A multi‐isotope investigation of diet and subsistence amongst island and mainland populations from early medieval western Britain

Abstract Objectives This is the first investigation of dietary practices amongst multiple early medieval populations (AD 500–1000) from Wales and the Isle of Man using carbon, nitrogen, and sulphur isotope analysis. The analysis will illuminate similarities or differences between the diets and subsistence strategies of populations occupying different geographical regions, specifically those living in marginal coastal regions in comparison to inland populations well‐connected to ecclesiastical centres and high‐status settlements. Materials and Methods One hundred and two human skeletons were sampled for carbon and nitrogen isotope analysis, and 69 human skeletons were sampled for sulphur isotope analysis from nine cemetery sites from western Britain (Isle of Man = 3, southwest Wales = 4, southeast Wales = 2). Thirteen faunal skeletons from St Patrick's Chapel (southwest Wales) were sampled for carbon, nitrogen, and sulphur isotope analysis. Results Human δ13C values range from −19.4‰ to −21.2‰ (δ13C mean=−20.4 ±0.4‰, 1σ, n = 86), and δ15N values range from 9.1‰ to 13.8‰ (δ15N mean = 10.8 ± 0.9‰, 1σ, n = 86). δ34S values range from 1.2‰ to 18.4‰ (δ34S mean = 11.6 ± 4.5‰, 1σ, n = 66). Significant differences were noted between the mean δ13C, δ15N and δ34S values according to geographic region: Isle of Man (δ13C = −20.7 ± 0.4‰, δ15N = 11.4 ±0.6‰, n = 13/86; δ34S mean = 17.1 ±0.6, n = 4/66), southwest Wales (δ13C = −20.5 ± 0.4‰, δ15N = 11.0 ±1‰, n = 32/86; δ34S = 16.1 ± 2.1, n = 21/66), and southeast Wales (δ13C =−20.3 ±0.4‰, δ15N = 10.4 ±0.7‰, n = 41/86; δ34S= 8.8 ±3‰, n = 41/66). Faunal δ13C values range from −23.1‰ to −21.2‰ (δ13C mean= −22.1 ±0.5‰, 1σ, n = 13), and δ15N values range from 6.3‰ to 9.8‰ (δ15N mean = 7.3 ± 1.1‰, 1σ, n = 13). δ34S values range from 4.7‰ to 18.4‰ (δ34S mean= 16.3 ± 3.6‰, 1σ, n = 13). Conclusions The data reveal a reliance on terrestrial protein, however differences are observed between the resource consumption of populations from southwest Wales and the Isle of Man in comparison to the populations from southeast Wales. Populations from the west coast have a marine sulphur signature that reflects their coastal proximity and may also include a reliance on seaweed as a fertiliser/food source. Populations in the southeast were connected to ecclesiastical centres and high‐status settlements and had access to inland‐grown produce. The data add support to the suggestion that δ34S can be used as a mobility indicator.


| I N T R O D U C T I O N
The reconstruction of diet amongst early medieval (5th-11th century AD) populations from Britain has, to date, focused on Anglo-Saxon populations from England. One of the earliest studies undertaken over a decade ago applied carbon and nitrogen isotope analysis to the early Anglo-Saxon cemetery population of Berinsfield, Oxfordshire, and sought to investigate a correlation between diet and status (Privat & O'Connell, 2001). Another study saw the analysis of the late-4th to mid-6th century AD population from Queenford Farm, Oxfordshire and revealed gendered dietary differences relating to roles within society (Fuller, Fuller, Harris, & Hedges, 2006). Other studies of Anglo-Saxon populations include the analysis of the 4th-7th century AD cemetery of Wasperton on the River Avon (Montgomery, Evans, Chenery, & M€ uldner, 2009), and the middle Anglo-Saxon burial ground of Belle Vue House, York (M€ uldner & Richards, 2007). Mays and Beavan (2012) collated carbon and nitrogen isotope data for 76 individuals from 18 early Anglo-Saxon cemeteries from England to investigate whether access to dietary resources was determined by the geographical location of the populations (i.e., inland, coastal, riverine), or the age or sex of the individuals. Pictish/Medieval (6th-15th century AD) populations from Portmahomack, Scotland have also recently been analysed (Curtis-Summers, Montgomery, & Carver, 2014). All of the aforementioned studies relied on the application of carbon and nitrogen isotope analysis only. Sulphur isotope analysis is, however, increasingly being applied to such studies and can offer further insight into palaeodietary reconstruction and, potentially, to mobility studies (Richards, Fuller, & Hedges, 2001;Nehlich et al., 2011, p. 4964).
In contrast to the palaeodietary analysis of populations from Anglo-Saxon England, only one investigation of diet amongst early medieval populations from Wales has been undertaken to date, primarily because of poor skeletal preservation amongst populations from western Britain. Five adult individuals excavated from the 10th-century monastic site of Ty Newydd, on Bardsey Island in northwest Wales were subject to carbon and nitrogen isotope analysis, and the results revealed a diet reliant on terrestrial resources (Arnold, 1998). Given the dearth of comparative studies for western Britain, there is a considerable lacuna in our understanding of dietary practices amongst early medieval populations from Wales and the Isle of Man that could be addressed through stable isotope analysis. There is all the more need for a palaeodietary isotope investigation of these populations since archaeological evidence for subsistence practices is also limited due to a scant amount of settlement evidence from this region (Edwards, Lane, & Redknap, 2011). This study will therefore investigate the diets and resource consumption of populations from Wales and the Isle of Man through the application of carbon, nitrogen and sulphur isotope analysis to nine cemetery populations from three specific geographical regions; the Isle of Man, southwest Wales (Pembrokeshire), and southeast Wales. Consideration will be given to whether populations in these different regions relied on the same dietary resources, or whether location may have influenced access to, and consumption of, different foodstuffs and the reasons for this. The isotope data will be interpreted in the context of evidence from the historical and archaeological record to better understand such an essential part of everyday life of those communities living in early medieval western Britain.

| Diet and subsistence in early medieval western Britain
The British cleric, Gildas, writing in the mid-6th century AD provides a contemporaneous account of life in early medieval Britain in his De Excidio Britanniae (Winterbottom, 1978). According to Gildas, the island of Britain occupied "the end of the world" and had wide plains and hills "excellent for vigorous agriculture" (Winterbottom, 1978, p. 17). Mountains were used for "alternate" pasturage implying that transhumancewhereby farmers and livestock would spend their time in the hills on a summer farm, and would return in winter to a lowland farm-was practiced in the 6th century (Davies, 1982, p. 40;Roberts, 1959). We also hear from Gildas of constant flowing fountains, "brilliant rivers" and lakes flowing "with a cold rush of living water" (Winterbottom, 1978, p. 17). The people of early medieval western Britain were therefore supported by a rich and diverse landscape, which allowed them to grow crops and support livestock including sheep, pigs, goats, and cattle. The number of references to cattle in written sources suggests that cattle were central to the agrarian economy and were more important than other animals, and we hear of the severe consequences of cattle disease and cattle raiding (Davies, 1982, p. 39). Indeed, exchanges of land and compensation were valued, and payments made, in terms of cattle (Davies, 1978, p. 53). The importance of cattle is further reflected by the number of references to milk and dairy produce; the late-10th century Latin learning aid De Raris Fabulis, written either in Wales or Cornwall, lists a range of vernacular terms for milk products suggesting a variety of diary produce and widespread use (Charles-Edwards, 2013, p. 647;Davies, 1982, p. 35). Writing in the late-12th century, the cleric Giraldus Cambrensis in his "Journey through Wales" notes that the people of Wales enjoyed a plentiful supply of meat and poultry, but consumed only a single meal a day that was served on a large, rolledout piece of thin baked bread akin to a modern-day pizza (Davies, medieval corn dryers were found alongside evidence for the use of dredge-a mixed crop of oat and barley often used as fodder or for bread, oatcakes, or brewing-and the practice of malting (Carruthers, 2010), while quernstones-used for grinding corn-were recovered from the high-status settlement of Dinas Powys in southeast Wales (Alcock, 1963;Davies, 1982, p. 35).
In addition to the cultivation of crops and the rearing of livestock, communities also relied on natural resources to supplement their diet.
According to the 11th-century Life of St David, the people of St Davids, Pembrokeshire were accustomed to collecting nuts, and at the site of Brownslade Barrow environmental samples indicate that hazelnuts and sloes were collected from hedgerows (Carruthers, 2012, p. 161). Fish, particularly riverine fish, were noted as a source of food; in southeast Wales, Saint Cadog, the abbot of the 5th/6th century monastery of Llancarfan, fought to secure the rights for his monastic communities to fish in the Usk and Neath rivers (Davies, 1982, p. 34). The late-6th to late-11th century Llandaff charters (Liber Landavensis) which record grants to the Church in and around southeast Wales, show that access to weirs and fishing rights could also be donated in addition to land (Davies, 1978). Giraldus Cambrensis wrote specifically about the availability of fish, referring to perch and eels in the northern lakes of Snowdonia and salmon, trout and greyling in the rivers of the south (Davies, 1982, p. 34). Bees were another utilised resource and an abundance of honey was seen as an indicator of fertile land; the 12th-century Life of Saint Illtud praises the monastery at Llantwit Major for having an "abundance of flowers and honey" (Davies, 1982, p. 34). Law tracts record the value of swarms, hives, and wax suggesting that apiculturerather than simply the collection of wild honey-was particularly important, most likely since honey was essential for making mead and Welsh ale, regarded as the drink of elites, poets, and heroes (Williams, 1980). Access to agricultural land was not equal however, and those occupying the lower strata of society were expected to work the land for others. Evidence from the Llandaff charters demonstrates that much of the cultivatable land in southeast Wales was organised into estates, ranging from 40 to 6,000 acres, which was worked by tenant farmers and slaves (Davies, 1982, p. 42). The charters, and the Welsh lawbooks, such as the Book of Cyfnerth compiled in the 12th century, record that food renders were paid to landowners and territorial rulers (Charles-Edwards, 2013, pp. 274, 280;Davies, 1982, p. 41). Food renders were given in the summer and winter, and consisted of three essential components; ale, bread and an accompaniment to the bread depending on the season, with a greater quantity of meat in winter and more dairy produce in summer. Honey was also given either to produce mead or as an accompaniment to the bread, while the winter render also had to include horse-fodder (Charles-Edwards, 2013, pp. 280-282). Some elite rulers during the 5th to 7th centuries AD also had the capacity to import luxury consumables (Campbell, 2007;Fulford, 1989;Thomas, 1959). For example, eastern Mediterranean amphorae produced in the Peloponnese, western Turkey, and Syria have been recovered from high-status settlement sites in Wales, including Dinas Powys, and in Cornwall (Campbell, 2007). These vessels are believed to have contained wine, olive oil, and madder, while evidence for repaired amphorae suggests that some vessels could have carried dried goods (e.g., grain, nuts, spices) and not just liquids (Alcock, Stevenson, & Musson, 1995, p. 84).
Members of the monastic communities-particularly those resident at large estates in southeast Wales-also enjoyed a varied diet; for instance in his "Preface on Penance," Gildas refers to the monastic diet consisting of a small portion of butter, cheese, milk, and buttermilk to drink alongside bread, broth, eggs, and garden vegetables, while the 5th/6th century monastic community at Llancarfan enjoyed fish and milk (Davies, 1982, p. 35). It is, therefore, apparent from the written sources that those who were in a position to demand the labor and produce of others could enjoy a rich and varied diet drawn from an agrarian economy, while some members of high-status society had access to "exotic" foodstuffs. One particular advantage of investigating dietary resource consumption from the analysis of human remains excavated from early medieval cemetery populations is that the evidence will reflect the dietary resources consumed by all members of society, and will complement the elite perspective provided by written accounts. The study sites from southwest Wales and the Isle of Man are located within 20 km of the Irish Sea, while in southeast Wales, Atlantic Trading Estate is situated near the mouth of the Cadoxton River which flows into the Bristol Channel, and the nearby site of Llandough overlooks the River Ely and is less than 20 km from the Severn Estuary. A summary of each site is provided below, and sampled skeletons date from the period between the early-5th and early-11th century AD.  (Freke, 2002). Radiocarbon dating confirmed an early medieval date for the cemetery, with the earliest burials dating to AD 650-960 (2r), and the latest dating to AD 1290-1440 (2r). A total of 23 burials were identified as pre-10th century or probable pre-10th century date on the basis of their phase within the cemetery, available radiocarbon dates, datable coins, or their overall characteristics (e.g., stone-lined cists) (Freke, 2002). Skeletal preservation was highly variable across the cemetery and many earlier burials were poorly preserved; Table 1 includes the osteological data from the original assessment (Rubin, 2002). Stable isotope analysis was undertaken on 10 skeletons believed to be of pre-10th century date (Table 1).

| Balladoole
The cemetery of Balladoole occupies a small hillock ("Chapel Hill") on the southeast coast of the Isle of Man (Figure 1). The site was excavated in 1945 by Professor Gerhard Bersu and is most well-known for the discovery of a boat burial believed to date to the Viking Age (Bersu & Wilson, 1966;Wilson, 2008). Below, and thus predating the boat burial, was a cemetery of stone-lined cist burials, which have been confirmed as dating to the 4th-7th century AD through radiocarbon dating (Fox, pers.comm). According to the original excavation report by Bersu, burials were oriented east-west (head to the west end), without grave goods, and individuals were buried supine with their arms by their sides and hands placed on the pelvis (Bersu & Wilson, 1966). A reassessment of the skeletal remains was undertaken by KH, which identified both adults and non-adults (<18 years of age) (Bunting & Verity, 1960;Hemer, 2010). In total, 28 individuals were identified, 18 of whom could be associated with the in situ burials recorded on Bersu's excavation plan, while a minimum of 10 individuals were identified as mixed/ comingled. Four individuals were analysed for carbon and nitrogen isotope analysis and two were sampled for sulphur analysis (Table 1).

| Cronk keeillane
The cemetery of Cronk keeillane ("Mound of the little church") lies enroute to the town of Peel on the west side of the Isle of Man ( Figure 1) (Kermode, 1926). Very little is known about the cemetery itself asides from that recorded in antiquarian accounts from the late-19th and early-20th centuries (Barnwell, 1868;Kermode, 1926;Oswald, 1860).
The antiquarian reports refer to a cemetery east of a small barrow, where graves of various sizes were arranged in a parallel fashion (Oswald, 1860). In 1925, excavation revealed the remains of a keeill, and a cross-inscribed stone dated by P.M.C. Kermode to the 6th century AD (Kermode, 1926). During the 1980s, the cemetery was excavated and human skeletal remains were curated by Manx National Heritage, Douglas, however, the actual excavation report was never published. Recent radiocarbon dating has confirmed an early medieval date for five skeletons; four date to AD 564-654 (2r) while another dates to AD 656-770 (2r) (Hemer, 2010(Hemer, , 2012Hemer, Evans, Chenery, & Lamb, 2014). An assessment of the human remains by KH identified 16 adults and two non-adults; seven adult individuals were selected for isotopic analysis (Table 1).
FIG URE 1 Map of Wales and the Isle of Man illustrating the location of the cemetery sites analysed by this study, and the location of Ty Newydd, Bardsey Island (Arnold, 1998)    total of 814 articulated skeletons and 212 disarticulated bone groups were excavated and represent the remains of males, females, and nonadults (Loe, 2003). Dietary stable isotope analysis was undertaken on 36 individuals from the cemetery (Table 1).

| Atlantic Trading Estate (southeast Wales)
The site of Atlantic Trading Estate once consisted of a low promontory with the Bristol Channel to the south, and a sheltered bay and Barry Island to the northwest (Figure 1). The present course of the Cadoxton River also runs through the site which consists of sand dunes and marshland (Price, 1996). It is thought that the promontory was linked to Barry Island via a causeway, which was part of a pilgrim route to the chapel of St Barruc (Price, 1996). Excavation at Atlantic Trading Estate during the 1980s revealed an early medieval cemetery consisting of east-west (head to the west end) oriented burials cut into the sand dune. A total of 45 graves were excavated, all of which were in keeping with the funerary rites of this period, including simple sand-dug graves, the use of shrouds, stone-lined cist burials and occasionally, the use of wooden planks (Price, 1996). Osteological analysis of the skeletal remains by Loe (2003) revealed a mixed cemetery population consisting of males, females, and non-adults. Radiocarbon dates confirm that the cemetery was in use for burial between the late-4th and early-6th centuries AD (Price, 1996). A total of 10 skeletons, including nine adults and one non-adult were sampled for stable isotope analysis (Table 1).

| Brownslade Barrow (southwest Wales)
Brownslade Barrow is a Scheduled Ancient Monument situated in the Castlemartin Parish on the south Pembrokeshire coast and lies approximately 1.6 km from the sea in an area of wind-blown sand ( Figure 1) (Groom et al., 2012). Considerable damage to the barrow by badger activity was observed in 2001, and a number of burials were disturbed (Groom et al., 2012, pp. 136-137). In response, the site was surveyed well-preserved skeletons. Radiocarbon dates confirm that the cemetery was in use between AD 430 to AD 1020 (2r), with a considerable phase of burial activity between the mid-7th to late-9th century AD (Groom et al., 2012). The east-west (head to the west end) orientation of the burials, the form of the burials including simple earth-dug graves and stone-lined cist burials, and the absence of grave provisions, were consistent with early Christian tradition. Analysis of the skeletal remains revealed a mixed community of adult males, females, and nonadults (Hemer, 2010). Carbon, nitrogen, and sulphur analysis was undertaken on 10 individuals excavated from the site (Table 1).

| Porthclew (southwest Wales)
The  (Schlee, 2009, pp. 1-3). Six trenches were excavated across the site, four of which revealed human remains, and simple earth-dug graves and stone-lined cist graves. Radiocarbon dates for five burials confirmed that the site was in use during the early medieval period, with the earliest burial dating to AD 430-610 (2r), and the latest dating to AD 680-900 (2r). Eight individuals, including five adults and three non-adults, were recorded during the osteological analysis by KH; five were sampled for stable isotope analysis (Table 1).

| West Angle Bay (southwest Wales)
The cemetery of West Angle Bay is located on a cliff overlooking the Milford Haven Estuary in Pembrokeshire (Schlee & Ludlow, 2012, pp. 167-177) (Figure 1). The cemetery consists of an egg-shaped earthwork enclosure defined by a stone bank, which is contained within a larger rectangular enclosure. The northern edge of the rectangular enclosure has been lost from the edge of the cliff due to coastal erosion, and it was this erosion that exposed burials and human remains in 1997 (Schlee & Ludlow, 2012, pp. 171-176

| St Patrick's Chapel (southwest Wales)
Excavation of the early medieval cemetery of St Patrick's Chapel, which overlooks Whitesands Bay in Pembrokeshire, begun in 2014 following the exposure of human remains during severe winter storms (Figure 1).
Work on the project is on-going, however more than 50 burials have been excavated so far and these predate the foundation of a later medieval stone chapel. Radiocarbon dates confirm that the cemetery was in use for burial from at least the 7th century AD. The cemetery population consists of males, females and non-adults, and there is a predominant use of stone-lined cists and evidence for the use of crossinscribed grave markers (Murphy, Shiner, Wilson, & Hemer, 2016). Stable isotope analysis was undertaken on 17 human skeletons and 13 faunal samples collected from the cemetery site (Tables 1 and 2).

| Principles of palaeodietary reconstruction
The principles of palaeodietary reconstruction and the use of carbon and nitrogen isotope analysis are well-established and discussed at length elsewhere (e.g., Ambrose, 1990;Lee-Thorp, 2008;M€ uldner & Richards, 2005;Pate, 1994;Sealy, 2001), and therefore, it is only necessary to provide a brief overview here. The use of carbon and nitrogen isotopes for palaeodietary reconstruction relies on the principle that carbon and nitrogen isotopes present in consumed foods (i.e., dietary proteins, carbohydrates, and lipids) are incorporated into body tissues (e.g., bone collagen, dentine, and hair). Since bone remodels during life, the isotope composition of bone collagen from the rib reflects the isotope profile of the foods consumed (predominantly the protein) during the last two to five years of life (Cox & Sealy, 1997;Sealy, Armstrong, & Schrire, 1995). Carbon isotope values recorded for bone collagen will reflect the d 13 C values of consumed plants and animals, and can be used to distinguish between C 3 and C 4 plants which follow two different photosynthetic pathways (O'Leary, 1988;Pate, 1994;Schwarz & Schoeninger, 1991 Nitrogen isotope analysis is applied in conjunction with carbon isotope analysis to further identify the consumption of animal protein, and to distinguish between terrestrial and marine resources. Nitrogen incorporated into body tissues is obtained from the breakdown of amino acids from the protein portion of the diet during biochemical processes (e.g., protein metabolism), and undergo a process of fractionation which occurs at each trophic level (DeNiro & Epstein, 1981).
While the amount of 15 N fractionated during metabolic processes can be influenced by various factors (e.g., physiological adaptations, environmental habitats, physiological stress), an increase of 3-4& in the d 15 N value of the consumer relative to its source of dietary protein are observed at each successive trophic level (e.g., between carnivores and herbivores) (Ambrose, 1991(Ambrose, , 2000Fuller et al., 2005;Hedges & Reynard, 2007;Schoeninger & DeNiro, 1984;Sponheimer et al., 2003).
Such a trophic shift can also be observed in breastfeeding infants who are consuming maternal milk (Fogel, Tuross, Johnson, & Miller, 1997;Fuller et al., 2006;Jenkins, Partridge, Stephenson, Farley, & Robbins, 2001). Nitrogen isotope analysis can also be used to identify the consumption of aquatic resources since the longer food chains within these environments increases the number of trophic levels resulting in significantly higher d 15 N values. The consumption of marine protein will also result in higher d 13 C values (Pate, 1994;Schwarz & Schoeninger, 1991) while freshwater fish consumption can appear to have an overall "terrestrial profile" since freshwater fish have d 13 C values comparable to terrestrial species (Schwarz & Schoeninger, 1991, p. 304).
A more recent addition to the investigation of palaeodiet is the use of sulphur isotopes, which are incorporated into bone collagen from the essential amino acid, methionine (Nehlich, 2015;Richards et al., 2001, pp. 1-17). Methionine is synthesised by plants from inorganic sulphur leached from the underlying geology into the soil, and is passed on to animals and fish as dietary methionine with minimal fractionation (<1&) at each trophic level (Nehlich, Boric, Stefanovic, & Richards, 2010, p. 1131Nehlich et al., 2011, pp. 4965-4966;Trust & Fry, 1992). As such, d 34 S of bone collagen in humans relates to the sulphur isotope ratios of the soil and plants available in a particular region (Nehlich, 2015, pp. 3-5;Nehlich et al., 2011, p. 4967), and knowledge of the sulphur isotope composition of the local geology and water sources can also be used to identify non-local individuals (Nehlich et al., 2010(Nehlich et al., , p. 1132. Sulphur isotope analysis is used to distinguish between consumers of marine and terrestrial resources since terrestrial organisms are said to have low, but more variable, sulphur isotope values ranging from 25& to 110& (Krouse, 1980), while marine organisms cluster around a d 34 S value of c. 120&, which is the value of marine sulphate (Richards et al., 2001, p. 186

| Laboratory methods
Carbon and nitrogen isotope analysis was undertaken on nine cemetery populations dating to the early medieval period from Wales and the Isle of Man (Figure 1). In total, 102 individuals-including adult males, females, and non-adults were sampled (Table 1). While preference was given to the analysis of collagen from rib bone-since this limited any damage to other, intact skeletal elements-in some instances, bone preservation was so poor that it was necessary to extract collagen from the primary dentine portion of a tooth.
Collagen was extracted from fragments of rib bone or tooth dentine according to the method of Longin (1971). For the collection of dentine, each tooth was cut in half using a diamond-edged rotary dental saw, and any secondary dentine was removed using a tungsten carbide dental burr. Remaining primary dentine was separated from the enamel surface and collected for analysis; the remaining enamel was

| R E S U L T S
In total, 115 samples of collagen were analysed for carbon and nitrogen isotope analysis (Tables 1 and 2; Figure 2), and of these, 99 samples were considered to have C:N ratios that correspond to suitably preserved collagen (C:N 2.9-3-6) according to DeNiro (1985) (Tables 1 and   2). Many of the lowest C:N ratios (2.8/2.9) were recorded for samples of dentinal collagen, which was only used when bone was unavailable or poorly preserved. The low C:N ratios for many of the dentinal  Since dentinal collagen forms during childhood and does not remodel, it provides a dietary signature for the individual at the time of tooth formation. As most samples of primary dentine were retrieved from the second permanent molars, the carbon and nitrogen isotope values will indicate the individual's childhood diet between three and seven years of age (Beaumont, Gledhill, Lee-Thorp, & Montgomery, 2013;Hillson, 1996). The analysis of bone and dentinal collagen provides the opportunity to compare the diets of adult individuals with the diets of non-adults, that is, individuals aged biologically below the age of 18. Mean carbon and nitrogen isotope values reflecting the dietary profiles of adults and non-adults are presented in Table 6. Comparison between the mean carbon and nitrogen isotope values  Chapel. It seems plausible that this individual's elevated nitrogen isotope value-which is almost 13& higher than the mean nitrogen isotope value for the sampled population-indicates that this infant was breastfeeding, or had only recently begun weaning at the time of death and was thus at a higher trophic position relative to its mother.

| Comparison between sites and geographic regions
Since the study included cemetery populations from three different

| DISCUSSION
The carbon and nitrogen results obtained for the study sample are consistent with a terrestrial diet since the majority of the sample have carbon isotope values that are below 220 61&, which is believed to be the threshold for a 100% terrestrial diet (Richards et al., 2006, p. 123) ( Figure 2). As this is the first multi-site palaeodietary study of early medieval populations from Wales and the Isle of Man, comparison between this study and data from other inland and coastal sites spanning the medieval period has the potential to elucidate meaningful similarities or differences that could relate to dietary trends or agricultural practices across time and location. As such, the palaeodietary results obtained as part of this study were compared with published data from other medieval populations from Britain including the 10th century cemetery of Ty Newydd, Bardsey Island (Arnold, 1998); the 5th-7th century Anglo-Saxon cemetery of Berinsfield (Privat & O'Connell, 2001), the Iron Age/Norse site of Cnip, Orkney (Richards et al., 2001), 6th-11th century burials from Portmahomack (Curtis-Summers et al., 2014), the 14th century Bordesley Abbey (Richards et al., 2001) and the 7th-17th century site of Auldhame, Scotland (Lamb et al., 2012). freshwater resources (M€ uldner & Richards, 2005, p. 44), or the use of fertiliser since manure is known to enrich soil with 15 N (Bogaard, Heaton, Poulton, & Merbach, 2007). In the context of this study, the use of fertiliser warrants further consideration particularly in light of the sulphur isotope data.
There is a significant difference between the mean carbon isotope values of individuals from the east and west regions of our study area.
Comparison of the faunal isotope data from St Patrick's Chapel and data from Madgwick, Mulville, and Stevens (2012) also reflects this regional difference ( Figure 4). As the differences are evident in the human and faunal data, this suggests that the variability is due to environmental factors such as temperature, humidity, evaporation rate and rainfall, which all effect plant metabolism (Schmidt, Robins, & Werner, 2015).
The analysis has revealed a distinct difference between the sulphur isotope values for those populations from the Isle of Man and southwest Wales in comparison to those populations from southeast Wales ( Figure 5). The two populations from southeast Wales have sulphur isotope values within an "inland" population range (1-14&) (Richards et al., 2006 (Lamb et al., 2012, p. 766;Richards et al., 2001, p. 186 Figure 6).
The use of seaweed as a fertilizer is also known to introduce marine sulphates to the soil resulting in high sulphur isotope values.
Published strontium isotope data for individuals within the same study sample, including those from Pembrokeshire, reveals a number of individuals with high strontium concentrations (>100 ppm) (Hemer et al., 2013;Hemer et al., 2014) century (Amorosi et al., 1998;Balasse, Tresset, Dobney, & Ambrose, 2005;Hallsson, 1961). It is therefore necessary to consider the use of seaweed as a fertiliser by the populations from western Britain to explain the occurrence of high sulphur isotope values. There is also further evidence for this from the archaeological record. Analysis of soil samples from Brownslade Barrow revealed the presence of clover-type weeds, and it was suggested that the sandy soils around Brownslade Barrow were of low nutrient content, and therefore it was necessary to fertilize the land with household waste or seaweed ash (Carruthers, 2012). The use of seaweed ash as a fertilizer is further suggested by the recovery of charred seaweed remains belonging to the Fucus spe-cies from the site (Carruthers, 2012). It is therefore possible that populations in and around the southwest coast of Wales, and possibly the Isle of Man, utilised seaweed as a fertiliser for crops and may explain the occurrence of elevated sulphur isotope values observed amongst the populations from this region by comparison to those "inland" populations from the southeast. In addition, seaweed may have been consumed directly by the coastal inhabitants.  (Tables 1 and 4). Of all the sites within the study sample, Llandough is the furthest away from the sea, and the region may have been under less influence from atmospheric marine sulphates, while seaweed may not have been used as a fertiliser or consumed by those living in this region. Such regional differences in the sulphur isotope values evident here lends support to Richards et al.'s (2001) suggestion that sulphur isotopes may offer a palaeomobility signal.
While geographical location and proximity to the sea and its resources is likely to be the primary factor influencing the regional differences in sulphur isotopes, individuals buried at Llandough may also have been consuming produce that was grown further inland and was available to those communities living at, or nearby to, this ecclesiastical centre. Historical evidence demonstrates that religious foundations in southeast Wales were surrounded by land used to support members of the lay and religious communities; for example, the monastic community of Llancarfan grazed their sheep on the island of Flatholm (Davies, 1982, p. 164), while a charter entry from AD 620 notes that the monastery near Caerwent was supported by its ager suburbanus, that is, the fields outside the city (Davies, 1982, p. 164). While some religious communities were self-supporting, other ecclesiastical centres owned land that was farmed and managed by members of the lay community who took some of the produce to meet their own needs, and provided the remainder to the monastery (Davies, 1982, p. 165). Some of the more powerful religious houses, like the bishopric of Llandaff, acted as landlords and were capable of extorting surplus produce in the form of food renders-including livestock, bread and ale-from very distant churches and their communities (Davies, 1982, p. 165). There is also the connection between religious houses and nearby high-status settlements, and the relationship that this created in terms of the collection and redistribution of surplus produce. For instance, Llandough is connected to the high-status settlement of Dinas Powys, less than 3 km away (Holbrook & Thomas, 2005). Dinas Powys has yielded a considerable volume of archaeozoological material as well as evidence for specialist craft working and imported Mediterranean pottery (Alcock, 1963;Seaman, 2013). Such evidence led to the suggestion that Dinas Powys was an elite stronghold that received tribute from local tenants in the form of food renders, including mature, fattened animals (Alcock, 1963;Gilchrist, 1988, p. 59). As such, the elite at Dinas Powys enjoyed a position within society which allowed them to accrue local agricul-tural surplus which in turn was consumed, alongside imported Mediterranean foodstuffs, as part of frequent, large-scale feasting events which were central to the establishment and maintenance of allegiances between elites and their subjects (Davies, 1982;Seaman, 2013).
The presence of the same Mediterranean amphorae at Llandough demonstrates that secular elites residing at Dinas Powys also made donations of imported consumables-such as wine and oil-to their neighboring religious community, and would benefit from the obliged counter-gift from the monastic brethren (e.g., prayers for the soul of the patron and his family) (Grierson, 1959, p. 137;Knight, 2005, p. 104). The donation of luxury items to the monastic community at Llandough raises the possibility that other agricultural produce collected as food renders by the elite of Dinas Powys may have also been passed on to its associated religious community at Llandough. The low sulphur isotope values recorded for the sample from this cemetery may therefore reflect individuals who consumed foodstuffs drawn from the hinterland.
Since Llandough was an important religious foundation in southeast Wales during the early medieval period, it is also necessary to consider the possibility that the cemetery served communities who did not live in the immediate vicinity of the coast. Indeed, a previous investigation of population mobility in Wales identified individuals buried at Llandough who may have grown up in other parts of Wales and the borders with England suggesting some degree of movement to the Llandough region (Hemer et al., 2013(Hemer et al., , p. 2356. As such, the low mean sulphur isotope value for individuals buried at Llandough may reflect the fact that this monastery was well-connected to populations from further inland, and that the cemetery was a foci of burial for a farreaching Christian community. By contrast, the cemeteries from southwest Wales and the Isle of Man served only local, coastal communities who relied on the consumable resources they were able to grow in their immediate vicinity.

| C O NC LU S I O N S
This study aimed to shed light on the use and consumption of dietary resources by communities living in Wales and the Isle of Man during the early medieval period. A limited amount of settlement and faunal evidence meant that the opportunity to investigate diet through stable isotope analysis would provide an invaluable perspective on the communities from western Britain. As such, the first multi-site investigation was undertaken, using carbon, nitrogen, and sulphur isotope analysis.
Despite many of the sites being in close proximity to the coast, the results revealed an overall reliance on terrestrial protein. Clear differences were observed, however, between the populations from the west coast of Wales and the Isle of Man in comparison to the populations from southeast Wales. A clear distinction was identified in the sulphur isotope values between these populations, with those from the west having a distinctly "marine" sulphur signature, while those from the east had a more "inland" signature. It was proposed that this difference may relate to different subsistence strategies employed by populations from these regions, and consideration is given to the possibility that HEMER ET AL. | 437 those in the west were reliant on the use of seaweed. Moreover, consideration was given to the fact that the cemetery sites themselves served very different communities; Llandough was an important ecclesiastical centre which was well-connected to communities in and around the southeast including the elites occupying the high-status settlement of Dinas Powys. As such, there was access to resources, and people, from a much-wider geographical area. In contrast, the cemeteries from the west coast were the foci for local communities who did not have the same networks as the large ecclesiastical centres, and as such these local populations relied far more on produce grown in and around the vicinity of the coast.

ACKNOWLEDGMENTS
This study was possible thanks to the support provided by the Cardiff, and Manx National Heritage for allowing access and permission to sample the skeletal collections included in this study. We would also like to thank the two anonymous reviewers who provided comments and feedback on this article.