Diet, Society, and Economy in Late Medieval Spain: Stable Isotope Evidence From Muslims and Christians From Gandía, Valencia

This article investigates the diets of neighboring Christians and Muslims in late medieval Spain (here 13th–16th centuries) through the analysis of the stable isotopes of carbon (δ13C) and nitrogen (δ15N) in adult human and animal bone collagen. Twenty-four Christians and 20 Muslims are sampled from two adjacent and contemporaneous settlements in the township of Gandía on the Mediterranean coast, together with the remains of 24 animals. Statistical differences in both δ13C and δ15N reveal that the diets of the two faith communities differed, despite living side-by-side. These differences may relate to inequalities in their access to foodstuffs, particularly to C3/C4 grain and/or possibly terrestrial meat sources, though cultural preferences are also highlighted. Isotopic values for animals were also found to vary widely, both between and within species, and this provides a window into the local livestock economy. Am J Phys Anthropol 156:263–273, 2015. © 2014 The Authors. American Journal of physical Anthropology published by Wiley Periodicals,Inc.

Stable isotope analysis of archaeological bone collagen is a well-established technique that can provide direct evidence of the diet of archaeological populations down to the level of an individual (Ambrose, 1993). This information needs to be properly contextualized to maximize its benefits and is best used alongside other sources of evidence such as faunal assemblages and written texts. A growing number of projects have now applied the technique to medieval Mediterranean populations (e.g., Salamon et al., 2008;Bourbou et al., 2011;Fuller et al., 2012;Reitsema and Vercellotti, 2012;Ciaffi et al., 2013), but major published studies from Spain remain restricted to the Balearic Islands Nehlich et al., 2012). Responding to this challenge, this article presents data from late medieval populations from the mainland peninsula and examines the diets of Muslims and Christians living in a single locale in the Kingdom of Valencia during the 13th-16th centuries AD. It is the first such study to attempt this.
The intermingling of faiths in medieval Spain can be explored from numerous angles, through literature, architecture, the development of science and the study of archaeological objects, to name a few (e.g., Mann et al., 1992). The study of diet using stable isotope evidence is particularly noteworthy, however, because it offers the prospect of detecting directly the dietary preferences of different faiths living side-by-side within the same community. In brief, Christians, Muslims and Jews co-existed in the Iberian Peninsula for much of the Middle Ages, at first under Islamic political control in al-Andalus after the 8th century AD, and later under Christian rule after the 'reconquest' gained momentum during the 12th and 13th centuries. This life together, or convivencia as it is termed, lent itself to some surprising cultural influences, for example the popularity of couscous among Christians and their adopted habit of dining when seated on a cushion on the floor (de Castro, 1993: 172). However, it ended in the forced conversion and subsequent expulsion of Jews and Muslims from Spain after AD 1492 and 1609, respectively.
The realities of religious co-existence within this pluralistic society have long been the subject of academic debate Soifer, 2009) and food, in particular, might be expected to gain a significance that went beyond physiological need. While dietary preference may be constrained by the local availability of foodstuffs, it is often used to express cultural and social identities such as status, ethnicity, gender, occupation and regional difference (e.g., Montanari, 1999;Super, 2002;Parker Pearson, 2003;van der Veen, 2003). Religious groups, in particular, may identify themselves by what they choose to consume or to avoid (MacClancy, 1992:42;Latham and Gardella, 2005). With this in mind, the aim of the research described here was to use stable isotopes to characterize the diet of two faith communities living in the same locality. To achieve this, human samples were obtained from Gand ıa (Valencia), a large urban settlement with good archaeological evidence for contemporaneous Christian and Muslim burials. Animal remains were also obtained to act as an isotopic baseline and in order to investigate aspects of economy and animal husbandry.

HISTORICAL BACKGROUND
Gand ıa lies on the Mediterranean coast of Spain, around 65-km south of the modern city of Valencia (Fig. 1). Urban settlement first developed here in the 13th century after the Christian reconquest in 1249 and it is in this region of eastern Spain that the largest mud ejar population could be found during the later Middle Ages (Sesma, 2008); mud ejares being the remaining Muslim population living in territory controlled by the Christians, an ethnic minority characterized by its adherence to Islam. By the end of the 15th century these two faith communities had lived sideby-side for over 250 years, the Muslims being tolerated by the Christians, by now segregated and closed away from the religious perspective but, at the same time, making vital contributions to agriculture, traditional crafts and commerce (Guiral, 1989:435-504;Hinojosa, 2002). Gand ıa, by this date a royal duchy, was a successful commercial town with an economy rooted principally in the growing of sugarcane and the refining of sugar which was then exported to an international market (Guiral, 1989:424;Aparisi and Alonso, 2011). Archaeological excavation and historical research has shown that the town was surrounded by a cluster of small settlements in which the local population were mud ejares; one of these was at Benipeixcar, originally an aljama, or self-governing community of Muslims, which has since been absorbed into the modern town of Gand ıa.
In the early 16th century, the mud ejares were forced to convert to Christianity, and these "crypto-Muslims" were known as moriscos; many of them retained their Islamic faith in secret. This generated friction with the Christian community (Vincent and Ortiz, 1978;Ehlers, 2006) and the moriscos were finally obliged to depart in the early 17th century. In the meantime, they continued to perform an important role in the local economy, especially in the sugar industry as well as in the making of sweet confectionery (Meyerson, 1990:118, 134;Vincent, 1992:1). For brevity, the Muslim and later population of Benipeixcar of the 15th-16th centuries are referred to as mud ejares in this article, while the existence of various Spanish kingdoms has been simplified here by reference to "Spain."

DIET IN MEDIEVAL SPAIN
The study of diet in medieval Spain is based on a wide range of available historical evidence. Taken together, these imply a diet based on cereals, with regional differences in the types of grain being consumed, as well as variation according to social status. Bread was made from unrefined wheat while grains such as barley and rye were eaten by those at the lower end of the social hierarchy together with C 4 grains such as foxtail and broomcorn millet and sorghum in both red and white varieties (de Castro, 1993:286;Garc ıa Marsilla, 1993:77, 259;Garc ıa S anchez, 2002:279;Tom as, 2009:466). Islamic texts also mention a long list of other grains or pulses for making bread of different qualities such as rice, barley, chickpeas and lentils (Garc ıa S anchez, 1995). Some of these grains are also referred to in historical texts as animal fodder (Rubio, 1995;Cortonesi, 1999) and archaeobotanical evidence confirms the presence of C 4 millets at medieval sites further north in Catalonia, although always in relatively minor quantities when compared against other cereals (Alonso, 2005;Al os et al., 200622007;Cubero et al., 2008). Although these grains will have formed a part of the diet of some proportion of the medieval population, current evidence falls short when considering their actual contribution. Another C 4 grain, maize, can be ruled out as a potential source of C 4 protein in this study. Maize was introduced to Europe after 1492 but historical evidence suggests it was not commonly grown in the Valencia region until the 17th and 18th centuries (G omez, 1974;Sauer, 1976: 824;Crosby, 2003:179).
Other foodstuffs were supplied from the kitchen garden like onions, garlic and legumes such as beans, chickpeas, lentils and peas. To these should be added a great variety of fruit such as pomegranates and oranges, honey, eggs, and dairy products such as milk and cheese which supplied additional protein. Meat was widely consumed but the type, quality, and quantity depended on status, geography, and faith. Oxen, for example, were considered to be the food of the poor (Garc ıa S anchez, 1983:139;de Castro, 1993;Tom as, 2009:466) whereas meat from young and suckling animals was a high status food in demand on the urban market (D ıaz, 1983;Mart ınez, 1996;Cortonesi, 1999). The most common types of meat consumed were mutton and lamb followed by kid, chicken, pork (in the case of Christians), beef, and game such as rabbits (Waines, 1992;de Castro, 1993;Garc ıa S anchez, 1996). Fresh fish would have been an important supplement to the diet along the coast and rivers as well as providing a catch for sale (de Castro, 1993). As many as 27 types of fish are recorded in local port books, from eels to sardines and tuna, and including also octopus (Guiral, 1989:388) while salted, dried, preserved fish (escabeche) and smoked sea-fish were also widely consumed and traded far inland (Gallart et al., 2005); sardines, for example, were imported to the Valencia region from Galicia in northern Spain (Ferreira, 1988:732). Remains of cuttlefish (Sepia officinalis) and possibly crab or crayfish from nearby sites in Alicante confirm that cephalopods and crustaceans also probably formed part of the Islamic diet (Benito, 1987).
Fish fulfilled a liturgical requirement for Christians; fast days (typically Wednesdays and Fridays) and Lent together accounting for over 150 days in the Christian calendar (Tom as, 2009:465), although other foodstuffs could be substituted for meat, among them cheese, eggs, legumes, nuts or vegetables (Garc ıa Marsilla, 1993:76;Adamson, 2004:189;Rodrigo, 2009:571;Grumett and Muers 2010:26-27). There are nuances to this generic description however, the Church in Spain, for example, allowed the consumption of eggs and dairy products in Lent officially only after 1491 (Tom as, 2009:465) while monasteries and religious orders had their own fasting practices, some prohibiting meat entirely (Sesma, 1977:67;Grumett and Muers 2010:43).
Muslims were prohibited from consuming pork and any animal that had not been slaughtered in such a way as to be considered halal but, in terms of fasting restrictions, Islamic dietary law was actually less restrictive than the dietary regimes prescribed by Judaism and Christianity in the strictest sense (Insoll, 1999:95-97;Zaouali, 2007). In that context it is interesting that some remains of pig were recovered from Benipeixcar. Zooarchaeological evidence indicates that the presence of pigs is not unusual for Muslim sites of this period and the very few remains that are recovered are usually attributed to the presence of a small number of Christians on site or else to the hunting of wild boar (Morales et al., 1988;Lentacker and Ervynck, 1999;Antunes, 1996;Benito, 2006;Davis, 2006). This reminds us that, although some correlation between foodstuffs and faith is expected, much of the medieval diet will be influenced by what was affordable and available locally, thereby demonstrating cultural affinities and social status, as well as religious practices.
In summary, while there are still relatively few detailed published studies of well-dated faunal assemblages from later medieval sites in Spain, we nevertheless have a reasonable impression of the foodstuffs available. We can say far less, however, about what was consumed by whom and in what proportions and this complicates any fuller examination of the differences in diet between faith communities. Stable isotope analysis, on the other hand, offers exactly that opportunity because it can examine the diet of individuals, and the technique is also well suited to applications in regions where C 4 plants were cultivated, as they were in Gand ıa.

RECONSTRUCTING DIET USING STABLE ISOTOPE ANALYSIS
Carbon (d 13 C) and nitrogen (d 15 N) isotopes in bone collagen are most frequently used for dietary recon-struction in archaeological populations; several reviews of the application of stable isotopes in archaeology are available (e.g., Katzenberg, 2000;Sealy, 2001;Schwarcz and Schoeninger, 2011). The isotopic composition of body tissues such as bone collagen reflects that of mainly the protein portion of the diet averaged over many years prior to death, although for low protein diets fats and carbohydrates may contribute carbon to the collagen (Hedges, 2004; and different protein sources of animal origin such as meat and milk cannot be distinguished (O'Connell and Hedges, 1999). Of significance here is the fact that carbon isotopes can discriminate between plants that use different photosynthetic pathways; C 3 plants include common crops such as wheat, barley and oats, most fruits, legumes and vegetables grown in temperate climates, whereas C 4 plants are better adapted to hotter, drier environments, including tropical grasses and crops such as sorghum, millet, maize and sugarcane (O'Leary, 1981). C 4 plants exhibit more enriched carbon values than C 3 plants, with mean values d 13 C of 213& and 227& respectively (Smith and Epstein, 1971;O'Leary, 1981). One confounding factor is the presence of marine foods more enriched in 13 C than C 3 terrestrial based resources (Schoeninger and DeNiro, 1984). As a result, carbon can be used to distinguish marine protein consumption in terrestrial C 3 -based diets but, when C 4 plants are involved, marine and terrestrial carbon values can overlap (Sealy, 1997). Stable nitrogen isotope ratios, however, do help to distinguish between the consumption of purely marine protein or C 4 plants, because the d 15 N values of tissues provide information about the trophic level at which an organism is feeding; nitrogen is enriched by around 3-5& with each step along the food chain (Bocherens and Drucker, 2003;Hedges and Reynard, 2007). Carbon, on the other hand, exhibits less enrichment with values of 0-2& as trophic levels increase (Bocherens and Drucker, 2003). As aquatic foodchains tend to be longer than terrestrial ones, marine and freshwater resources usually exhibit higher d 15 N values than terrestrial sources due to greater enrichment in nitrogen (Schoeninger et al., 1983). As noted elsewhere, however, this may not be the case in the Mediterranean as the ranges of d 15 N values of marine and terrestrial animals from this region can overlap considerably . Terrestrial plants can also demonstrate elevated d 15 N values as a result of natural environmental conditions such as salinity and aridity or anthropogenic factors such as the spreading of manure as fertilizer (Bogaard et al., 2007;Fraser et al., 2011).
There is some uncertainty in the reconstruction of the protein part of the diet of an individual using collagen d 13 C and d 15 N. Feeding studies indicate that there may have been underestimation in the magnitude of the diettissue offset in nitrogen with each trophic step (D 15 N dietcollagen ), which may be up to 16& (O'Connell et al., 2012). Furthermore, due to macronutrient routing, there may be an unequal contribution of carbon from different dietary sources, which could lead to inaccuracies when estimating the amount of particular foods such as marine protein in archaeological populations . While precise reconstruction of an individual's diet is more problematic, carbon and nitrogen stable isotope analyses of bone collagen remains a powerful and informative tool for investigating relative differences between the protein parts of the diets of past populations and individuals.

SAMPLES AND METHODOLOGY
Samples of adult human bone were taken from the burials of 20 mud ejares in an Islamic cemetery dating to the 15th to 16th centuries associated with the settlement of Benipeixcar (Fig. 1), excavated by J. Cardona in 1993Cardona in -1994. All were buried SW-NE with the head facing east towards Mecca, in accordance with Islamic tradition. For comparison, 24 Christian skeletons were also sampled from three grave pits uncovered during excavations at the Colegiata de Santa Mar ıa in nearby Gand ıa prior to restoration of the building (Vidal, 2006). The archaeological contexts from which these skeletal remains came are only broadly dated to between the 13th and 16th centuries, so there is a slight discrepancy in the date ranges of the mud ejar and Christian samples which cannot be resolved. Most importantly for this study, they represent two faith groupings from the same geographical location. Together with this human bone, terrestrial animal bones (n 5 24) from a variety of species were also sampled from domestic contexts excavated within the settlement of Benipeixcar. The poor recovery of fish bones on site prompted the inclusion of additional marine fish from an 11th-12th century site at Albarrac ın in Teruel province (Fig. 1) in order to provide a contemporaneous isotopic baseline.

Collagen extraction and stable isotope analysis
Rib bones were sampled from each adult human skeleton and cortical bone was chosen from animal bones, ensuring that only one sample was taken from each individual. Collagen was isolated from bone samples following a modified Longin (1971) method including an additional ultrafiltration step (Brown et al., 1988;Richards and Hedges, 1999). Samples of bone (0.2 g) were cleaned by air-abrasion and demineralized (0.5M HCl at 4 C for up to 10 days), the resulting insoluble fraction was gelatinized (HCl at pH 3) for 48 h at 75 C and the supernatant ultrafiltered to isolate the high molecular weight >30 kDa fraction which was then lyophilized. The purified collagen was analyzed in duplicate using continuous-flow isotope ratio mass spectrometry (CFIRMS) using a Delta XP mass spectrometer at either the Alaska Stable Isotope facility (University of Alaska Fairbanks's Water and Environmental Research Centre; WERC) for humans or at the Department of Human Evolution, Max Planck Institute for Evolutionary Anthropology (MPI-EVA, Leipzig, Germany) for animals. Isotopic values are reported here as the ratio of the heavier isotope to the lighter isotope ( 13 C/ 12 C or 15 N/ 14 N) as d values in parts per mille (&) relative to internationally defined standards for carbon (VPDB: Vienna Pee Dee Belemnite) and nitrogen (AIR) following the equation [d5 (R sample -R standard )/R standard 3 1000] (Coplen, 1994). The instrument precision for d 13 C and d 15 N was 60.2& (1r) or better at both laboratories, determined by replicate analysis of internal laboratory standards. Five samples were compared between the laboratories yielding mean differences of 0.6& 6 0.2& (1r) for d 13 C and 0.0& 6 0.6 & (1r) for d 15 N. The small difference in d 13 C is insignificant at the magnitude of dietary interpretation discussed here. Sufficient collagen (>1%) was found to be preserved in all the samples and the collagen quality met published criteria (DeNiro, 1985;van Klinken, 1999). Statistical tests were performed using SPSS for Windows version 15. The non-parametric Mann-Whitney U and Kolmogrov-Smirnov tests for non-normally distributed data were employed.

RESULTS
The human and animal stable isotope data are provided in Tables 1 and 2, respectively and plotted together in Figure 2; summary statistics for both humans and animals are listed in Table 3. An immediately noticeable feature of these results is that many of the humans in both populations register enriched carbon values suggestive of a measurable dietary input of C 4 plants and/or marine protein. The most depleted d 13 C values, on the other hand, suggestive of a largely C 3 terrestrial-based diet, are exhibited by the Christians sampled from Gand ıa. Overall, despite the relatively small difference (<1&) in absolute means for d 15 N and d 13 C between the populations of Christians (n 5 24) and mud ejares (n 5 20) shown in Table 3, the distribution of their d 15 N and d 13 C values is statistically different (two- sample Kolmogorov-Smirnov z 5 1.486, P 5 0.024, and z 5 1.486, P 5 0.024, respectively). Comparison in Figure  3 indicates that mud ejares demonstrate more enriched d 13 C and d 15 N values than their Christian contemporaries and this implies that the long-term diets of each group were significantly different. Comparison of human values with herbivores recovered from Benipeixcar reveals that human values from both sites are on average 2& higher in d 13 C and 5.4& higher in d 15 N than the means for herbivores (Table 3). The dietary spacing between humans and herbivores is wider than might be expected for trophic level enrichment (Bocherens and Drucker, 2003). This implies that both populations also consumed other enriched sources of 15 N, perhaps chickens and their eggs, pork and marine fish, although those individuals with enrichment in both d 13 C and d 15 N values may have consumed marine fish in greater quantities. However, taking the potential dietary spacing for D 15 N diet-collagen being 6& (O'Connell et al., 2012), the consumption of herbivore protein cannot be ruled out as a major dietary source.
More precise determination of individual dietary preference is complicated by several factors. First, isotopic values are distinct between species such as ovicaprids and cattle, implying different management strategies. However, there is also evidence of multiple management and feeding regimes for the same species. Two cows, GBC1 and GBC4, possess enriched carbon values of 215.2& and 214.3& indicative of a significant dietary intake of C 4 plants when compared to other cattle which ate predominantly C 3 diets. Two chickens, GBGa1 and GBGa5, also appear to have had a significant C 4 plant contribution to their diet, with d 13 C values of 214.9& and 213.3&, respectively; C 4 crops such as millet and sorghum, which make excellent feed for poultry, are likely to be responsible. Overall, however, the wide range of isotope values exhibited by the species in the sample makes it very difficult to determine the proportional contribution of different foods to the human diet, that is, marine foodstuffs, C 4 -fed herbivores, and C 3 -fed herbivores. This is particularly noteworthy in this case study because terrestrial fauna exhibit d 15 N values that are similar to those of Mediterranean fish, which in turn tend to exhibit lower values than those elsewhere, in the Atlantic for example (Jennings et al., 1997;Garcia-Guix e et al., 2010;Barrett et al., 2011;Vika and Theodoropoulou, 2012;Craig et al., 2013).

Economic and social explanations for dietary difference
Despite the inevitable uncertainties in determining diet, and the broad range of dates available for the Christian samples selected for this study, the differences observed between Christians and mud ejares in both d 15 N and d 13 C reflect their contrasting access to foodstuffs over long periods of their lives. In particular, the enriched d 13 C values exhibited by mud ejares would imply that this Muslim or "crypto-Muslim" community relied more heavily on C 4 plants, in their own diet and/ or that of their animals.
This result is open to several possible interpretations. We have already noted that the mud ejar population of Gand ıa was closely involved with sugarcane cultivation and processing and it is probable that workers on the sugarcane plantations chewed on the raw canes to extract the sweet juice; indeed, inquisitorial records specifically indicate this was common practice among mud ejares in the 16th century (Galloway, 1989:16;Ehlers, 2006:96). This habit, however, cannot be the explanation for the elevated C 4 signatures; sugarcane has an extremely low protein content and it would remain under-represented in the bone collagen provided Marine fish from Albarrac ın are marked with an asterix (*). that the diet contained adequate quantities of high protein foodstuffs such as meat. It is far more likely then that the adults with enriched carbon values obtained their C 4 plant signatures indirectly through the consumption of animals foddered on bagasse, the fibrous waste from sugarcane crushing, and the green tops.
C 4 crops, it should be noted, were generally regarded as a low-status foodstuff by society at this time; sorghum and millet were consumed only when the supply of other grains was lacking, particularly in rural settlements (Braudel, 1972:595;Glick, 1982:82;Sarasa, 1995:195;Adamson, 2002:8), so the observed dietary differences may reflect socio-economic disparities between rural (Benipeixcar) and urban (Gand ıa) and/or between mud ejares and Christian populations. Historical sources indicate the Duke of Gand ıa instructed the Muslim workers on his land to farm sugarcane due to its profitability and that this policy led to such a serious shortage of staple crops, such as wheat, that grain had to be imported from Valencia (Aparisi et al., 2009). A conventional interpretation of the results might therefore be that the Christian inhabitants of Gand ıa had privileged access to these wheat imports while the rural mud ejar population fell back on the cultivation and consumption of C 4 crops. A similar emphasis on C 4 crops has been noted among mud ejar populations in the Sierra de Espad an in the north of the Kingdom of Valencia in the 14th-16th centuries (Butzer et al., 1986: 372). Written records and archaeobotanical evidence indicate that communities there relied on the cultivation of millets and sorghum, and this has been linked to the "social segregation" of Muslims and, in particular, to a deterioration in their social standing following the Christian Reconquest (Ruiz, 2001). We should also remember that dietary differences between crypto-Muslims and Christians may have become more acute during the 16th century when halal meat became unavailable, at least legally, due to the prohibition of Muslim butchery practices (Vincent and Ortiz, 1978;Ehlers, 2006). Alternatively, it is also possible that the isotopic variation in both d 13 C and d 15 N is explained by a stronger preference for marine fish in the mud ejar diet. There is some historical evidence to suggest that meat rose in price in the Kingdom of Valencia during the 16th century, when the king prohibited the slaughtering of animals several times after 1525 (Garc ıa C arcel, 1977:100) and this may have led to a preference for fish, chicken and eggs among those of lower socio-economic status.
A word of caution is required, however, before these socio-economic interpretations are uncritically accepted. According to Islamic sources, millet (a C 4 crop) was much appreciated by Muslims (Mar ın, 1997:10); it is mentioned, together with sorghum, in Islamic agricultural treatises (Hern andez Bermejo et al., 1998:22) and it was cultivated together with wheat and barley in regions of the peninsula that lay outside Christian political control, such as in 13th-century Granada (Ari e, 1992:158). Millet is also specifically listed in Islamic treatises as a licit product in which to trade (Constable, 1994:161) and this suggests that differences in cultural Standard deviations are shown when sample size is at least 3. traditions in cooking and eating may also have had their part to play in the observed isotopic differences.
The results presented here strongly suggest that the evident dietary differences between mud ejares and Christians originated in the home. A comparison of Muslim and Christian females reveals that the enrichment in 13 C in particular is most pronounced in the mud ejar population (Mann-Whitney U 5 21.5, P 5 0.001; Fig. 4), although the observation of corresponding higher d 15 N values among mud ejar females compared to Christian females is not statistically significant (Mann-Whitney U 5 48.5, P 5 0.067). If it can be assumed that the task of food preparation was, for the most part, taken up by women in medieval Christian and Muslim societies (Bynum, 1987:192;Garc ıa Marsilla, 1993:152;Boone and Benco, 1999:67;Insoll, 1999:62-63), then this may have heavily influenced the difference in diet. Unfortunately, there were too few males (n 5 5) in our mud ejar sample to explore this further and compare between males of different faith. The question might be resolved in the future by larger and better-balanced samples.
Turning finally to trends in diet within the two faith groups, statistical analysis indicates that there is no difference between the d 13 C and d 15 N values of males (n 5 12) and females (n 5 12) in the Christian population (Mann-Whitney U 5 61.5, P 5 0.544 for d 13 C and U 5 62.5, P 5 0.583 for d 15 N). This is not the same as saying that Christian households always ate together, only that their overall diet did not differ greatly or, at least, any differences are not observable isotopically with these sample sizes. Likewise, there is no written evidence to suggest that female and male mud ejares had a different diet; though special arrangements were made for women during pregnancy or lactation (Aubaile-Sallenave, 1997;Garc ıa S anchez, 2006). Again, the insufficient sample number prevents meaningful comparison between sexes among mud ejares.

Implications for animal husbandry strategies at Gand ıa
This case study has also demonstrated significant differences in nitrogen values between ovicaprids and cattle. This may be due, on the one hand, to the stocking of cattle closer to settlements, where soils are likely enriched with manure and domestic refuse (Hedges et al., 2005;Bogaard et al., 2007;Commisso and Nelson, 2010). The sheep, on the other hand, may have been grazed further afield in transhumant flocks (Montalvo, 1992:162) as they were in the mountains of the Sierra de Espad an, north of Valencia, where there were both small, locally owned herds and large inter-regional flocks that grazed seasonally (Butzer et al., 1986).
The enriched carbon values exhibited by certain cattle probably reflect foddering on C 4 plants such as millet, sorghum or sugarcane. In his 12th century Treatise on Agriculture, Ibn al'Awwam mentions that the byproducts of sugarcane, among them molasses and discarded cane, supply excellent cattle fodder (Ibn al-Awwam, 1802) and this is exactly the practice in western India today where tops are fed green to dairy animals, as well as being dried and stored and fed like cereal straws (Mena, 1987;Yadar and Solomon, 2006). Given what we know of the local historic landscape, it seems highly likely that at least some of the C 4 contribution to the diet of cattle came from sugarcane, specifically the draught animals put to work in the plantations and sugar mills. Other animals too had access to these C 4 crops and while some animals, such as chickens and pigs, were foddered on millet, sorghum and/or sugarcane by-products; the cats possibly hunted rodents that would have scavenged C 4 resources.

CONCLUSION
Overall, the isotopic results presented here point to a complex, mixed C 3 /C 4 -based economy. The isotopic signatures of the animals sampled from Benipeixcar show a surprising degree of separation between species and within species, and our interpretation is that certain animals are likely to have grazed particular pastures and/or were deliberately fed significant quantities of C 4 plants, a feeding regime which may be linked to their use as traction animals in sugarcane agriculture. Notably, the use of C 4 crops in the study area was not restricted to cattle; chickens, cats and the single pig sample all exhibited enriched d 13 C values that probably reflect the availability of C 4 foodstuffs either through deliberate foddering or scavenging. While the use of C 4 plants to fodder animals has previously been hypothesized for this period, as mentioned above, the data here represent some of the first clear isotopic evidence for this practice in medieval Europe.
We have also shown that both the Christian population from Gand ıa and the mud ejares from Benipeixcar consumed significant amounts of C 4 plants directly and/ or indirectly, with some individuals exhibiting an enrichment in both carbon and nitrogen which suggests an input of marine protein into their diets. In addition, there are significant differences in both carbon and nitrogen values between the two faith communities, indicating dietary differences which may be linked to the differing cultural and/or socio-economic status of the two populations. After weighing up the available historical and archaeological evidence, we have suggested that mud ejares on the whole may have eaten more fish from the Mediterranean and relied more heavily on C 4 resources, stemming from their involvement in the cultivation and processing of sugarcane, a C 4 crop, or from a cultural preference for particular grain types. Alternatively, their particular dietary signature may be the result of limited access to more desirable C 3 plants, perhaps due to their declining socio-economic status at this time. Further clarification on this point could be sought through the application of single compound amino acid analysis (Corr et al., 2005;Choy et al., 2010) or in cases of excellent preservation, analysis of bone mineral apatite (Fernandes et al., 2012).
For the moment, while we cannot be certain about the food sources that underlie their dietary differences, we can demonstrate that the two contemporary faith communities ate different diets and begin to identify where those differences lay. Given the association of Christianity with fish consumption, it might be surprising that the Christians do not appear to have consumed more fish than the mud ejares, but perhaps this illustrates how differences in faith and culture can be subsumed by discrepancies in the regional availability of resources.
Our study illustrates well the challenge of teasing out medieval identities from archaeological data (D ıaz-Andreu et al., 2005;Hinton, 2009). Dietary signals in isotopes intermesh food habits which may result from religious persuasion, gender, regional and community traditions, the availability of produce, kinship, economic grouping, and social ambition (M€ uldner, 2009); choosing between these and understanding how and why social groups selected some foodstuffs above others requires robust historical and theoretical underpinning, for example to understand the role of food in the social construction of memory (e.g., Jones, 1997;papers in Counihan and van Esterik, 1997;Sutton, 2001;Mintz and Du Bois, 2002). In that sense, this project is only a beginning. As yet we know little about how medieval diet differed between Spanish regions, the effects of social and economic change on eating patterns, the impact of racial laws and prohibitions, gender dynamics, food allocation, taboos, food symbolism, culinary memory, and much else besides. For 16th century Spain we should consider how many mud ejares there were in the region, the nature of the community and their contacts with Christians; not all populations were as embedded as the rural Valencian case study discussed here. There is a long chronology to reflect upon too; Valencia fell to the Reconquest in 1238 but Granada only in 1492, so mud ejares of the 16th century across the peninsula had very different experiences of Christian colonisation and conversion which would be interesting to compare. The "slow shipwreck" of Islam on the Iberian Peninsula (Braudel 1976, 781) is potentially one with an intriguing archaeological signature but there is still a considerable way to go before the full contribution of stable isotope techniques is fully realised.