Multi-isotope evidence shows extensive free-roaming pig husbandry endured in medieval England

1 Introduction

The 14th and 15th centuries in England were marked by successive demographic and climatic shocks. The Great Famine (1315–1317), linked to sustained precipitation anomalies, caused harvest failure, grain price inflation, livestock mortality, and increased human mortality. The Black Death (1347–1351) removed an estimated half of the population, sharply reducing labour supply (Broadberry et al., 2010). Recurrent plague outbreaks constrained demographic recovery, while cooler and more variable climatic conditions associated with the early phases of the Little Ice Age introduced instability in agricultural yields in some regions (Campbell, 2000). Husbandry practices need to be situated within the wider restructuring of English agriculture in response to the shocks of the 14th century. National-scale agrarian syntheses provide the economic and organisational backdrop against which local husbandry changes, intensification, altered land use, shifting labour costs and changing market demand, can be interpreted (Campbell, 2000). How communities navigated the crises of the 14th and 15th centuries remains a central question. In particular, understanding how core elements such as animal management and land-use practices operated under demographic contraction and climatic variability is critical to assessing patterns of continuity and change in Medieval England.

Pigs are omnivorous and opportunistic feeders capable of exploiting a wide range of resources and therefore landscapes. Under extensive or free-roaming systems, they may forage on woodland mast (e.g., acorns and nuts), roots, invertebrates, carrion, pasture, wetland vegetation or post-harvest stubble. Under more controlled or confined systems, they can be foddered with legumes, cereals, crop residues as well as household refuse and food-processing by-products such as whey or spent grain, and animal or human derived waste. These divergent feeding pathways integrate distinct environmental and trophic inputs into pig tissues. Consequently, while pigs provide a sensitive archive of landscape use and management intensity, their isotopic signatures reflect multiple overlapping carbon, nitrogen and sulfur sources, necessitating a multi-isotope framework to resolve equivocality in interpretations (see Madgwick et al., 2023).

Pannage, the seasonal turning of pigs into woodland to consume acorns and other mast, is often regarded as the characteristic north-west European medieval system. The practice of herding pigs in forested environments in Europe may extend back to at least 5,000 BCE (Rowley-Conwy, 2006). However, some have argued that these practices have been historically overemphasised in the English context (Rackham, 1980; 1986). It has been suggested that, where practised, pannage systems diminished by the 11th and 12th century due to extensive deforestation and subsequent tighter control over forest use (Wiseman, 2000). It has thereby been argued that in the later Medieval period, there was a move to tighter human control of pigs, including more consistent foddering and increased confinement through stying (Drew, 2010, Ervynck and Dobney, 1999, Ervynck and Lentacker, 2007, Halley and Rosvold, 2014, Hamilton and Thomas, 2012). In some regions this shift is identified in the Early Medieval Period (Guiry et al., 2025). Others have argued for the persistence of extensive free-roaming practices for longer (Margetts, 2022, Wealleans, 2013). The transition from forest dwellers to styed urban pigs in Medieval Europe, marks a key transition in the evolutionary history of the European domestic pig (Ervynck et al., 2007). Rather than indicating a uniform transition, the evidence points to uneven adoption of strategies shaped by woodland access, urban demand, and local regulatory capacity (Jørgensen, 2013, Keene, 2012), with significant regional variation (Albarella, 2006). Finally, work on pigs' cultural meanings shows that husbandry operated within value systems that framed pigs as simultaneously useful and problematic, particularly in urban settings where concerns about nuisance and cleanliness were heightened. In later Medieval England this was reflected in urban regulations restricting pig movement and requiring enclosure within sties, limiting their free roaming within towns and their immediate surroundings (Jørgensen, 2013, 2024).

The objective of our study is to determine how pig husbandry and land-use strategies operated in England across the Medieval period and specifically during the 14th and 15th centuries crises, a period marked by demographic collapse and climatic instability. The focus on small-town contexts reflects the study's origin within a broader interdisciplinary investigation of small-town lifeways in Medieval England (Figure 1). We test whether isotopic evidence reflects continued extensive free-roaming system in pig husbandry, increased urban confinement, or mixed and regionally variable regimes. To achieve this, we apply integrated carbon (δ¹³C), nitrogen (δ¹⁵N) and sulfur (δ³⁴S) and strontium (⁸⁷Sr/⁸⁶Sr) isotope analysis.

As pigs exhibit complex and flexible feeding behaviour, δ¹³C, δ¹⁵N and δ³⁴S values reflect both dietary inputs and physiological conditions. We therefore outline the principal drivers of isotopic variation in pigs in the English Medieval context before inferring husbandry strategies.

2 Drivers of bioavailable δ¹³C, δ¹⁵N and δ³⁴S in woodlands, wetlands and towns

Stable δ¹³C, δ¹⁵N and δ³⁴S isotope ratios in animal tissues reflect those of consumed foods, as dietary isotopic signatures are transferred from diet to consumer tissues. Some variation in diet–collagen isotopic spacing may occur due to factors such as diet composition and physiology, but these effects are probably small (Sponheimer et al., 2003, Warinner and Tuross, 2009, 2010). Interpreting isotopic data requires understanding the environmental and anthropogenic processes that shape bioavailable isotope values. This section therefore reviews the key drivers of δ¹³C, δ¹⁵N and δ³⁴S in woodland, wetland and urban environments, which provide the ecological baseline for interpreting the pig isotope data. ⁸⁷Sr/⁸⁶Sr is not discussed here, as it reflects underlying geology and is not subject to substantial biological fractionation; instead, it is primarily used to investigate animal mobility and geographic catchments.

2.1 Carbon

δ¹³C values in pig tissues primarily reflect the carbon isotope composition of dietary resources. Consumption of animal protein can lead to slightly higher δ¹³C values in omnivorous pigs due to minor trophic enrichment during assimilation (ca. 0.5‰ per trophic level) (Caut, 2009; DeNiro and Epstein, 1978).

UK wetlands are dominated by C₃ vegetation, with cordgrass (Spartina maritima) the principal C₄ exception, largely confined to Essex and parts of Suffolk and the south coast (Adam, 1978). Across freshwater–marine gradients these systems generate wide δ¹³C variation. Freshwater wetlands are typically ¹³C-depleted due to strong photosynthetic discrimination (Andrews, 2008; Cloern et al., 2002; Finlay et al., 1999), whereas brackish and marine systems are relatively enriched through mixing with ¹³C-enriched marine bicarbonate (Cloern et al., 2002; Peters et al., 1978), consistent with broader freshwater–marine contrasts (Robson et al., 2016). Pigs mediate these gradients through wetland foraging, including rooting for below-ground tissues and invertebrates, consumption of overground vegetation, and occasional scavenging of freshwater or marine fish (Canright et al., 2024).

Freshwater carbon cycling can generate substantial δ¹³C heterogeneity (Guiry, 2019), as vegetation may utilise atmospheric CO₂, ¹³C-depleted decomposition-derived CO₂ (Bade et al., 2004; Finlay, 2001), or relatively ¹³C-enriched bicarbonate from carbonate weathering (Finlay and Kendall, 2007; Mook et al., 1974). Pigs foraging on marine macrophytes or algae may likewise show relatively elevated δ¹³C values (Raven et al., 2002). δ¹³C also differs between pelagic and littoral zones (France, 1995; Hecky and Hesslein, 1995): pigs exploiting open-water or fast-flowing resources may exhibit more depleted signatures, whereas margin-based foraging may favour ¹³C-enriched littoral values. Such variability can overlap marine signatures (Guiry, 2019). Hydrology and seasonality further modify pig δ¹³C values: floodplain foraging may reflect terrestrial inputs, tidal margins marine carbon, and stagnant systems ¹³C-enriched recycled pools (Bouillon et al., 2011). Summer foraging may yield elevated δ¹³C values, whereas winter foraging may produce relative depletion. As strong swimmers, pigs can cross rivers, lakes and tidal channels and thereby access a range of ecosystems (Ishizuka et al., 2025).

In woodlands, dense canopies generate lower δ¹³C values than open grasslands due to ¹³C-depleted CO₂ and reduced light (Farquhar et al., 1989; van der Merwe and Vogel, 1978). Pigs foraging in closed woodland may therefore show more ¹³C-depleted values than those feeding in open or pasture settings (Drucker et al., 2008). However, pig δ¹³C values will depend on diet composition: mast, nuts, fungi, and other non-photosynthetic tissues are relatively ¹³C-enriched compared to leaves (Cernusak et al., 2009; Guiry et al., 2021a, 2021b; Hamilton et al., 2009; O’Regan et al., 2016), and roots and tubers are less depleted than foliage (O’Leary, 1981). As pigs preferentially consume mast and below-ground tissues, their δ¹³C values are likely to reflect these relatively ¹³C-enriched woodland resources (Madgwick et al., 2012).

2.2 Nitrogen

Bioavailable nitrogen baselines in wetlands are highly dynamic and spatially heterogeneous in δ¹⁵N due to open nitrogen cycling characterised by continual external inputs and fractionating losses. Plant δ¹⁵N values may therefore be weakly coupled to soil nitrogen pools, reflecting uptake from multiple shifting sources (Hu et al., 2022). Redox conditions are central to this variability: under waterlogged, reducing conditions, denitrification preferentially removes ¹⁴N and enriches remaining nitrogen in ¹⁵N, whereas nitrification under oxic conditions produces smaller fractionation effects (Kendall, 1998; Robinson, 2001). Flooded and poorly drained systems thus tend to favour ¹⁵N-enrichment (Hu et al., 2022). Peat development further enhances microbial processing, progressively ¹⁵N-enriching residual nitrogen. Vegetation also modifies baselines, as peat δ¹⁵N varies with botanical composition; nitrogen-fixing taxa such as alder (Alnus glutinosa) and bog myrtle (Myrica gale) may contribute isotopically light atmospheric nitrogen (Krumins et al., 2019). For pigs, rooting in peat-rich, anaerobic, or disturbed wetlands may therefore yield elevated δ¹⁵N values, whereas foraging in nitrogen-fixing systems may produce comparatively lower values. Wetland plants often display elevated δ¹⁵N due to coupled nitrification-denitrification within rhizospheres rather than salinity alone (Guiry et al., 2021a, 2021b), and invertebrates and fish are further ¹⁵N-enriched through detrital and trophic pathways (Deegan and Garritt, 1997; Finlay and Kendall, 2007). Anthropogenic inputs such as intense grazing, manuring, refuse disposal and fertiliser run-off may further elevate baselines (Bogaard et al., 2007; Szpak, 2014). Because δ¹⁵N increases by ca. 3‰ to 4‰ per trophic level (Caut et al., 2009), even modest intake of animal protein can elevate pig δ¹⁵N values, meaning wetland, anthropogenic inputs, and non-coastal environments may overlap isotopically.

In contrast, closed, undisturbed woodlands typically exhibit lower bioavailable δ¹⁵N due to tight microbial recycling (Aber et al., 1998; Högberg, 1997). Enrichment occurs where soils are disturbed; in wood pasture, pig dung and rooting may create localised ¹⁵N-enriched hotspots (Bardgett and Wardle, 2010). Woodland fungi are often relatively ¹⁵N-enriched compared to foliage and may modestly elevate pig δ¹⁵N values (Trudell et al., 2004).

Finally, very low δ¹⁵N values may derive from nitrogen-fixing plants. In medieval Britain, pigs could access these through cultivated legumes (e.g., pea (Pisum sativum), broad bean (Vicia faba), beans (Vicia spp.) or wild pasture taxa such as clover (Trifolium spp.) and hop clover (Medicago lupulina), consumed as fodder, stubble, or grazing resources.

2.3 Sulfur

Sulfur isotope ratios (δ³⁴S) in animal tissues reflect those of consumed foods, and these values are largely determined by the environment in which the food was produced than by food type, with minimal trophic fractionation (Krajcarz et al., 2019; Nehlich, 2015). In terrestrial systems, bedrock and associated weathering, redox, hydrological, and atmospheric processes generate variable δ³⁴S values (c. −7‰ to +8‰) (Tarrant and Richards, 2024).

Marine systems are enriched in δ³⁴S (c. 17‰ to 21‰) due to seawater sulfate, and this signal can extend inland via sea spray, precipitation, and flooding, sometimes overriding local geology (Bataille et al., 2021). In Britain, coastal values can range from 8‰ to 20‰ (Evans et al., 2022; Hemer et al., 2017), and even aerosol deposition alone can elevate terrestrial signatures (Guiry and Szpak, 2020). Elevated δ³⁴S in pigs may therefore indicate coastal or marine-influenced foraging, though variability in marine and brackish systems can complicate distinction between marine and freshwater fish intake (Aguraiuja-Lätti et al., 2025, Guiry et al., 2025, Nehlich et al., 2014).

Freshwater wetlands differ markedly from both terrestrial and marine systems. They are isotopically dynamic sulfur systems, with δ³⁴S baselines varying in response to redox conditions and hydrology (Guiry et al., 2022; Lamb et al., 2023; Stevens and Wexler, 2025). Strongly negative plant and herbivore values (−30‰ to −25‰) are commonly documented (Guiry et al., 2022; Lamb et al., 2023), indicating that pigs foraging in freshwater wetlands may exhibit markedly depleted δ³⁴S signatures. In wetlands, microbial sulfate reduction and sulfide oxidation alter sulfur pools, producing heterogenous δ³⁴S values (Stevens et al., 2022; Wexler and Stevens, 2025). Hydrological regimes play a central role (Wexler and Stevens, 2025). In flow-through systems with continuous groundwater sulfate input, microbial reduction generates isotopically light sulfides that plants incorporate, producing persistently low δ³⁴S values. In standing-water wetlands with limited sulfate replenishment, progressive reduction enriches residual sulfate and alters sulfide pools over time, generating highly variable plant signatures from ³⁴S-depleted to ³⁴S-enriched. Following drainage, oxidation of stored low-δ³⁴S sulfides can produce extremely depleted sulfate that plants rapidly assimilate, particularly under repeated flood-drain cycles. Where flooded and drained conditions coexist due to topography or hydrology, concurrent reduction and oxidation produce highly variable δ³⁴S baselines. Wetlands commonly exhibit metre-scale δ³⁴S variability (Antler et al., 2019; Bottrell and Novak, 1997; Novák et al., 2005). Variations in iron content, organic matter, and temperature directly affect wetland sulfur cycling, further producing spatially and seasonally variable δ³⁴S signatures. Further, wetland-adapted plant species may access isotopically light sulfur under flooded conditions, whereas plants following drainage more often reflect oxidised sulfate sources (Rolando et al., 2024). For pigs, δ³⁴S values in freshwater wetlands may therefore shift with hydrological state, season of foraging, and type of foods foraged.

Human management can further modify sulfur baselines. Marine biofertilisation elevates δ³⁴S and may mimic marine diets, whereas peat manuring can generate ³⁴S-depleted values resembling wetlands (Blanz et al., 2024; Gröcke et al., 2021; Schlütz et al., 2025).

Urban environments introduce an additional layer of complexity when interpreting pig isotopic values. Towns functioned as consumption hubs drawing on regional and long-distance supply networks, including imported grain, fish, and fodder, alongside intensive manuring and waste recycling. As a result, δ¹³C, δ¹⁵N, and δ³⁴S values may reflect provisioning networks rather than the immediate environmental context.

3 Materials & methods

The archaeological and environmental contexts of the sites, zooarchaeological recording procedures, isotope sampling and analytical methods (δ¹³C, δ¹⁵N, δ³⁴S and ⁸⁷Sr/⁸⁶Sr), chronological assignment, herbivore baseline construction, and statistical analyses are described below. The study sites were chosen to represent a broad range of Medieval English towns, spanning contrasting regional, environmental, and economic settings. We focused on small towns, which have historically received relatively little attention but offer potential because of the way that they are situated at the interface of the urban-rural divide (Dyer, 2003).

3.1 Glastonbury (Somerset)

Pig remains (n = 9) analysed derive from excavations at Bove Town and Silver Street, dating from the 10th to late 15th century (Figure 2A). Glastonbury, laid out by the late 12th or early 13th century around its Abbey, functioned as a service centre and market for its estates, with evidence for metalworking, pottery and tile production, and tanning. At Bove Town evidence for suburban ironworking and ceramic production was recovered, while Silver Street lies in the town core near the abbey precinct (Ellis, 1982; Hollinrake and Hollinrake, 2018). The town sits on the low-lying Somerset Levels, a peat- and alluvium-rich floodplain subject to seasonal flooding and high groundwater, supporting wet grassland, fen, reedbeds, and woodland. The cool-temperate, humid climate is largely influenced by terrestrial inputs. The Levels overlie Triassic mudrocks (⁸⁷Sr/⁸⁶Sr: 0.7085–0.7155), bordered by Carboniferous limestones of the Mendips and Triassic sands of the Quantocks, with Jurassic mudrocks to the south and east grading into the Cretaceous Chalk of Wiltshire (based on excel data from Evans et al., 2022).

3.2 Grimsby (Lincolnshire)

Pig remains (n = 7) analysed derive from excavations at Cartergate, dating from the 12th–16th century. Grimsby, a small medieval port on the Lincolnshire coast beside the Humber Estuary, lay within mineral-rich marshlands and estuarine wetlands shaped by tidal inundation, river input, rainfall, and early embankment. The climate is cool-temperate with strong maritime influence. Historical evidence indicates Grimsby experienced economic decline in the later Middle Ages following the rise of the port of Hull, coastal change, and the reorientation of the Sea trade toward London. Population and trade contracted, though the town remained a fishing centre, with archaeological evidence suggesting reduced land-use intensity. At Cartergate, 12th-century plots show intensive backyard activity and redevelopment until a 15th-century downturn (Rigby, 1993; Rowe, 2011). The town and coast overlie Cretaceous carbonate bedrock (⁸⁷Sr/⁸⁶Sr: 0.7075–0.7101) (based on excel data from Evans et al., 2022). Estuarine areas extend into the Humberhead Levels on Triassic mudrock, while to the south the Lincolnshire Wolds give way to the Central Lincolnshire Valleys and Lincolnshire Fens on Jurassic mudrock. Historically, the Fens comprised freshwater and brackish wetlands with seasonal flooding, permanent meres, and limited cultivation on higher and reclaimed ground.

3.3 Huntingdon (Cambridgeshire)

Pig remains (n = 9) analysed derive from excavations in Huntingdon town centre, dating from the 11th–14th century. Situated at a crossing of the River Great Ouse on the main London – York route, Huntingdon had become a substantial regional centre by the 13th century. Excavation and documentary evidence indicate a later medieval downturn, with reduced land-use intensity in central plots and formerly settled suburbs reverting to agriculture by the 15th century (Clarke, 2004, 2006; Jervis et al., 2025). Huntingdon lies within the low-lying Bedfordshire and Cambridgeshire Claylands, characterised by ancient woodland and heath over Jurassic mudrock (⁸⁷Sr/⁸⁶Sr: 0.7081–0.7114) (based on excel data from Evans et al., 2022). The landscape is bordered by the East Anglian Chalk and lies near the Jurassic mudrock of the Cambridgeshire Fens. Hydrology is strongly seasonal, with winter flooding and persistently high groundwater. The climate is cool-temperate and inland, dominated by terrestrial atmospheric inputs.

3.4 Maldon (Essex)

Pig remains (n = 4) analysed derive from excavations at Maldon Bus Station, 143–147 High Street, and The Rose & Crown, dating from the 12th–16th century. Maldon lies on low-lying terrain at the head of the Blackwater Estuary within the Northern Thames Basin, characterised by wooded claylands and the creeks, mudflats, salt marsh, and reclaimed grazing marsh of the Greater Thames Estuary. To the north, the landscape grades into the South Suffolk and North Essex Claylands. These areas are underlain by Cenozoic mudrocks (⁸⁷Sr/⁸⁶Sr: 0.7081–0.7102), transitioning to Cretaceous carbonate bedrock (0.7075–0.7101) in the East Anglian Chalk and Chilterns (based on excel data from Evans et al., 2022). The region has a cool-temperate maritime climate with relatively low rainfall, strong seasonal flooding, and persistent marine aerosol input. Founded in the 10th century, Maldon became one of the most important medieval towns in Essex. Later medieval evidence indicates piecemeal contraction, abandonment, and redevelopment, followed by post-crisis prosperity marked by new building activity. Unlike many towns in this region, Maldon did not develop a textile industry, instead focusing on its estuarine harbour. The Bus Station site lay at the towns periphery and appears to have reverted to cultivation by the 15th century, while the High Street sites show clear evidence of later medieval redevelopment (Brooks and Wightman, 2010; Ennis, 2015, 2016; Jervis et al., 2025).

3.5 St Albans (Hertfordshire)

Pig remains (n = 9) analysed derive from excavations at 12 Chequer Street and Checquer Street, dating from the 13th to late 15th century. St Albans lies in the River Ver valley on the margin of the Chilterns, underlain by Cretaceous carbonate bedrock (⁸⁷Sr/⁸⁶Sr: 0.7075–0.7101) (based on excel data from Evans et al., 2022). The well-drained chalk supported woodland, grazing, and arable use, while the valley floor comprises wetter alluvium. To the north and east, the landscape rises onto the Hertfordshire Plateau, formed on Cenozoic mudrock and historically characterised by woodland, pasture, and wet heath. The climate is cool-temperate, relatively dry, and dominated by terrestrial atmospheric inputs. Developing as a market town under the control of its Abbey and positioned on a major route to London, St Albans prospered through grain trade to the capital. Excavations on Chequer Street reveal later medieval rebuilding along the street frontage, with multiple malting ovens indicating household involvement in brewing (Niblett and Thompson, 2005).

3.6 Reading (Berkshire)

Pig remains (n = 9) analysed derive from excavations at the Oracle and Friar Street, dating from the 12th–15th century. Reading lies in the broad Thames Valley at the confluence of the Rivers Thames and Kennet, underlain by Cenozoic mudrock (⁸⁷Sr/⁸⁶Sr: 0.7078–0.7114) (based on excel data from Evans et al., 2022). The surrounding landscape comprises floodplains, river terraces, and gently undulating clay lowlands typical of the Thames Basin. To the north and west, higher ground of the Chilterns and Berkshire Downs is formed on Cretaceous carbonate bedrock, while to the south the Thames Basin heaths overlie Cenozoic sand–silt deposits. Founded in the 12th century following the establishment of its abbey, Reading became a prominent regional centre, benefiting from riverine trade links with London and the sheep pastures of the Berkshire Downs. The town prospered as a centre of cloth production. Samples derive from a domestic site on Friar Street in the town centre and from excavations on the Kennet floodplain at the Oracle, an area used for industrial activity (Astill, 1978; Ford et al., 2013; Ford and Ford, 2005).

3.7 Romsey (Hampshire)

Pig remains (n = 8) analysed derive from excavations at Church Street, dating from the 13th–15th century. Romsey lies on the floodplain of the River Test within the Hampshire Lowlands. The underlying geology comprises Cenozoic sand-silt (⁸⁷Sr/⁸⁶Sr: 0.7084–0.7104) and mudrock (0.7081–0.7102) (based on excel data from Evans et al., 2022). To the northeast the Hampshire Downs and to the northwest Salisbury Plain are formed on Cretaceous carbonate bedrock. Similar Cenozoic sediments extend west into the New Forest and south across the South Coast Plain. The climate is cool-temperate maritime with moderate rainfall and no direct tidal influence. Originating as a market centre at the gates of its early medieval abbey, Romsey expanded significantly in the later Middle Ages. Cloth production, particularly dyeing, was central to its prosperity, supported by proximity to Southampton and Winchester and their wider trade networks. Excavations at Church Street revealed a complex sequence of later medieval domestic and industrial activity (Jervis, in press).

3.8 Zooarchaeological recording of samples

Pig mandibles and maxillae were recorded to assess taphonomy, dental wear, biometrics, and sex (Supplementary Data Sheet S1). Taphonomic criteria included weathering (Behrensmeyer, 1978), gnawing and burning (Lyman, 1994), fragmentation (Outram, 2001), and bone surface preservation (Ruebens et al., 2023). Measurements were taken to 0.01 mm using iGaging OriginCal IP54 callipers following von den Driesch (1976) and Payne and Bull (1988). In total, 504 measurements were recorded from 57 pigs (Sus) (Supplementary Data Sheet S1). Data were converted to log-size index (LSI) values (Meadow, 1999) using Payne and Bull’s (1988) wild boar standard (Kızılcahamam, Turkey) and an internal site mean (Evin et al., 2023). Where multiple measurements were available, mean LSI values were calculated. Analysis focused on WA and WP molar dimensions (Supplementary Data Sheet S1) to provide a robust proxy for size variation while minimising wear effects. Eruption and wear stages followed Grant (1982) and Lemoine et al. (2014), and sex was assessed from canine morphology (Mayer and Brisbin, 1988). Specimens were also examined for enamel hypoplasia (Dobney and Ervynck, 1998).

3.9 Temporality of sampled tissues

In pigs, mandibular M1 crown formation begins ca. 3 weeks before birth and completes by ca. 3 months in domestic pigs and ca. 4 months in wild boar; mandibular M2 formation begins ca. 45–50 days after birth in domestic pigs and ca. 64–70 days in wild boar, completing by ca. 6.5–7 months and ca. 9 months respectively (Emken et al., 2023, 2024). Dentine mineralises shortly after matrix secretion and slightly ahead of enamel, whereas enamel undergoes prolonged maturation during which most isotopic incorporation occurs (Dean, 1998; Emken et al., 2021; Frémondeau et al., 2012; Zazzo et al., 2010). As weaning may occur several weeks to several months after birth, bulk M1 and M2 enamel and dentine primarily reflect post-weaning juvenile diet, with slight temporal offsets between tissues. Both maxillary and mandibular M1 and M2 were included in our study, as eruption timing differences between mandibular and maxillary teeth in pigs are small and tooth-specific, and are unlikely to result in meaningful isotopic differences at the temporal resolution examined here (Tucker and Widowski, 2009).

The pig mandible shows rapid growth and remodelling in the first five months of life, followed by slower secondary remodelling in adulthood (Lin et al., 2014; Powell et al., 1973). As skeletal elements can exhibit intra-element isotopic variability (de Gruchy et al., 2024), and different mandibular regions remodel at different rates, some inter-specimen offsets may occur due to sampling location. However, these are expected to have a much smaller effect on isotope ratios than age-at-death. Bone isotope values represent moving dietary averages from months to several years before death, with shorter integration windows in younger individuals. Age-at-death in this study ranges from 8 months to 6 years (Supplementary Data Sheet S1).

3.10 δ¹³C, δ¹⁵N & δ³⁴S analysis of bone and dentine collagen

For sulfur, carbon and nitrogen isotope analysis, sample preparation and collagen extraction was undertaken at the Cardiff University BioArchaeology (CUBA) laboratory. The collagen-extraction protocol followed a modified version of the Longin method (1971). Dentine samples (ca. 30–300 mg) were cut from maxillary and mandibular predominantly M2s, and where not available from P3, P4, M1 or M3, using a diamond saw. Sample length from the Root-Enamel Junction (REJ), width, and tooth crown height were recorded (Supplementary Data Sheet S1).

Dentine samples were mechanically abraded using a diamond burr to remove surface contamination. For bone, 300–1,900 mg of mandibular or maxillary bone from the same individuals as the sampled teeth was collected from each specimen using a diamond rotary wheel, with sampling locations varying between individuals as they were selected from areas of recent breakage to minimise destruction. The cortex was abraded using a burr to remove adhering contaminants. Dentine and bone samples were then demineralised in 8 ml of 0.5M HCl at 4°C. Demineralised specimens were thoroughly rinsed in deionized water and then gelatinized in a pH3 solution of HCl at 70°C for 48 h. The supernate containing the soluble collagen was collected using an 8 μm ezee-filter (Elkay, Basingstoke) and transferred to a polypropylene test tube for freeze-drying. For δ¹³C and δ¹⁵N bone collagen isotope analysis, 0.75 mg (±0.1 mg) of collagen was weighed per sample, with 20% analysed in duplicate.

Stable carbon (δ¹³C), nitrogen (δ¹⁵N), and sulfur (δ³⁴S) isotope ratios of tooth dentinal collagen were measured using a Delta V Advantage continuous-flow isotope ratio mass spectrometer coupled to an IsoLink elemental analyser (Thermo Scientific) at the Scottish Universities Environmental Research Centre (SUERC), following established protocols (Sayle et al., 2019). Samples were combusted at 1020°C in the presence of oxygen, and resulting gases (N₂, CO₂, SO₂) were separated by gas chromatography prior to isotopic analysis. Isotope values were normalised using IAEA reference materials USGS40 and USGS41a for carbon and nitrogen, and in-house and international standards (IAEA-S-2, IAEA-S-3, USGS88) for sulfur. Results are reported relative to VPDB, AIR, and VCDT standards. Analytical precision was ±0.17‰ for δ¹³C and δ¹⁵N and ±0.44‰ for δ³⁴S, based on repeated measurements of gelatin standards.

Stable carbon (δ¹³C) and nitrogen (δ¹⁵N) isotope ratios of bone collagen were analysed using continuous flow-elemental analyser-isotope ratio mass spectrometry (CF-EA-IRMS) for δ¹³C and δ¹⁵N stable isotope composition. Isotope analysis was undertaken at the School of Earth and Environmental Sciences at Cardiff University using a Flash 1105 elemental analyser with ThermoFinnigan Delta V advantage. Carbon and nitrogen stable isotope ratios are given as δ¹³C and δ¹⁵N, relative to internationally defined standards Vienna Pee Dee Belemnite (VPDB) for carbon, and atmospheric air for nitrogen, expressed in units per thousand (‰). Stable carbon and nitrogen ratios were measured in accordance with international standards (Bond and Hobson, 2012). Carbon and nitrogen isotope ratios were calibrated against in-house caffeine (laboratory grade, 98.5%, Acros Organics, lot A0342883) and supermarket gelatine standards, which are calibrated against IAEA-600 (δ¹³C and δ¹⁵N), IAEACH-6 (δ¹³C) and IAEA-N-2 (δ¹⁵N). The 1σ (n = 35) standard reproducibility was ±0.11 for δ¹³C and ±0.11 for δ¹⁵N.

3.11 ⁸⁷Sr/⁸⁶Sr analysis

Initial sample preparation for ⁸⁷Sr/⁸⁶Sr analysis was undertaken at Cardiff University BioArchaeology (CUBA) laboratory. Enamel samples (c. 10–70 mg) were cut from teeth using a diamond saw. Dental enamel samples were mechanically abraded using a diamond burr to remove surface contamination and dentine and ultrasonicated in deionized water and dried prior to weighing. The sampled enamel surface was attempted to be standardised through only sampling mandibular or maxillary M2 of pigs (Sus). The lobe and side sampled were recorded, along with the length (measured from the REJ) and width of the enamel removed (Supplementary Data Sheet S1). Tooth wear and initial crown height were also recorded (Supplementary Data Sheet S1). Enamel was sampled close to the REJ.

Sample pretreatment was conducted at the Cardiff Environmental Laboratory for Trace element and Isotope Chemistry (CELTIC). Enamel samples were digested overnight in 8M HNO₃ on a hotplate at 120°C. Following digestion, enamel pretreatment followed the ⁸⁷Sr/⁸⁶Sr extraction protocol in Scorrer et al., (2021) Approximately 100 μl of pre-cleaned Eichrom Sr-Spec resin is loaded into extraction columns. Digested samples were then loaded into the resin columns, matrix elements, including Ca and trace Rb, were removed in several washes of 8M HNO₃ before Sr was eluted and collected in 0.05 M HNO₃. Samples dried overnight on a hotplate at 120°C, then the process was repeated for a second time for the effective removal of any remaining traces of Ca. Once dry, purified samples were re-dissolved in 2% HNO₃ prior to isotopic analysis. Procedural blanks and standards were processed alongside samples to monitor contamination and analytical consistency.

Strontium isotope ratios were measured using a Thermo Fisher Scientific Neoma MC-ICP-MS/MS equipped with an ESI Apex Omega desolvating nebuliser. Instrument tuning was optimised to ensure stable and intense Sr signals. Analyses were conducted using a sample–standard bracketing approach with the certified strontium isotope standard NIST SRM 987. Isotopes ⁸⁸Sr, ⁸⁷Sr, and ⁸⁶Sr were measured in static multi-collection mode, while ⁸⁵Rb was monitored to correct for isobaric interference from ⁸⁷Rb. Data were acquired with 200 cycles of 4 s integrations. Instrumental mass bias was corrected by internal normalization to the invariant ⁸⁸Sr/⁸⁶Sr ratio of 8.375209, and corrections were applied for Rb and Kr interferences where necessary. Final ⁸⁷Sr/⁸⁶Sr ratios were calculated relative to the bracketing standard. Analytical accuracy and precision were monitored through repeated measurements of NIST SRM 987, and results are reported as ⁸⁷Sr/⁸⁶Sr ratios with uncertainties expressed as 2 SE.

3.12 Chronological assignment

The samples used in this study are currently dated based on pottery typology and therefore span chronological ranges (Supplementary Data Sheet S1). To facilitate inter-site comparability, samples were binned by calculating the midpoint of each assigned date range. These midpoint values were then used to classify samples into three chronological phases: pre-crises (<14th century), crises (14th and 15th centuries), and post-crises (>15th century).

3.13 Data & statistical analysis

All data processing, statistical analyses, and figure generation were performed using R (version 4.5.0). The R scripts used in this study are publicly available at: https://github.com/GwenMaurer-arch/ENDURE/tree/main/medieval_pig_husbandry_2026.

The isotope statistical analysis can be found in Supplementary Data 2. Data distributions were initially explored using ridge plots, boxplots with overlaid individual data points, and Q-Q plots for each site and isotope (Supplementary Data 2). Normality of isotope distributions was assessed using Shapiro–Wilk tests in combination with visual inspection of Q-Q plots. Homogeneity of variance between groups was evaluated using median-centred Levene's tests. Where assumptions of normality and homogeneity of variance were met, differences between groups were assessed using one-way analysis of variance (ANOVA). In cases where homogeneity of variance was violated, Welch's ANOVA was applied. Where normality assumptions were not met, differences between groups were assessed using Kruskal–Wallis tests. Post hoc comparisons were conducted using Tukey's HSD tests following ANOVA, Games–Howell tests following Welch's ANOVA, or pairwise Mann–Whitney U-tests following Kruskal–Wallis tests. For Mann–Whitney comparisons, p-values were adjusted using a Bonferroni correction to control for Type I error. Statistical significance was assessed at α = 0.05.

Dentine-bone isotopic offsets were calculated as the difference between bone and dentine collagen values (bone-dentine), representing dietary differences between early life (dentine) and later life (bone). Spearman's rank correlation was used to test for relationships between age-at-death category and dentine-bone isotope shifts. This analysis can be found in Supplementary Data Sheet S1, Table D and E. Two measures of the shift were calculated: the raw difference (bone-dentine), which preserves whether isotope values increase or decrease through life, and the absolute difference (bone-dentine), which measures the magnitude of dietary change regardless of direction. Interpretations focus on the absolute difference.

3.14 Herbivore baselines

Cattle and sheep dentine (mandibular and maxillary M2) were analysed from each site to establish local herbivore isotopic baselines. For each site, mean δ¹³C and δ¹⁵N values were calculated using all analysed herbivore samples and used as reference values for comparison with pigs from the same site. Sample sizes and calculated mean δ¹³C and δ¹⁵N values are presented in Table 1 and Supplementary Table C in Supplementary Data Sheet S3. Temporal variation in herbivore δ¹⁵N values between the 12th and 16th centuries was assessed statistically. Statistical testing found no significant differences in δ¹⁵N values across centuries (p = 0.43), with post hoc analysis also non-significant. This indicates no temporal change in herbivore δ¹⁵N values and supports the use of pooled site-level means as baseline reference values (Supplementary Data Sheet S3). The herbivore data provide a comparative baseline against which the degree of herbivory and omnivory in the analysed pigs is assessed.

Table 1. Summary statistics for δ¹³C, δ¹⁵N and δ³⁴S analysis in dentine and bone collagen across the analysed sites.

4 Results

4.1 Demography and biometry of samples

The analysed tooth samples from each site show distinct age-at-death profiles, though these reflect only the sampled material rather than complete assemblages (Figure 3A and Supplementary Data Sheet S1). The analysed samples from Glastonbury are dominated by older individuals (>52 months), those from Grimsby by juveniles (8–12 months), those from Huntingdon and St Albans by subadults (12–30 months), and those from Maldon by adults (30–52 months), with no juveniles present. The analysed samples from Reading exhibit a balanced distribution across age categories, while those from Romsey are dominated by subadults with some juveniles but no older individuals. Overall, 98% of dental measurements of the analysed specimens fall below the wild-sized reference (LSI = 0), clustering around −0.1 and indicating predominantly domestic-sized pigs (Figure 3B). The only specimen exceeding the wild threshold is GLAST-S-006 (subadult). Scaling to a pooled internal mean reveals significant inter-site size differences (p = 0.03; see Supplementary Data Sheet S1) between our analysed samples. Sex determinations were limited (n = 9: six males, three females; Supplementary Data Sheet S1).

4.2 Isotope analysis

4.2.1 Carbon & nitrogen

Collagen preservation was good in both dentine and bone (Guiry and Szpak, 2021), with C:N ratios of 3.1–3.4 (dentine), 3.1–3.5 (bone), C:S ratio of 217–598 (dentine) and N:S ratio of 67–187, %C of 35.9–43.5 (dentine) and 17.3–51.3 (bone), %N of 13.1–15.2 (dentine) and 6.2–18.8 (bone), and collagen yields of 2.6–24.7 wt% (dentine) and 0.8–28.2 wt% (bone), respectively. None of the dentine collagen samples failed quality control, whereas four bone collagen samples (GLAST-S-008, GRM-S-001, RDG-S-009 and ROM-S-007) failed and where therefore excluded from further analysis (Supplementary Data Sheet S1). The dentine and bone collagen δ¹³C, δ¹⁵N and δ³⁴S values from all sites are summarised in Table 1 and presented in full in Table 2 and Figure 2. The comparison between the analysed pig and herbivore dentine and bone collagen δ¹³C and δ¹⁵N values are summarised in Figure 4. Temporal changes are summarised in Figure 5 and Supplementary Figure S4.

Table 2. Results of stable isotope (δ¹³C, δ¹⁵N, δ³⁴S) and strontium (⁸⁷Sr/⁸⁶Sr) analyses for pigs from the study sites.

At Glastonbury, mean pig dentine δ¹⁵N lies 1.3‰ below the local herbivore baseline, while mean bone δ¹⁵N is likewise lower (−1.6‰) (Figure 4 and Table 1). Mean dentine and bone δ¹³C are slightly higher than those of local herbivores (Figure 4). Dentine and bone δ¹⁵N values show a strong positive correlation between dentine and bone collagen (r = 0.89; R² = 0.80; p = 0.02). In contrast, δ¹³C values show no relationship between dentine and bone collagen (r = −0.05; R² = 0.003; p = 0.92) (Supplementary Data Sheet S1, Table D). Mean dentine–bone offsets are minimal for both isotope ratios (δ¹⁵N = −0.1‰; δ¹³C = + 0.2‰). Spearman's rank correlation between age-at-death category and dentine-bone shift magnitude indicates a moderately negative but non-significant relationship for δ¹⁵N (ρ = −0.68; p = 0.14). No relationship is observed for δ¹³C shift magnitude (ρ = −0.19; p = 0.73) (Supplementary Data Sheet S1, Table E).

At Grimsby, mean dentine δ¹⁵N lies 1.7‰ above the local herbivore baseline, while mean bone δ¹⁵N is only marginally higher (+0.5‰). Mean dentine and bone δ¹³C are slightly ¹³C-enriched relative to local herbivores. Dentine and bone collagen isotope values show weak, non-significant positive relationships for both δ¹⁵N (r = 0.38; R² = 0.14; p = 0.40) and δ¹³C (r = 0.44; R² = 0.19; p = 0.33). Mean dentine-bone offsets are comparatively large for δ¹⁵N (−1.8‰), whereas δ¹³C values show moderate enrichment (+0.6‰). Spearman's rank correlation indicates no relationship between age-at-death category and δ¹⁵N shift magnitude (ρ = 0.06; p = 0.90). A moderate but non-significant positive relationship is observed for δ¹³C shift magnitude (ρ = 0.64; p = 0.12), suggesting greater dietary shifts among older individuals.

At Huntingdon, mean pig dentine δ¹⁵N lies below the local herbivore baseline. Mean bone δ¹⁵N values are similarly lower (−1.1‰). Mean dentine δ¹³C is comparable to the herbivore mean, whereas mean bone δ¹³C is slightly lower (−0.2‰). Dentine and bone collagen δ¹⁵N values display a moderate positive correlation (r = 0.71; R² = 0.51; p = 0.03). A similar positive relationship is observed for δ¹³C (r = 0.66; R² = 0.43; p = 0.05). Mean dentine-bone shifts indicate decreasing δ¹⁵N values (−1.3‰) and enrichment in δ¹³C (+0.7‰) between early and later life. Spearman's rank correlations show moderate positive but non-significant relationships between age-at-death category and shift magnitude for both isotope ratios (δ¹⁵N: ρ = 0.50; p = 0.20; δ¹³C: ρ = 0.57; p = 0.14), suggesting somewhat greater dietary change among adult individuals.

At Maldon, all dentine and bone δ¹⁵N values lie below the local herbivore baseline. Mean dentine and bone collagen δ¹³C values are enriched by approximately 0.7‰ relative to local herbivore means. Dentine and bone δ¹⁵N values show no relationship between tissues (r = −0.22; R² = 0.05; p = 0.78). In contrast, δ¹³C values exhibit an extremely strong positive correlation (r = 1.00; R² = 1.00; p < 0.01), indicating strong continuity between early and later life. Mean dentine–bone offsets are small for both isotope ratios (δ¹⁵N = −0.6‰; δ¹³C = +0.1‰). Spearman's rank correlations indicate no relationship between age-at-death category and isotope shift magnitude for either δ¹⁵N (ρ = −0.32; p = 0.68) or δ¹³C (ρ = 0.32; p = 0.68), suggesting no consistent age-related pattern in dietary change.

At Reading, mean dentine δ¹⁵N lies ca. 1.7‰ above the local herbivore baseline. Bone δ¹⁵N similarly exceeds the baseline by ca. 1.4‰ on average. Dentine and bone δ¹⁵N values show a weak positive relationship (r = 0.28; R² = 0.08; p = 0.50). In contrast, δ¹³C values display a stronger positive correlation (r = 0.76; R² = 0.57; p = 0.03). Mean dentine–bone shifts indicate a slight decrease in δ¹⁵N values (−0.3‰) and enrichment in δ¹³C (+0.6‰). Spearman's rank correlations indicate no relationship between age-at-death category and shift magnitude for δ¹⁵N (ρ = −0.31; p = 0.45) or δ¹³C (ρ = 0.28; p = 0.51).

At Romsey, the mean dentine δ¹⁵N value lies slightly below the local herbivore baseline by approximately 0.4‰. Mean dentine collagen δ¹³C values for pigs correspond closely to the local herbivore mean, whereas mean bone collagen δ¹³C values are comparatively higher (+0.6‰). Dentine and bone δ¹⁵N values display a strong positive correlation (r = 0.84; R² = 0.70; p = 0.04. In contrast, δ¹³C values show no relationship between dentine and bone collagen (r = 0.12; R² = 0.01; p = 0.83). Mean dentine–bone offsets are none for δ¹⁵N (0.0‰) and minimal for δ¹³C (+0.5‰). Spearman's rank correlations indicate no relationship between age-at-death category and isotope shift magnitude for δ¹⁵N (ρ = −0.13; p = 0.80) or δ¹³C (ρ = 0.39; p = 0.44).

At St Albans, mean dentine δ¹⁵N lies ca. 0.6‰ below the local herbivore baseline. Mean bone δ¹⁵N is similarly lower (−0.5‰). Dentine and bone δ¹⁵N values show a weak positive relationship (r = 0.42; R² = 0.17; p = 0.30). In contrast, δ¹³C values exhibit a strong positive correlation (r = 0.73; R² = 0.54; p = 0.04). Mean dentine–bone shifts indicate slight enrichment in both isotope ratios (δ¹⁵N = +0.2‰; δ¹³C = +0.7‰). Spearman's rank correlations indicate no relationship between age-at-death category and isotope shift magnitude for δ¹⁵N (ρ = 0.00; p = 1.00) or δ¹³C (ρ = −0.15; p = 0.72).

Statistical analyses reveal significant inter-site variation in dentine δ¹⁵N values but limited isotopic variation through time. Significant differences in dentine δ¹⁵N values were observed between several sites (Supplementary Data 2). Post hoc assessment indicates significantly different dentine δ¹⁵N values at Grimsby and Huntingdon compared with Glastonbury (p < 0.001). In addition, Romsey and St Albans differed significantly from Grimsby (p < 0.021), and both sites also differed from Huntingdon (p < 0.045). No other pairwise comparisons were statistically significant (Supplementary Data 2). Dentine δ¹³C values did not differ significantly between sites (p = 0.32). Bone collagen δ¹³C (p = 0.10) and δ¹⁵N (p = 0.10) values likewise did not differ significantly across sites. Post hoc assessments detected no significant differences in dentine δ¹³C, bone δ¹³C and δ¹⁵N values between sites (Supplementary Data 2). Temporal comparisons likewise indicate limited isotopic variation. No significant differences were detected in dentine δ¹⁵N (p = 0.13) and δ¹³C (p = 0.61) values or bone δ¹⁵N (p = 0.44) and δ¹³C (p = 0.35) values between centuries. Post hoc assessments also detected no significant differences in dentine δ¹³C, bone δ¹³C and δ¹⁵N values between centuries (Supplementary Data 2).

4.2.2 Sulfur

Dentine δ³⁴S values differed significantly between sites (p < 0.001) (Supplementary Data 2). Post hoc assessments indicated that Huntingdon significantly differed from Glastonbury (p = 0.01), Reading (p = 0.01), Romsey (p = 0.02), and St Albans (p = 0.002). No other pairwise comparisons were significant before and after Bonferroni correction.

Significant tests indicated no difference between δ³⁴S values across phases (p = 0.45) (Supplementary Data 2). Post hoc assessments likewise did not identify any statistically significant differences between the 12th–16th century (Supplementary Data 2).

4.2.3 Strontium

The ⁸⁷Sr/⁸⁶Sr ratios from all analysed pigs are shown in Table 2 and Figure 6. ⁸⁷Sr/⁸⁶Sr ratios are further contextualised with δ³⁴S values in Figure 7. There is no change in regional catchment areas between 11th and 16th century at any of the sites analysed, even though this is compounded by small sample sizes (Figure 8).

5 Discussion

The multi-isotope evidence indicates marked regional variation in medieval pig husbandry, but little systematic change through time. Across seven English towns spanning different ecological and economic settings, pigs were managed through varying combinations of landscape-based foraging and urban provisioning. At Glastonbury, Huntingdon, Maldon, Romsey and St Albans, diets were predominantly herbivorous, whereas pigs from Grimsby and Reading were more omnivorous, indicating sustained access to animal-derived foods. These differences persist across the pre-crisis, crisis and post-crisis phases, suggesting that pig husbandry remained regionally structured despite the demographic and environmental disruptions of the 14th and 15th centuries.

The key interpretive issue is whether these predominantly plant-based diets reflect extensive foraging or controlled foddering with plants. Herbivory alone cannot distinguish free-ranging access to landscape resources from provisioning with plant foods. That distinction instead emerges from the combined interpretation of δ¹³C, δ¹⁵N, δ³⁴S and ⁸⁷Sr/⁸⁶Sr data, and from comparison with archaeobotanical evidence for fodder plants, crop-processing by-products and urban waste streams. Archaeobotanical evidence, except for dung analysis, does not provide a direct record of pig diet, but it is useful in assessing whether systematic foddering is a plausible explanation for the isotope results. At sites with predominantly herbivorous diets, assemblages are largely composed of domestic refuse, background arable material, and low-level crop-processing residues (e.g., chaff and weed seeds), offering little firm evidence for concentrated fodder resources. In contrast, at Grimsby and, to a lesser extent, Reading, archaeobotanical remains point to urban waste streams and possible fodder by-products, consistent with more confined forms of pig management within towns.¹

At Glastonbury, pigs are predominantly herbivorous, with only rare evidence of omnivory, consistent with limited opportunistic intake of animal protein during free-roaming foraging. The broad δ³⁴S range also reflects the wide foraging of pigs, along the heterogeneous sulfur baselines of the Somerset Levels, generated by fluctuating freshwater-marine interactions, redox conditions and fine-scale topographic variation. Strontium isotope ratios are consistent with the wider Somerset region and extending into Wiltshire (Evans et al., 2022). This is in line with documentary evidence for the circulation of pigs across Glastonbury Abbey's estates in Somerset and Wiltshire (Keil, 1964). Archaeobotanical assemblages from Glastonbury provide little support for systematic foddering. Although some of these plant remains could have been available to pigs, they are dominated by cereal grains and arable weeds characteristic of mixed domestic waste rather than concentrated fodder deposits, while charred seeds of great fen-sedge (Cladium mariscus) underline the engagement with the nearby wetland environments (Allen, 1993; Farr, 2014; Hart, 2006; Law, 2014; Pearson, 2014; Pelling, 2004).

A similar pattern is evident at Huntingdon, where the data points to extensive Fenland foraging. Highly depleted δ³⁴S values indicate foraging in the wetlands of the Cambridgeshire basin (Lamb et al., 2023). ⁸⁷Sr/⁸⁶Sr ratios are consistent with the Fenlands' low-lying sedimentary geology (Evans et al., 2022). Diets are predominantly herbivorous across all phases, but several individuals (HNT-S-008, HNT-S-009 and HNT-S-007) show markedly elevated δ¹⁵N values, among the highest in the dataset. Given the consistently lower δ³⁴S values, this enrichment is more plausibly explained by freshwater-derived animal protein obtained through wetland foraging on fish or carrion, or through fish and fowl waste associated with human use of the Fenlands, than by a shift to systematic household provisioning (Hallam, 1961). One individual (HNT-S-005) shows particularly low δ³⁴S (−25.8‰) and δ¹³C (−23.6‰) values consistent with foraging in alder carr wet woodland, commonly found along Fenland margins (Andrews, 2008; Cloern et al., 2002; Finlay, 2004; Finlay et al., 1999). Historical evidence likewise emphasises the close integration of Huntingdon and nearby Godmanchester with their agricultural and Fenland hinterlands (Masschaele, 1997; Raftis, 1982). Frequent legumes often purposely grown as pig fodder, especially peas, occur in the archaeobotanical assemblages, but the isotopic data do not support intensive legume foddering to pigs (Clapham, 2006; Fosberry, 2011; Mundin et al., 2006). However, occasional legume foddering cannot be ruled out for pigs at the lower end of the δ¹⁵N range.

Maldon and St Albans are consistent with the same broad pattern of predominantly extensive husbandry. At Maldon, consistently low δ³⁴S values indicate inland rather than coastal husbandry despite the town's estuarine setting, while strontium isotope ratios are compatible with pigs being managed in the inland Essex claylands (Evans et al., 2022). Diets remain herbivorous throughout the medieval sequence and across the life course. The relative absence of chaff in the archaeobotanical assemblage suggests off-site cereal processing and thus reliance on the agricultural hinterland rather than local fodder preparation, consistent with extensive hinterland-based pig management (Branch and Gray, 2008; Ennis, 2009; Vitolo, 2017; Vitolo and Allot, 2016). At St Albans, isotope values likewise indicate largely plant-based diets with only minimal omnivorous input. Strontium isotope ratios are compatible with origins in the wider surrounding region, including the Bedfordshire Greensand Ridge, the northern Thames Basin, the Upper Thames Clay Vales and the Chilterns (Evans et al., 2022). The broad δ³⁴S range indicates that pigs foraged across diverse surrounding landscapes rather than being maintained on plant fodder drawn from a restricted local zone. Archaeobotanical evidence is limited to charred barley grains and a small number of arable weed seeds and provides little further resolution (Murphy, 1993).

Romsey shows a similar dietary pattern but a more spatially restricted sulfur signal. Diets are predominantly herbivorous, with only limited evidence of omnivorous input where present, while ⁸⁷Sr/⁸⁶Sr ratios are consistent with pigs having been raised within the surrounding landscape, including nearby sedimentary formations and the Chalk of the Hampshire and Wiltshire Downs (Evans et al., 2022). The tight but elevated δ³⁴S values indicate a restricted sulfur zone, compatible with either sheltered coastal settings, including wooded areas of the New Forest, or inland chalk landscapes. However, such limited δ³⁴S variability could also be consistent with more restricted foddering practices. Archaeobotanical assemblages were dominated by wheat, barley, oat, and small amounts of rye, which may indicate local use. Charred peas and lentils were also identified (Leivers et al., 2006; Slator and McConnel, 2006; Stevens, 2008). Documentary evidence for the management of livestock on the Bishop of Winchester's extensive estates in Hampshire and Wiltshire demonstrates that the movement of pigs between manors for fattening was a part of demesne estate management (Hare, 2006). Romsey Abbey's estate was considerably smaller but includes land in the New Forest and the chalklands and it is plausible that similar management practices were undertaken here, although at a smaller scale.

The contrast with Grimsby and Reading is marked, because in these towns the isotope and archaeobotanical evidence point more clearly towards provisioning within urban waste economies. At both sites, pigs are substantially more omnivorous throughout life, indicating sustained access to animal-derived foods rather than occasional supplementation alone. This matters because it implies a recurrent rather than incidental source of ¹⁵N-enriched dietary inputs. At Grimsby, marine inputs can be excluded, as δ¹³C, δ¹⁵N and δ³⁴S values fall outside marine ranges despite the town's maritime location. The isotopic pattern instead suggests consumption of terrestrial household refuse, including meat scraps, freshwater resources and other urban food waste. Strontium isotope ratios are consistent with the Lincolnshire Central Valleys and Wolds, suggesting that pigs were raised in the hinterland before entering the settlement, although one individual shows strontium isotope ratios and sulfur isotope values compatible with wetter lowland environments, possibly the Lincolnshire Fenlands (Evans et al., 2022). Archaeobotanical evidence supports this interpretation, indicating abundant mixed organic waste within the settlement, including cereal chaff, straw, bran and other food refuse, together with deposits tentatively identified as stable manure (Fryer, 2018; O’Brien, 2011; Tann and Palmer-Brown, 2009). There is also evidence for taxa indicative of wetlands and saltmarsh environs (Carrott et al., 1994). In this case, the archaeobotanical and isotopic evidence do not merely coexist; they point in the same direction, indicating a provisioning regime in which pigs were maintained on concentrated urban by-products (e.g., animal derived food waste and manure) rather than relying primarily on open foraging.

Reading shows a similar but less tightly organised pattern. Pigs are predominantly omnivorous throughout life, indicating sustained access to urban food waste rather than short term dietary enrichment. Strontium isotope ratios suggest origins within the surrounding region, especially the chalk landscapes of the Chilterns and Berkshire Downs, as well as the Thames Valley and the Upper Thames clay vales, while the broad δ³⁴S range indicates inputs from multiple contrasting environmental zones (Evans et al., 2022). This variability could be reflective of in-town provisioning based on pigs, feed resources, or waste streams drawn from a geographically diverse hinterland. Archaeobotanical assemblages support this interpretation. Routine domestic food waste predominates, but possible fodder, including hay-meadow plants and probable fodder vetch, indicate that a range plant resources were also available within the town (Druce, 2007; Hunter, 2005; Palmer, 2020; Pelling, 2013). Historical evidence indicates that Reading lay within a wider regional pork-producing zone, consistent with strong links between urban demand and hinterland supply (Murphy and Galloway, 1992).

Documentary evidence² from 777 Crown seizure lists offers a limited but useful indication of the scale and distribution of pig-keeping in later medieval small towns: pigs appear in 91 lists, and piglets in 30, with a predominantly eastern distribution and notable concentrations in Lincolnshire and the Thames valley (Supplementary Figure S5). Despite the limitations of these records, this pattern supports the isotopic evidence in identifying Reading and Grimsby as being situated in regions with significant urban pig husbandry (Chick et al., forthcoming).

Regional variation in pig husbandry is reinforced by life-course isotope evidence, which distinguishes between broadly continuous feeding histories and more structured shifts in management through life. At Maldon, Romsey and St Albans and to a lesser extent Glastonbury, small dentine–bone offsets and the absence of clear age-related patterning suggest that pigs consumed broadly similar resources from early to later life. This continuity is consistent with extensive husbandry systems in which pigs remained embedded within surrounding landscapes throughout life. The absence of an age-related pattern suggests that dietary variation was not structured by life stage but instead reflects relatively loose and opportunistic management. This broadly accords with documentary evidence for the keeping of pigs within small towns which shows that whilst pigs were the most abundant animals possessed by urban households, these households represent only a small proportion of the urban population (Chick et al., forthcoming). Critically, our data suggest that where households are recorded as possessing only one or two pigs that these are likely to have been raised in surrounding areas, being brought to town immediately prior to slaughter. At Glastonbury, slightly greater variation among subadults suggests that juvenile foraging may have been somewhat more variable before diets stabilised in their foraging areas, reinforcing the idea of loosely managed early-life access to landscape resources. Huntingdon and Grimsby both show population-level dentine-bone differences, indicating that diet changed over the life course at these sites. At Huntingdon, however, this pattern appears to reflect increasing dietary variability with age, suggesting that resource use became more variable with age, perhaps as pigs matured and exploited a wider range of wetland environments. Whereas at Grimsby the decline in δ¹⁵N from dentine to bone suggests a more consistent shift away from stronger animal-derived inputs after early life, while δ¹⁵N values are consistent with omnivorous diets in both tissues. Because this pattern is unrelated to age-at-death, it appears to reflect a shared change in feeding conditions rather than gradual individual variation, supporting the interpretation of a relatively structured urban provisioning regime. By contrast, δ¹³C offsets are smaller overall, although their tendency to increase in older pigs may suggest slightly greater change in plant-based dietary inputs among animals kept for longer. Reading occupies an intermediate position. Pigs show omnivorous diets in both tissues, and the weak dentine-bone correlation together with the very small mean δ¹⁵N shift suggests little overall change in protein intake through life, despite individual variation. The lack of any relationship between δ¹⁵N or δ¹³C offsets and age-at-death further suggests that dietary change was not systematically structured by life stage. These patterns suggest that Reading pigs had sustained access to urban food waste throughout life, but within a less tightly structured provisioning system than at Grimsby. Overall, these life-course patterns show that isotopic differences between dentine and bone correlated with age-at-death data do not merely add biographical detail, but help identify different modes of pig management, from relatively continuous landscape-based foraging to more organised urban feeding systems.

6 Conclusion

We challenge the idea that pig husbandry in later medieval Europe followed a simple trajectory from extensive free-roaming systems to increasingly confined urban management. Evidence from seven English sites instead points to the long-term continuity and co-existence of different pig husbandry strategies between the 12th and 16th centuries.

Carbon, nitrogen, and sulfur isotope data indicate both extensive free-ranging systems and urban provisioning regimes, with no consistent chronological shift across the sites examined. Rather than indicating a general shift away from extensive management, the results highlight the stability of site-specific husbandry practices through time. Strontium isotope data similarly show that the catchment areas and regional supply networks associated with pig management remained stable from the 12th to the 16th century. This suggests that the spatial organisation and landscape use of pig production and movement was not substantially reorganised over time. This study represents the first application of a multi-isotope approach to Medieval English pig husbandry, demonstrating how isotopic evidence integrated across multiple tissues can resolve subtle variation in management, mobility, and provisioning.

That these patterns persisted through the social, economic, and demographic disruptions of the 14th and 15th centuries is especially significant. Despite these crises, small towns in relation to pig husbandry and landscape use appear to have been marked by resilience rather than fundamental transformation. Overall, our findings support a model of medieval pig husbandry in England defined by long term resilience and regional variation.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fearc.2026.1832112/full#supplementary-material

Isotope dataset. Supplementary material at the source

Isotope statistics. Supplementary material at the source

Herbivore statistics. Supplementary material at the source

Stable isotope values for pig dentine and bone collagen by site and phase (pre-, crises, and post-crises); points show individual pigs and bars show group medians. Supplementary material at the source

Distribution of pigs and piglets in lists of goods seized by the Crown from felons residing in small towns recorded in the records of the Escheator, dating 1357–1479. Data gathered from The National Archives E153 series by Joe Chick. Supplementary material at the source

Footnotes

Edited by: Juliette Waterman, University of Reading, United Kingdom

Reviewed by: Magdalena Blanz, University of Oxford, United Kingdom

Sean Doherty, Queen's University Belfast, United Kingdom

Footnotes

¹The archaeobotanical study is ongoing as part of Kate Evetts’ doctoral at the University of Leicester, Understanding the plant economy of small towns in medieval England (c.1200–1600): A synthetic analysis and interpretation of archaeobotanical evidence.

²This evidence comprises lists of goods seized by the crown from felons, recorded in the records of a royal official known as the Escheator between 1357–1479. Data are derived from analysis of National Archives series E153. See Chick et al., forthcoming.