurica (Augst/Kaiseraugst), and built a network of roads connecting them, allowing for the integration of this region into the imperial economy. A great number of rural settlements was established, including villae rusticae, and agriculture and animal husbandry flourished during Roman times, in the context of a market-oriented economy (e.g. Ebnöther and Monnier 2002; Vandorpe et al. 2017). Roman agriculture in this region was largely based on a system of villae rusticae and smaller farmsteads that produced goods for urban towns and military settlements (e.g. Matter et al. 2016).

At the end of the Roman period, however, the importance of towns (e.g. Berger 2012) and the number of villae rusticae declined (Schwarz 2011), during a transition period that was characterised by a political crisis, human migration and climatic deterioration (e.g. Marti 2000; Windler and Fuchs 2002; Büntgen et al. 2016). From the 3rd century onwards, archaeological evidence shows that substantial transformations occurred (Flutsch et al. 2002; Windler et al. 2005), with important changes in settlement patterns and dynamics. For instance, many villae were abandoned or re-structured into smaller settlements, urban centres declined (Schwarz 2011; Berger 2012), and Germanic people and influences arrived from the North, when Franks and Alamanni crossed the river Rhine (Marti 2000, 2008, 2009). From the 4th century onwards and during the earliest centuries of the Early Middle Ages, this region shifted between different political entities (Windler and Fuchs 2002), although after the 5th century, the scarcity of documentary sources has hindered historical research on these processes. All of these aspects had important effects on farming in this region.

Zooarchaeological research

Romans had a great interest in improving cattle husbandry and, as a result, cattle size increased considerably, a process that has been documented in most of the Roman Empire (e.g. Lauwerier 1988; Grant 2004; Albarella et al. 2007; Pigière 2017; Frémondeau et al. 2017; Valenzuela et al. 2017; Trentacoste et al. 2021; Grau-Sologestoa et al. 2022), including Switzerland (Breuer et al. 1999; Grau-Sologestoa et al. 2021). To a certain extent, a similar process has also been suggested for pig in this region, although less marked than for cattle (Breuer et al. 2001; Grau-Sologestoa et al. 2021). Livestock size increase in Roman times occurred in the context of “improvement” of agricultural practices in order to increase agrarian production, which was largely aimed at specializing the production and at supplying a growing market (e.g. Grant 2004; Groot and Deschler-Erb 2015; Groot 2017; Pigière 2017), perhaps through the importation of large cattle breeds (e.g. Schlumbaum et al. 2006). One possibility is that intensive (i.e. high level use of labour, in comparison to the land area) animal husbandry was practiced in order to achieve greater productivity, and, in this context, a close control of feeding times and foodstuffs, as well as keeping animals in stables or sties, were possible factors behind livestock size increase in the Roman period. However, some researchers have suggested that Roman farming practices were rather extensive (MacKinnon 2001; Albarella et al. 2019; Lodwick et al. 2020). In Roman Switzerland, while some animal stables have been found, and pathological marks on pig feet bones have been interpreted as possible evidence of them being tied with a rope, material culture evidencing that cattle were allowed to graze outdoor has been identified as well (Deschler-Erb et al. 2002: 168–169). The evidence is still too fragmentary to be able to generalise that Roman animal husbandry practices in the study region were intensive, but they were certainly operated at a large scale, oriented to the market.

On the other hand, during the Early Middle Ages, agricultural activities were mainly aimed at subsistence, and therefore were carried out at a smaller scale. The shift from large-scale to small-scale agricultural production after the collapse of the Roman Empire and the fragmentation of early medieval economies are phenomena that have been thoroughly investigated by historians and archaeologists (e.g. Wickham 2005 and 2009; Hodges 2012; Quirós-Castillo 2013; Bianchi 2015). Zooarchaeological research all over Europe has highlighted a generalised process of livestock size reduction during this transition (Yvinec 1991; Stephan 2008; O’Connor 2010; Hammon 2011; Holmes 2014; Grau-Sologestoa 2015; Salvadori 2015; Rizzetto et al. 2017; Rizzetto and Albarella 2022), and the same pattern has been observed in Switzerland (Breuer et al. 1999, 2001; Rehazek 2010; Marti-Grädel 2012, 2013; Frosdick 2014; Grau-Sologestoa et al. 2021). Different reasons have been proposed to explain this size decrease, such as climate deterioration, a relaxation of the Roman taxation system, a genetic bottleneck due to the termination of livestock trade, a deliberate selection of smaller animals for breeding, and, of key importance for this paper, a change in the scale of animal husbandry practices and different feeding techniques or foodstuffs. In Switzerland, during the Early Middle Ages, there was a reduction of the cultivated areas for cereals, in contrast with an increase of grassland and forests (e.g. Wick 2015; Akeret et al. 2019), which may have been used to feed animals. If less land was being used for cultivating crops, then more land would have been available for feeding livestock. These changes might have meant that early medieval livestock was perhaps more often kept with some degree of freedom for roaming and grazing. This was the case in England, where early medieval pigs were quite likely taken to woodland areas to be fed on roots, acorns and beech mast (Albarella 2006: 77). In other European areas, recent research suggests that, in fact, herbivores and omnivores were largely kept free-range during the Early Middle Ages (e.g. Sirignano et al. 2014; García-Collado 2020). Looser feeding and breeding habits, together with a relatively more mobile lifestyle (in terms of seasonal movements or free-range practices), are possible explanations behind the livestock size decrease throughout the Early Middle Ages (Grau-Sologestoa et al. 2021), but, at the moment, this is just a hypothesis.

Research on stable isotopes

Dietary patterns of domestic animals are relevant for understanding animal husbandry practices. Stable isotope analysis of bone collagen is a well-known quantitative technique for reconstructing the diet, i.e. the plant and animal protein content in the diet, of past humans and animals. This is based on the principle that food sources contain different isotopic signatures that travel along the food chain from food to consumers and from prey to predators (e.g. Schwarcz and Schoeninger 1991; Van Klinken et al. 2000). Bone preserves dietary signatures that reflect the nature and timing of incorporation of dietary constituents used in their construction. Bone remodels over the life of animals; therefore, the stable isotope signatures in bone collagen reflects a long-term dietary average. Stable carbon and nitrogen isotope analysis in bone collagen disproportionally reflects dietary protein relative to lipid and carbohydrate contributions (Krueger and Sullivan 1984; Ambrose and Norr 1992). Stable carbon isotope values from bone collagen are useful for distinguishing between diets based on C3 vs. C4 plants (Van der Merwe and Vogel 1978; O’Leary 1988), and between foods from marine and terrestrial environments (Chisholm et al. 1982). Due to the canopy effect, woodland and land diets can be discriminated in animal skeletal remains (Van der Merwe and Medina 1991; Drucker et al. 2008). Stable nitrogen isotope values become elevated by ca. 3–5‰ at each step ascending a food chain (Ambrose and DeNiro 1986; Hedges and Reynard 2007). For this reason, they help differentiate between herbivorous, omnivorous and carnivorous diets. Suckling of young animals can also be identified due to an increase of isotopic signatures by one trophic level (Balasse and Tresset 2002). 15N enrichment can further be the result of the manuring of crops (Bogaard et al. 2007) and of environmental processes in the soil like seasonal flooding, aridity and salinity (Stroud 2022 citing Handley et al. 1999; Hartman and Danin 2010; Heaton 1986; Yousfi et al. 2010).

In recent years, the analysis of the carbon and nitrogen isotope composition of animals is becoming increasingly popular, in particular in relation to reconstructing past animal husbandry practices and wild game management (e.g. Castaños et al. 2014; Price et al. 2017; Ventresca Miller and Makarewicz 2018; Dai et al. 2019; Trentacoste et al. 2020; Jones et al. 2020; Groot et al. 2020; Gillis et al. 2020; Price et al. 2020; Takken Beijersbergen et al. 2021; Alagich et al. 2022). Isotopic approaches to animal diet in the Roman and early medieval periods, however, are still very rare (an exception is, for example, Sirignano et al. 2014), and many works deal with animal diet only secondarily, when using them as baselines to investigate human diet (e.g. Knipper et al. 2013; Iacumin et al. 2014; López-Costas and Müldner 2016; Lubritto et al. 2017; García-Collado 2020). Only few works have focused on carbon and nitrogen data from animals of the Roman period (e.g. O’Connell et al. 2019), the Middle Ages (e.g. Richards et al. 2006; Hamilton and Thomas 2012; Fisher and Thomas 2012; Hammond and O’Connor 2013; Halley and Rosvold 2014), or on both these periods diachronically (e.g. Müldner and Richards 2007; Fuller et al. 2012; Müldner et al. 2014).

Stable isotope analysis for studying diet is well developed in archaeological research on Swiss materials. Like in other countries, this research tradition is better established for analysing human remains (e.g. Lösch et al. 2013; Moghaddam et al. 2016, 2018; Knipper et al. 2017; Peacock et al. 2019; Bourbou et al. 2019; Siebke et al. 2020; Varalli et al. 2021; Gerling and Doppler 2021). In respect to animals, important research has been developed regarding freshwater fish (Häberle et al. 2016a, b) and prehistoric cattle and red deer (Doppler et al. 2015, 2017; Gerling et al. 2017a, b; Reitmaier et al. 2017 but no work has been done on Roman and early medieval faunal remains so far.

Studies in isotope geochemistry looking at changes in animal husbandry practices for the periods under study can be considered in their infancy. At the moment, it can only be speculated that a change in the scale of animal husbandry practices led to changes in livestock feeding practices, and these were at least partly responsible for the size decrease of livestock.

Aims of the paper

The present paper was conceived to test the hypothesis (e.g. Grau-Sologestoa et al. 2021) that feeding practices for cattle and pigFootnote 1 changed (from large herds to keeping fewer animals, and from being very closely controlled, with stable foddering, to being raised in a more free-range way) between the Late Roman times and the Early Middle Ages. This hypothesis assumes that Roman animal husbandry practices were more intensive than early medieval ones, which were in comparison more extensive. In economics, intensive agriculture is a system of cultivation using large amounts of labour and capital relative to land area, while intensive animal husbandry is any concentrated, confined animal growing operation for meat, milk or egg production located in pens or houses wherein the animals are provided with externally sourced feed (Britannica 2017). Extensive agriculture, on the other hand, is a system of crop cultivation using small amounts of labour and capital in relation to area of land being farmed and, normally, it produces a lower yield per unit of land (Britannica 2011). Extensive livestock production is an animal farming system characterised by a low productivity per animal and per surface, using small amounts of inputs, capital, and labour compared to the farmed land area, essentially based on grazing (permanent grasslands, natural pastures, transhumance, etc.).

By integrating zooarchaeological evidence and stable isotope geochemistry, this paper aims answer the following research questions:

  • What was the diet of cattle and pig in Roman and early medieval times like, and what can this tell us about husbandry practices for these livestock species?

  • Can changes in feeding practices explain the livestock size changes for this transitional period?

Materials and methods

The site of augusta Raurica

Augusta Raurica (Colonia Augusta Rauracorum) was founded in the second half of the 1st century BC, and is located in north-western Switzerland, between the modern-day municipalities of Augst (Kanton Basel-Land) and Kaiseraugst (Kanton Aargau) (Fig. 1). It is one of the most well-known Swiss archaeological sites, including its environment (Wick 2015) and zooarchaeology (e.g. Schibler and Furger 1988; Deschler-Erb 1991a, 1992; Vogel Müller and Deschler-Erb 2006; Deschler-Erb et al. 2021). As part of the Roman province of Germania Superior, from the end of the 1st century AD, Augusta Raurica was a prosperous commercial trading centre and a dynamic city in the region. At around 200AD, the population of Augusta Raurica is estimated to have been 9000–12,000 people (Bossart et al. 2006). In the central years of the mid-3rd century AD, some parts of the city were destroyed (the possibility of an earthquake has been suggested, although disputed – Schatzmann 2013) and the Germanic tribe of the Alamanni arrived (Favrod 2002; Schwarz 2011). In the late 3rd century, the Limes Germanicus, the frontier that marked the northern border of the Roman Empire, retracted to the so-called “knee” of the Rhine (Flutsch et al. 2002), roughly following the present-day borders of France, Germany, and Switzerland. As part of the new frontier defences, the Roman military forces built then a fortress, named Castrum Rauracense, located just north of the ancient city and on the south bank of the Rhine river, with an associated civilian settlement located in its vicinity (a suburbium), that lasted at least until the early 7th century. In the 7th or 8th century, the bishopric seat of the region was moved from Augusta Raurica to Basel, causing the latter to gain more importance in the subsequent centuries, until nowadays.

Fig. 1
figure 1

Map of Augusta Raurica with the location of all the sites included in the analysis. Image by Claudia Zipfel

Full size image

A transition from agricultural surplus production to subsistence agriculture has been proposed for north-western Switzerland between Late Roman times and the Early Middle Ages (e.g. Akeret et al. 2019; Grau-Sologestoa et al. 2021). Archaeobotanical data and pollen diagrams also attest to this transition, based on the diversification of the cereal spectrum, the absence of imported foods (Vandorpe et al. 2017; Akeret et al. 2019), and the increase of tree species (particularly beech, oak and alder) at the expense of open-land species (Hadorn 1994; Wick 2015).

Zooarchaeology

In this paper, a review of available zooarchaeological literature on Augusta Raurica has been performed. Two types of zooarchaeological information have been collected:

  • Number of Identified Specimens (NISP) of cattle, sheep/goat and pig, in order to look at the taxonomic frequencies of the main domesticates. Assemblages with less than 100 NISP (adding cattle, sheep/goat and pig) have been excluded.

  • Biometric dataFootnote 2 (of cattle and pig), in order to understand the role of cattle and pig husbandry at the siteFootnote 3. The biometrical analysis has been performed using the log-ratio (or LSI) technique, a size-index scaling technique that transforms logarithmically (in our case, with logarithm base 10 – LSI10) the ratio between measurements and a set of standards measurements (in this paper, the standard measurements are a Hinterwälder female cow, 17 years old, and a 2–3 year old male wild boarFootnote 4). Table 1 summarises the composition of the biometrical dataset that has been analysed.

Table 1 Sites in Augusta Raurica (Augst-Kaiseraugst, Switzerland) used for biometrical analysis, references and chronological information
Full size table

Stable carbon and nitrogen isotope analysis

Cattle (Bos taurus) and pig (Sus domesticus) bones (see Table 2 for details on the skeletal elements), recovered in contexts dated between Late Roman times and the Early Middle Ages from different sites in Augusta Raurica (Theater NW-Ecke, Kastelen, Implenia Mühlegasse, Gasthof Adler and Jakobli-Haus), have been sampled (Fig. 1; Table 2). Samples were taken from well-dated contexts, spanning between the mid-2nd and early 7th centuries, chronologically as precise as possible, in order to allow for diachronic comparisons, and were grouped into four different phases for the analysis:

  • Phase 1: 150–250 (mid-2nd -mid-3rd c. AD).

  • Phase 2: 250–300 (second half of 3rd c. AD).

  • Phase 3: 300–400 (4th c. AD).

  • Phase 4: 450–620 (mid-5th -early 7th c. AD).

Phase 1 represents the peak of the Roman settlement of Augusta Raurica (Berger 2012), when it was one of the main urban centres in the region. In the middle of the 3rd century, large parts of the city were destroyed, and therefore, Phase 2 represents the moment of the demise of the upper city and “enceinte réduit” of the settlement. In Phase 3, the population (a mixture of Roman military and civilians) concentrated in the newly built Castrum Rauracense. In 401 AD, the Roman army left the settlement, and during Phase 4, the population was a mixture of Roman and Germanic people. All additional information about the samples is provided in the Supplementary Materials (SM3).

15 samples per taxon and phase were taken where possible, for a total of 117, of which 111 were successful in collagen extraction. Whenever possible, we tried to minimise the possibility that the sampled remains belonged to the same individual by choosing the same skeletal elements. Moreover, we made sure that the remains belonged to adult animals, either through the epiphyseal fusion state or by the mandibular wear stage, to avoid isotopic signatures of suckling individuals.

Table 2 Sampled sites for isotopic analysis, references and chronological information. * Number of samples successfully analysed for stable isotope analysis
Full size table

Selected bones, stored at the museum of Augusta Raurica, were transferred to the IPNA/IPAS, Department of Environmental Sciences, University of Basel, for sampling and collagen extraction. Sample preparation followed Longin (1971) with modifications as described in Knipper et al. (2017). Compact bone portions were cut and the surfaces removed. Between 500 and 800 mg of sample were demineralized in 10 ml of 0.5 N HCl at 4˚C and room temperature, rinsed to neutrality and reacted with 10 ml of 0.1 M NaOH for 24 h at 4˚C, rinsed to neutrality and gelatinized in 4 ml of acidified H2O for 48 h at 70˚C. Insoluble particles were separated using EZEE filter separators, and the collagen frozen and lyophilized. About 0.8 mg of collagen per sample was weighed into tin capsules and analysed with an INTEGRA2 EA-IRMS instrument (Sercon Ltd., Crewe, UK) in the laboratories of the Aquatic and Isotope Biogeochemistry at the Department of Environmental Sciences, University of Basel. Samples were measured in duplicates. Raw nitrogen and carbon isotope data were blank-, linearity, and drift-corrected and then normalized to the AIR-N2 and VPDB (Vienna Pee Dee Belemnite) scales by means of two-point calibrations based on an in-house EDTA standard and IAEA-N-2 or EDTA and IAEA-CH-6, respectively. The resulting nitrogen and carbon isotopic compositions are reported in δ-notation as δ15N and δ13C in per mil relative to AIR-N2 and VPDB (Vienna Pee Dee Belemnite), respectively. Reproducibility of internal and external standards was better than ± 0.25‰ for δ15N and better than ± 0.1‰ for δ13C.

Statistical analyses

The biometrical analyses have been complemented with statistical analyses performed in SPSS. The normality tests (Kolmogorov-Smirnov and Shapiro-Wilk) suggested that biometrical data was not normally distributed and, therefore, statistical significance was tested using a Kruskal-Wallis H test, and pairwise comparisons were done using a Mann-Whitney U-test post hoc (adjusted with Bonferroni correction). The statistical analysis of the isotopic data was performed in R, with a parametric test of Tukey multiple comparisons of means.

The results of these analyses are available in the Supplementary Materials (SM4 and SM5). The confidence levels were set at 95% and the significance of the p values are reported as follows: ***: ≤ 0.001; **: ≤ 0.01; *: ≤ 0.05.

Results

Zooarchaeological analysis

Table 3 shows the NISP of cattle, pig and sheep/goat at different sites in Augusta Raurica. Cattle predominates in most sites of Roman chronology, with proportions between 50 and 70% in the majority of cases, although in Theater NW-Ecke, cattle represent more than 90% of the assemblage, and there are also a few exceptions in which either pig (Insulae 28, 30, and 31, Mansio, Jakobli Haus − 150–200), or sheep/goat predominate (Latrinengruben and Amphitheater). This Roman pattern in which cattle was the more frequent species, changed during the Late Antiquity: in the first centuries of the Early Middle Ages, pig is the predominant taxon in all sites (with proportions of 40–50%), followed by cattle and sheep/goat with similar proportions. When all the sites are combined and ordered chronologically (Fig. 2), it becomes apparent that this shift in taxonomic proportions is clear by the 5th -6th centuries, although the decrease in cattle might have started already in the 4th century. This means that cattle predominated during phases 1 to 3, while pig was more common in phase 4.

Table 3 NISP of cattle, pig and sheep/goat at different sites in Augusta Raurica
Full size table
Fig. 2
figure 2

Taxonomic frequencies of cattle, pig and sheep/goat in Augusta Raurica, based on sites in Table 3, combined and arranged chronologically. In brackets, NISP of cattle + pig + sheep/goat

Full size image

The biometrical analysis is hindered by the relatively small sample size, which is quite probably behind the low statistical significance of observable differences between some of the phases, but the results are nonetheless interesting. Both cattle (Fig. 3) and pig (Fig. 4) increased in size between the 1st and the 3rd centuries (statistically significant, with p ≤ 0.001 for cattle, and 0.035 for pig). Also, the size of both livestock species remained quite stable during the rest of the Roman period. Cattle size decreased in the 5th-early 7th centuries (statistically significant, with p = 0.002). This is not the case for pig, with average values very similar to those of the preceding phases, although with an increase in variability, with individuals smaller and larger than in the other periods. Although the inclusion of some wild boar might explain some of the large values, it seems quite clear that the size of pig did not decrease as much as cattle. This is in line with previous publications dealing with biometrical data from this region (Breuer et al. 1999, 2001; Marti-Grädel 2012; Grau-Sologestoa et al. 2021).

Fig. 3
figure 3

Box plots showing the LSI10 values of cattle postcranial widths from Augusta Raurica. In brackets, number of measurements. ***: α ≤ 0.001; **: α ≤ 0.01; *: α ≤ 0.05

Full size image
Fig. 4
figure 4

Box plots showing the LSI10 values of pig postcranial widths from Augusta Raurica. In brackets, number of measurements. ***: α ≤ 0.001; **: α ≤ 0.01; *: α ≤ 0.05

Full size image

Isotopic analysis

The stable carbon and nitrogen isotope data of cattle and pig bone collagen are listed in the Supplementary Materials (SM3), and the descriptive statistics are summarized in Table 4. Despite considerable fragmentation rates of the bones, overall collagen preservation was good, with only 6 samples out of 117 not meeting the quality criteria for ancient collagen (Ambrose 1990; van Klinken 1999); these were therefore excluded from further analysis. 111 samples fulfilled the collagen quality criteria.

Table 4 Descriptive statistics of carbon and nitrogen stable isotope data for cattle and pig from Augusta Raurica
Full size table

The direct comparison between the carbon and nitrogen isotope ratios of cattle and pig from Augusta Raurica (Fig. 5) is consistent with differences between an herbivore species (cattle) and an omnivorous one (pig), although pig isotope values show that they were largely following an herbivorous diet. For carbon isotope values, the difference between cattle and pig is greater for phases 3 and 4, that is, for the 4th and 5th -early 7th centuries, and in both cases this difference is statistically significant (p ≤ 0.001). The level of nitrogen isotope values in pigs is mostly as low as that in the cattle values of the same period, suggesting that their diet was (almost) entirely based on plants. There is however a notable difference between the nitrogen values of cattle and pig from phase 2, the second half of the 3rd century, and it is statistically significant (p ≤ 0.001). There is no significant difference in the diet of cattle and pig of phase 1 (mid-2nd – mid-3rd century), suggesting that both species ate similarly.

Fig. 5
figure 5

Box plots showing the carbon (top) and nitrogen (bottom) isotope values of cattle (red) and pig (blue) in the different occupation periods, in Augusta Raurica

Full size image

Figures 6 and 7 show the average values and 1σ standard deviations of carbon and nitrogen ratios for pig and cattle in comparison to other isotopic data available from two sites located less than 15 km from Augusta Raurica, dated to the Late Iron Age (Basel-Gasfabrik) and the High Middle Ages (Basel-Barfüsserkirche). Figure 6 shows that the pig isotopic ratios were generally quite similar over time, with the exception of phase 2, the second half of the 3rd century, when nitrogen isotope values are relatively high, and in phase 3, the 5th century, with low nitrogen and high carbon values.

Fig. 6
figure 6

Graphical representation of the statistical summary of stable carbon and nitrogen isotope data of pig skeletal remains from north-western Switzerland (Late Iron Age: Basel-Gasfabrik; High Middle Ages: Basel-Barfüsserkirche; others: Augusta Raurica). Centroids (dots for values from Augusta Raurica and triangles for values from Basel) represent the averages (mean values) of each group, and the lines represent 1σ standard deviations

Full size image

Cattle isotopic values (Fig. 7) show a significant overlap between the different phases. However, some trends are visible. The highest nitrogen levels in cattle are documented for the Late Iron Age, and seem to decrease in later periods, reaching their lowest in the High Middle Ages. A similar decreasing trend is observed in carbon isotope values from the Late Iron Age to phase 4 (although more notably between phases 2 and 4), and then an increase in the High Middle Ages.

Fig. 7
figure 7

Graphical representation of the statistical summary of stable carbon and nitrogen isotope data of cattle skeletal remains from north-western Switzerland (Late Iron Age: Basel-Gasfabrik; High Middle Ages: Basel-Barfüsserkirche; others: Augusta Raurica). Centroids (dots for values from Augusta Raurica and triangles for values from Basel) represent the averages (mean values) of each group, and the lines represent 1σ standard deviations

Full size image

When carbon and nitrogen data for each taxon are plotted separately for the periods between the Iron Age and the High Middle Ages (Supplementary Materials – SM1 and SM2), changes between phases are visually clearer, although very few of the statistical tests have resulted as significant, due to the small sample sizes. Cattle and pig carbon isotopic values in north-western Switzerland seem to have followed opposite patterns through time. Late Iron Age and phase 1 cattle have the highest median carbon isotope values, only comparable to the high medieval ones. For pig, the lowest median carbon values are those from Late Iron Age and High Middle Ages, while the highest ones are of the Late Roman period. The decrease in carbon values between the early and the high medieval times is statistically significant for pig (p = 0.008). Cattle nitrogen levels did not substantially change across time, and only the Late Iron Age ones were considerably higher. For pig nitrogen contents, the decrease of the values between phases 1 and 3 is statistically significant (p = 0.01 for the pair Phase 1 – Phase 2, and p < 0.001 for the pair Phase 2 – Phase 3). For the medieval periods, a slight increase in nitrogen values can be observed.

To recapitulate, Table 5 offers a summary of the main results of the zooarchaeological and stable isotope analyses in Augusta Raurica between phases 1 and 4.

Table 5 Summary of main zooarchaeological and isotopic results for Augusta Raurica phases 1–4 (mid-2nd to early 7th cent. AD)
Full size table

Discussion

When all the results of our analysis are considered together in the context of the important changes that Augusta Raurica underwent during the transition from the Roman times to the Early Middle Ages, some interesting issues arise. The zooarchaeological analysis shows a clear shift from an economy largely predominated by cattle in the Roman period (phases 1 to 3) (likely linked to their importance as sources for raw material – Deschler-Erb et al. 2021), to more balanced proportions between the three main domestic taxa, with pig being predominant by the 5th -6th centuries (phase 4). The biometrical analyses have evidenced that the size of cattle and pig increased during Roman times, but both species followed different paths in the periods afterwards: while cattle size decreased in the 5th -early 7th centuries, the average size of pig remained stable, but with increased variability.

Both the continuity and increased variability of pig size between the Late Roman times and the Early Middle Ages might be related to the interbreeding between domestic pigs and wild boars, an issue that has been discussed for other early medieval contexts (Albarella et al. 2019). Our previous work (Grau-Sologestoa et al. 2021) supported this idea, as revealed by a greater decrease of pig teeth size than postcranial bones, since smaller teeth in relation to postcranial bones are a characteristic of wild boars (Albarella et al. 2019: 28; Tecce and Albarella 2020: 121). Interbreeding between wild boars and domestic pigs might have been facilitated by free-range practices and, therefore, the biometrical analysis seems to suggest that such practices were more widespread in the early medieval times than in Roman times.

The isotopic evidence shows changes in livestock management (especially in respect to pig) through time. Augusta Raurica as a Roman city was mainly a consumer rather than a producer site. This means that the cattle and pigs that we have examined isotopically might have come from different herds to be sold at the market, raised in different areas, and kept under different conditions. The co-existence of animals raised free-ranged and those being stalled is expected to perhaps lead to an increased variation in stable isotope values. If our hypothesis was correct, and free-range practices became generalised in the early medieval period, stable isotope analysis should show differences between the Roman and the early medieval samples. Roman feeding practices should be characterised by e.g. higher nitrogen isotope values for pigs as a result of omnivorous animals being fed with/on domestic food scraps, and/or by carbon isotope values due to an (increased) intake of C4 plants through food scraps and/or purposely cultivated fodder. In contrast, lower carbon and nitrogen isotope values would be expected during the Early Middle Ages based on a more herbivorous diet due to free-ranging food availability (as suggested, for example, by García-Collado 2020). Increased variability in the diet can also be expected in animals that are free-ranged, in comparison to those that are kept in stables. If livestock was kept in stables in Roman times, as part of intensive animal husbandry regimes, the breeder would have had a close control of what the animals were eating. In the case of an omnivorous animal, like pig, if they were kept close to humans, a high intake of proteins could be expected, as these animals can be easily fed with human food-scraps. If, in the early medieval period, in turn, animals were raised outdoors and free-range, pigs would show a highly herbivorous dietary signal. However, there are two potential issues with these assumptions, arising from the fact that more experimental data on pig diet is still needed: first, our interpretation of the results of the isotopic analysis assumes that the diet of a free-ranged pig would be similar to that of wild boars which, although often eat mainly vegetables, being omnivorous and opportunistic animals, also eat quite a lot of protein (insects, worms, small vertebrates, carrion, etc.) (e.g. Schley and Roper 2003; Ballari and Barrios-García 2014); and, second, stabled pigs might not have been fed with food scraps at all, and even if they were, how much of these would actually be protein, when human diet was also predominantly composed of vegetables (for example, García Moreno et al. 2022)?

Another important issue is that, while the process of livestock size decrease in the transition to the Early Middle Ages is very generalised in the Western Roman Empire (despite regional differences and timing variations), we do not know much about livestock management practices in different regions, neither during the Roman times, nor the early medieval period, and we do not know if what holds true for one region would be applicable to another. The area for which most research has been carried out is Roman Italy, facilitated by the fact that Roman agricultural writers produced a good amount of literature on the topic. Historical sources (Dion. Hal., Ant. Rom., 3.70; Polyb. 2.15; Strabo 1917, Geogr. 5.1.12) attest extensive pig management in Roman times, taking advantage of pannage in local woodlands (MacKinnon 2001; Trentacoste 2020). This has been suggested by archaeological evidence from Archaic and Imperial Italy (Alldritt et al. 2019; Trentacoste et al. 2020). Despite this, increased interbreeding and hybridization between domestic pigs and wild boars has also been suggested, based on archaeological evidence, for early medieval Rome (Albarella et al. 2019), which might be related to an (even further) extensification of pig management practices in this period.

Despite these limitations, there are some interesting observations in the isotope values of both species:

1) Nitrogen isotope data show no significant changes in manuring levels through time. Nitrogen isotope values between the two species are comparable, except for phase 2 when pigs show significantly enriched δ15N values. Feeding on products from manured fields is one possible explanation for this enrichment. But, more likely, pigs might have been kept in the settlement (an idea supported by the presence of pig perinatal individuals – Deschler-Erb 2002), fed partially on food scraps, including protein leftovers, which would explain the relatively high level of nitrogen. Perhaps the instability caused by the first incursions of Germanic people in the region, the move of the limes to the river Rhine, and the move of the population to the castle of the Upper Town –Kastelen- led to the need to keep the pigs close by (Schwarz 2011; Berger 2012). Cattle δ15N values show a remarkable overlap between the different phases, suggesting that the diet of cattle did not change substantially across time. This is not surprising, considering that these ruminants are herbivores, with all their diet consisting of grass, leaves crops and grains, whether they were kept stalled or not.

2) In respect to carbon isotope values, naturally, all sampled cattle show an isotopic signature characteristic for an herbivorous diet. δ13C values suggest that C3 plants constituted most of their diet. Pigs show comparatively higher δ13C values, which cannot be due to metabolic differences, as it is ruminants that should show enriched carbon values (Hamilton and Thomas 2012: 251). This difference might suggest that millets were perhaps artificially fed to pigs, even though C3 plants constituted the vast majority of their diet. In our study region, naturally occurring C4 plants are relatively rare (Pyankov et al. 2010), and the only cultivated ones in Roman times and the Early Middle Ages were broomcorn milletFootnote 5 (Panicum miliaceum) and foxtail millet (Setaria italica); both are relatively common in archaeobotanical samples of Roman Augusta Raurica (Vandorpe et al. 2017; Akeret et al. 2019; Deschler-Erb et al. 2021) and could have been provided occasionally (probably during the winter) to the animals as artificial, complementary fodder in phases 3 and 4, coinciding with the moment when the population concentrated in the Castrum Rauracense, the population of Germanic origin increased, and the army left the city. Similar patterns of greater consumption of C4 plants in the period between the 5th -8th centuries than in Roman and high medieval times have been observed in the Iberian Peninsula (García-Collado 2020), and evidence for the generalised use of dried fodder to feed stabled sheep and goats in early medieval Catalonia (Gallego-Valle 2022). It is also possible, however, that the higher carbon values observed in pig for the later phases in Augusta Raurica are related to an increase in the consumption of fungi (Hamilton and Thomas 2012; Stroud 2022).

3) There are some indications suggesting slight changes in cattle management through time. These observations are mainly based on the relatively high nitrogen (and carbon) isotope values for Late Iron Age cattle at Basel-Gasfabrik (Knipper et al. 2017). The authors suggested that this might be related to intensive agricultural practices and the import of cereals and animals from the surrounding area, e.g. the fertile Sundgau. The suggested use of cattle as draught animals in agriculture at the site (Stopp et al. 1999) may have led to natural manuring (and in consequence higher δ15N values). Also haymaking and stall keeping of cattle is indicated for Basel-Gasfabrik (Stopp et al. 1999) and may have led to variations in stable isotope values of cattle. The predominance of cattle in zooarchaeological assemblages from these periods might be indirectly related to the higher nitrogen values: larger cattle herds and larger number of cattle per land plot might have resulted in increased manuring. In contrast, the isotopic values of late Roman and early medieval cattle are consistent with animals that might have been kept roaming freely and in smaller numbers, with low nitrogen and carbon isotope values.

4) Environmental change and worsening climate conditions may serve as an explanation for the decreasing trend of cattle carbon isotope values from phases 2 to 4. Cattle, being herbivores, might reflect environmental and climatic conditions better than pigs. Carbon values of plants growing in cold and wet places are lower of those growing in warm and dry areas (Farquhar et al. 1989; Van der Merwe and Medina 1991). Palaeoclimatologists (Büntgen et al. 2011) have documented, in our study region, increased precipitation for the period ca. 280–500 AD, followed by an abrupt decrease of temperatures during the 6th c. This coincides well with the observed decrease in cattle carbon isotope values.

Conclusion

This paper aimed at exploring the previously proposed hypothesis (e.g. Grau-Sologestoa et al. 2021) that changes in livestock management and feeding practices, in the transition between the Roman times and the Early Middle Ages, were related to changes in animal size, due to a possible link between livestock size and diet (Hammond 1960; Widdowson and Lister 1991). Our analyses have revealed important changes in cattle and pig husbandry at Augusta Raurica during this transitional period.

Our proposed hypothesis assumed that Roman animal husbandry practices were more intensive than early medieval ones, which were in comparison more extensive. These are however analytical categories, while the reality of farming and agricultural systems is far more complicated: many farmers would practice a combination of both systems (for example, by leaving their herds to graze during the summer, and stabling them during the winter), or raise an animal species extensively and another one intensively, for instance. In this line, our evidence does not support a clear, generalised and complete change from intensive to extensive husbandry practices, but it does suggest that a slight shift occurred towards more free-range livestock management, even if some animals were still being kept in sties at least part of the time. The signs for greater hybridisation of pig and wild boar and the possible increased consumption of fungi by pigs go in line with the generalisation of pannage as the system to raise pigs in the early medieval period, and the decreasing values of carbon in cattle might be related to forest pasturing. However, these are just some of the possible explanations for the observed trends and the evidence remains difficult to be disentangled.

Although isotopic analyses of archaeological faunal remains have been carried out for quite some time now, our understanding of the complex feeding mechanisms of these animal species and their reflection in the isotopic values, e.g. free-ranged vs. stalled pig, is far from being complete. Data from modern animal feeding and keeping experiments (cf. e.g. Webb et al. 2016 and 2017) would certainly help to better understand isotopic variation in archaeological cattle and pig diets and their relation to influencing factors like food sources and keeping conditions but also metabolism and age. An important direction in respect to future stable isotope studies is the combination of domestic animal and their potential food sources as well as the extension of modern feeding experiments. We hope that our paper contributes to the understanding of these fascinating topics.