Beyond faith: Biomolecular evidence for changing urban economies in multi-faith medieval Portugal

Objectives: During the Middle Ages, Portugal witnessed unprecedented socioeconomic and religious changes under transitioning religious political rule. The implications of changing ruling powers for urban food systems and individual diets in medieval Portugal is poorly understood. This study aimed to elucidate the dietary impact of the Islamic and Christian conquests. Materials and Methods: Radiocarbon dating, peptide mass fingerprinting (ZooMS) and stable isotope analysis ( δ 13 C, δ 15 N) of animal (n = 59) and human skeletal remains (n = 205) from Muslim and Christian burials were used to characterize the diet of a large historical sample from Portugal. A Bayesian stable isotope mixing model (BSIMM) was used to estimate the contribution of marine protein to human diet. Results: Early medieval (8 – 12th century), preconquest urban Muslim populations had mean (±1 SD ) values of (cid:1) 18.8 ± 0.4 ‰ for δ 13 C 10.4 ± 1 ‰ for δ 15 N, indicating a predominantly terrestrial diet, while late medieval (12 – 14th century) postconquest Muslim and Christian populations showed a greater reliance on marine resources with mean (±1 SD ) values of (cid:1) 17.9 ± 1.3 ‰ for δ 13 C and 11.1 ± 1.1 ‰ for δ 15 N. BSIMM estimation supported a significant increase in the contribution of marine resources to human diet. Discussion: The results provide the first biomolecular evidence for a dietary revolu-tion that is not evidenced in contemporaneous historical accounts. We find that society transitioned from a largely agro-pastoral economy under Islamic rule to one characterized by a new focus on marine resources under later Christian rule. This economic change led to the naissance of the marine economy that went on to characterize the early-modern period in Portugal and its global expansion.

estimation supported a significant increase in the contribution of marine resources to human diet.

Discussion:
The results provide the first biomolecular evidence for a dietary revolution that is not evidenced in contemporaneous historical accounts. We find that society transitioned from a largely agro-pastoral economy under Islamic rule to one characterized by a new focus on marine resources under later Christian rule. This economic change led to the naissance of the marine economy that went on to characterize the early-modern period in Portugal and its global expansion. and Christians co-existed, characterized by diversified political settings, economic strategies, religious beliefs and population movements (Collins, 1994;Santos et al., 2014).
Between the 8th and 14th centuries, the Iberian Peninsula witnessed unprecedented structural, societal and religious changes following the Islamic conquest in 711 CE and the Christian conquest from the mid-12th century (Kennedy, 1996). Muslims ruled over much of the peninsula, creating a division between the northern Iberian Christian Kingdoms and the southern Islamic caliphate. During the early medieval period, historical sources indicate an economy based on agriculture and pastoralism throughout the Peninsula, while marine resources were exploited on a small scale, characterizing the subsistence strategies of coastal settlements (García-S anchez, 1992).
Archaeology and historical sources often reveal a gradual and continuous passage from one society to the other between the Muslim and the Christian period and although archeological evidence for destruction and abandonment levels have been identified; cultural practices continued in a more fluid way, sometimes facilitated by influential Mozarab (Christians living under Islamic rule) elites (Simões, 2012).
The increasing number of isotopic studies on medieval populations, however, shows a composite picture of food practices with high dietary variability at the local level L opez-Costas & Müldner, 2018;Lubritto et al., 2017;MacKinnon et al., 2019). In the later medieval period, Christian Kingdoms of the north gradually pushed south and conquered Islamic territory (Kennedy, 1996) and by the mid-13th century, the Kingdom of Portugal lay under Christian rule. The expanding reach of the Kingdom of Portugal saw the creation of new trading routes to Northern Europe, the Americas and Africa, with an emphasis on seafaring activities (Amorim, 2009).
While much previous historical research has focused on the political and religious transitions in Iberian medieval society, research on their impact on individual diets has been limited. Aquatic resources became economically important food items during the late medieval and early modern periods in Portugal, however their relative contribution to diet is poorly understood and often subject to sweeping generalizations (Amorim, 2009). In the light of political and societal turmoil that characterized the medieval period, the extent to which different food sources contributed to the diet of different groups, before and after the Christian conquest, also remain uncertain. It has been assumed that Northern European populations exploited marine and freshwater resources to sustain growing populations during this time period, although it is a matter of debate if this could be applied to southern Europe (Barrett et al., 2011). Remains of fish are scarcely recorded from archeological sites in Portugal, which may in part be due to recovery techniques and are therefore not sufficient to explore this issue alone. Stable carbon and nitrogen isotope analysis of bone collagen is a well-established technique to extrapolate dietary information from human, faunal and plant remains and has been employed to reconstruct medieval diets across the Iberian Peninsula (L opez-Costas & Alexander, 2019). However, the dietary practices of Islamic populations before and after the Christian conquest has rarely been explored (Alexander et al., 2015;Toso et al., 2019).
Here we report a novel contribution to disentangle diets in medieval Portugal, and specifically the importance of marine resources as a supply of dietary protein. New radiocarbon dates and a large body of bulk collagen stable isotopic (carbon and nitrogen) data from human and animal populations from three major urban sites: Lisbon, Beja and Silves have been produced. The relative contribution of terrestrial and marine resources to diet have been estimated using a Bayesian stable isotope mixing model (BSIMM, FRUITS) to provide a probabilistic model for the proportion of calories and protein derived particularly from marine resources.

| Archeological background and materials
Lisbon, Beja and Silves were three major urban settlements in medieval Portugal that played central roles in the economy during both Islamic and Christian rule (8-15th centuries) ( Figure 1). All three cities were under Islamic control from 711 CE but were conquered by the Christians at different times between the mid-12th and mid-13th centuries (Marques, 1993). They are all situated on rivers, although Lisbon has closest access to the sea. At all three locations, pre-and post-Christian conquest burials of both Muslim and Christian individuals have been analyzed, serving a threefold purpose of elucidating food access and distribution in urban environments in relation to faith, economy and chronology. Potential sex related differences in diet were also considered.
Funerary practices in Medieval Iberia were varied. A number of different juridic texts highlight the variability of laws and their interpretation that regulated burial practices throughout the Islamic rule of the Iberian Peninsula (Fierro Bello, 2000). Despite regional variations, however, there is a broad consensus on the treatment of the dead during this period that includes the preparation of the corpse, the funeral and the preparation of the grave (Petersen, 2013). Muslim burials were usually placed in single graves oriented toward Mecca (SE orientation in Portugal) with the body lying on the right-side facing east and cemeteries themselves were commonly outside the city walls near one of the main gate of the city (Insoll, 1999). Early medieval Christian burial customs from other parts of the Iberian Peninsula were initially similar in placing cemeteries in the outskirts of the cities, however, after the Christian conquest, Christian parishes and their churches developed which provided burial space in their vicinity (Ruiz-Taboada, 2015). The typology of Christian graves varied in burial depth and could present the use of bricks, which were forbidden by Islamic law (Fierro Bello, 2000). Each tomb would include the remains of a single individual, as in the Islamic tradition, however the body was placed in a supine position with the head looking up. All tombs usually followed an east-west orientation. Due to these different practices and additional contextual information from historical and archeological sources, cemeteries are usually assigned to one of the three main religions that were practiced in Iberia at the time that is, Christianity, Islam, and Judaism, even when multiple cemeteries are found in relatively close proximity as in the case of Toledo (Ruiz-Taboada, 2015). Lisbon and Beja also provided rare evidence of multifaith burial sites where Christians and Muslims were buried together; however, the contemporaneity of these burials was unclear.
In the following section the archeological background of the sites is presented including the sampled individuals. Additional details about the archeological background of the sites can be found in Data S1.

| Lisbon
In Lisbon, five archeological sites were sampled representing different chronologies: one site dates to the early medieval period (8-12th centuries) (Largo de Santa Marinha) while four sites (Largo das Olarias, Quarteirão dos Lagares, Poço do Borratem and Casa da Severa) date to the late medieval period (13-15th centuries).
In Largo de Santa Marinha, a total of eight rock-cut graves were identified following commercial archeological work, seven of which were primary Muslim burials located at the northern limit of the city's eastern Islamic cemetery. The burials were poorly preserved and only three individuals were analyzed from this site. No materials or grave goods were found in association with the individuals. The suggested chronology embraces the entire Islamic period in Lisbon (8-12th centuries) (Filipe et al., 2020).  (Toso et al., 2019) and are not representative of the urban population because of their small sample size and peculiar location. The excavations at the site of Largo das Olarias are ongoing and therefore an in-depth study of the chronology, use and development of the site is underway. However, the complex stratigraphy of the site, and archeological finds such as coins and pottery, suggest a late chronology for the main part of the cemetery (13-15th century). A total of 22 Muslim and 15 Christian individuals were analyzed from these two sites.
Poço do Borratem is a Christian burial site located just outside the Mouraria walls connected to the nearby hermitage of São Matheus, erected between the 12th and 13th century. The burial site is believed to be located at the edges of the cemetery in close proximity to a street and the walls of the Mouraria, and for this reason, archeologists suggest that people buried there would have had a F I G U R E 1 Distribution of the analyzed sites within the Iberian Peninsula. Full circles indicate the sites analyzed in this study, empty circles indicate comparative contemporary sites lower social status. The use of this space has been dated to the 14th and 15th centuries (Filipe et al., 2018). A group of 31 human individuals has been analyzed from this site as well as 13 faunal remains of similar chronology.
Finally, 22 faunal remains were sampled from an excavation that took place in a residential area at Casa da Severa. This site is located in the heart of the Mouraria, just over a 100 m from the Muslim cemetery of Largo das Olarias. The materials and the stratigraphic analysis from the excavation provided a chronology of use between the 12th and the 15th centuries (Valente & Marques, 2017).

| Beja
In Beja the only burial site analyzed was excavated at the Escola Secu-

| Silves
In Silves three sites were analyzed: Rua 25 Abril of early medieval chronology, and Rua Miguel Bombarda and Largo da Sé dating to the late medieval period. Rua 25 Abril is a Muslim burial site and was the main cemetery in use under Islamic rule (Santos et al., 2008). Nearly 200 individuals have been recovered and one radiocarbon date from the older layers provided a date of 895-926 cal CE (Serra, 2012). A total of 26 Muslim burials were analyzed from this site. In Rua Miguel Bombarda a Christian cemetery has been identified, probably pertaining to the Ermida de Nossa Senhora dos Martires outside the city walls. Although the church is mentioned soon after the first Christian conquest of the city (1189-1191), the materials recovered during the excavation dated to the 13-14th centuries (Casimiro et al., 2008). A sample of 19 Christian individuals was analyzed from this site. Lastly, another Christian cemetery located at Largo da Sé was sampled (n = 21) and pertains to the Cathedral of Silves. Its construction started in the 13th century however it remained unfinished until the 15th century due to the situation of political instability (Gamito et al., 1997).
Excavations suggest that this space was used as a burial site between the 13th and 15th centuries (Casimiro et al., 2008). Faunal remains (n = 16) were recovered at all sites and also from a small pit located in Rua A, in the vicinity of the Cathedral, associated with pottery dating between the 11th and 14th centuries. Because of the wide chronology they have been treated as one medieval baseline and used for both Muslim and Christian individuals from Silves.

| Diet in medieval Portugal
A detailed overview of the medieval diet in Portugal is provided in the Data S1, however dietary expectations are briefly outlined here.
Throughout the medieval period, agriculture and farming were the central economic activities and provided the main food sources for the population at large (García-S anchez, 1983(García-S anchez, , 1986(García-S anchez, , 1996. The south of Portugal did not grow much wheat and cereals were imported from the Maghreb and northern Africa during the Islamic and Christian periods (Constable, 1996) or from other regions in Portugal such as Entre-Douro-e-Minho, Beiras, Ribatejo and Estremadura during the late medieval period (Gonçalves, 1984). Among crops, wheat, a C 3 plant, was the most desirable, however secondary crops such as barley, rye and millet, a C 4 plant, were more affordable and provided food security when other resources were lacking (Catarino, 1998;Marques, 1987). The Muslim minority of Largo das Olarias living in Lisbon under Christian rule, or the lower status Christian population buried at Poço do Borratem could be likely candidates for the consumption of secondary crops as has been found in other low-status medieval Iberian populations .
Livestock farming was also one of the most important economic activities through the Middle Ages in the Iberian Peninsula (Estaca-G omez et al., 2018). During the Islamic period, the most common meats were lamb and mutton, followed by fowls and cow (García-S anchez, 1986). Goat seems to be consumed more often in the Islamic period compared to the late medieval Christian period in Iberia (Waines, 1994). This trend is also confirmed by a number of zooarchaeological studies showing a preference for sheep/goat, followed by cow, chicken and rabbit during the Islamic period and a scarce presence of pig (Morales-Muñiz et al., 2011;García-García, 2017), while cattle, sheep/goat and pig are the most recovered species at Christian sites (Grau-Sologestoa, 2017). An important food source for the lower classes, would also be secondary animal products such as cheese, milk and eggs, however these cannot be discriminated from meat by their isotopic values.

Finally, marine resources have a distinctive trajectory in medieval
Portugal. Arab authors have different opinions on whether fish was good for human consumption (García-S anchez, 1986;Pellat et al., 2012). Fish and seafood recipes are scarce in medieval Andalusian cookbooks, although we expect small-scale coastal fishing could have been undertaken by lower classes to supplement their diet (García-S anchez, 1983). Historical sources are more generous in information about the later Christian medieval period fishing practice and several species appear in city markets including the more expensive croaker, red porgy, red bream, and hake and freshwater catches such as salmon, lamprey and eel (Coelho, 1995) but also less expensive fish like sardines, sole and allis shad (Catarino, 1998;Martins, 2016). This trend could be associated with Christian abstinence from meat, however, economic ventures including contacts with the northern shores of the ocean led to the expansion of the fishing industry and brought Atlantic cod onto the table, which was salted (with salt from Aveiro and Viana do Castelo) or dried, to be preserved during the long journey back to Portugal (Azevedo, 1982;Cole, 1990). A virtual absence of fish remains from the majority of known medieval sites due to a lack of sufficient recovery techniques hinders a systematic assessment of the consumption of marine resources in Medieval Portugal.

| Sample selection
Specimens in this study were collected from three main urban sites dating to the Medieval period, both from the Islamic (8-12th century) and Christian (13-15th century) rule aiming at exploring faith and chronological changes in diet. Several sites in Lisbon, Beja and Silves were sampled to provide a representative sample of the population with equal numbers of male and female individuals whenever possible (Table 1). Only individuals that were classified as adult (over $18 years of age) in the anthropological site reports were sampled. Further adult age groupings based on age-at-death estimations were not available.
Individuals were considered as Christian or Muslim according to their funerary practice which include a different position of the body, orientation as well as layout of the grave. In addition, historical and archeological sources collected in previous excavations were used to contextualize the burial sites. Assigning religious affiliation solely on the basis of burial practices can be misleading and problematic: a person could be buried following a rite but not have lived following the specific mandates of that religion and some of these cases include individuals that converted or were forced to convert throughout their lives (crypto Christians and crypto Muslims). Although the archeological and historical sources may seem a relatively limited way of identifying the potential complexities of religious affiliation (Christys, 2009), burial rite is unfortunately the only information we have on an individual's religious affiliation, and it has been commonly used by archeologists in Iberia to differentiate between religious groups. The individuals studied here followed the standard traditional rites described for these religious communities during this period in Iberia, however, since there is no certain way to assign religious affiliation, whenever we mention a Muslim or Christian burial, we are actually referring to individuals buried with a Muslim or Christian rite. Any incongruences for example, crypto Christians and crypto Muslims, would have probably been a rare occurrence in our dataset and is unlikely to have had a significant impact on our population study. Unfortunately, no mozarab (Christians under Muslim rule) burials were available for analysis and so the diets of these communities cannot be explored.

| Radiocarbon dating
The Western Atlantic coast of the Iberian Peninsula is influenced by a dynamic upwelling which strongly affects the fluctuations in the  Soares, 1993).
Because the individuals herein presented a mixed diet with the inclusion of marine protein in their collagen (discussed below), the calibration considered the percentage of marine carbon to the diet in order to account for the marine reservoir effect. This percentage has been calculated with a BSIMM and is reported in Data S1 (Table S2)

| Peptide mass fingerprinting
Faunal remains whose species identification by macroscopic observation was hindered by fragmentation were analyzed with collagen peptide mass fingerprinting (ZooMS), largely resolving sheep/goat identifications. ZooMS was performed at the BioArCh lab facilities following a slightly modified methodology as described in (Buckley et al., 2010). In brief, a small sample ( Table 3.
T A B L E 2 Radiocarbon dates for intercutting graves of Muslims and Christians at Beja. The methodology followed for calibration is provided in Data S1  et al., 2006). We opted for a two-food source model where the relative contribution of terrestrial and marine protein to diet was considered (Cubas et al., 2018). This methodological approach was considered the most suitable to the research question as well as the most conservative, as models with multiple source foods and only two proxies (δ 13 C and δ 15 N) are often inconclusive and present larger uncertainties.
Additional details can be found in Data S1 Methodology, however, in brief, dietary proteins were considered to provide 74 ± 4% of bulk collagen carbon, while lipids and carbohydrates (i.e., energy) contributed the remainder (26%) (Fernandes et al., 2012). The model assumptions and parameters were following (Fernandes, 2016)

| Isotope analysis
Well preserved collagen was extracted from all sites in Lisbon, Beja and Silves, with collagen yields of >1% and C:N ratios between 2.9-3.6 (DeNiro, 1985) and atomic parameters within standards (C:N ratios, C% and N%) (Ambrose, 1990). The raw data for bulk collagen stable isotope analysis with indication of faith, chronology and sex for each individual is given in Data S1 (Table S5).
Isotopic analysis of a large number of animal remains (n = 57) provides both a baseline for the human diet but also information on husbandry practices. The summary data for animals from Lisbon and Silves are provided in and from 9.4‰ to 13.7‰ (Silves EM) and from 9.4‰ to 13.5‰ Any correlation present between δ 13 C and δ 15 N values in the earlier periods is weak and insignificant by comparison (Data S1, Table S1). In the case of Silves, the early medieval population shows statistically significant correlation (p = 0.032), however, the later medieval group demonstrates a stronger correlation (p < 0.000). The lack of baseline values for the most commonly consumed crops (including C 4 ) and freshwater fish did not allow for a model with additional food sources. Although BSIMMs have clear limitations, they are a powerful tool to estimate the relative importance of food groups to past lifeways (Cheung & Szpak, 2020) and help visualize the difference in protein intake between the early and late medieval period in this study. The generated individual estimates for the two models (calories and protein) for all food sources deviated less than 5% at all sites, indicating that the model outputs were robust following the given parameters.
The model provides estimations on the caloric and protein contribution of each food source (marine resources and terrestrial resources) to diet. Tables S3 and S4 show that terrestrial resources (including terrestrial animals, its derivates and plants) contributed the majority of the caloric and protein intake across periods. This is not surprising, however while the model's predictions confirm the importance of crops and domestic animals to the diet, it also shows an increasing reliance on marine resources from the early to the late medieval period. The output from FRUITS indicates that the dietary (calories and protein) contribution of marine resources show a statistically significant change at all three sites from the early to the late medieval period (Data S1, Table S3 and Table S4). The protein estimates generated by the BSIMMs reflect this trend of increasing importance of marine resources as a protein source in the late medieval period (Figure 4). The model assigns a range of 2% to 4% protein gained from marine resources in early medieval Lisbon compared to 13% to 25% of protein in the late medieval period. Similarly, early medieval Beja shows ranges of marine protein between 2% to 5% compared to 2% to 10% for the early and late medieval periods, respectively. Silves also shows an increase in the amount of protein obtained from marine resources, going from 4% to 14% in the early medieval period to 4% to 39% in the late medieval. All estimations are significantly different between the early and late medieval period at all sites (Data S1, Table S4).
No difference was found between female and male individuals at any of the new sites presented in this study and detailed results and statistical tests for each group and at each site can be found in Data S1. Social status also could not be inferred; however However, the transition in Southern Portugal is less clear since the prolonged use of Islamic pottery at many sites, hinder the identification of a precise chronology (Valente, 2019). In addition, because of the lack of direct dating of animal remains, changes in husbandry practices between the early and the late medieval period cannot be explored with this dataset.
The isotopic data indicate that these urban human populations were subsisting on C 3 plants, and terrestrial animal resources as their main dietary source especially in the early medieval period. There is some evidence for direct C 4 plant consumption by humans, although we exclude C 4 fed protein as animal diets are exclusively C 3 . Outliers Historical sources suggest that men and women had a different role in Islamic and Christian medieval societies that extended to both the public and private sphere of life (Marín, 2000). However, the isotopic results indicate that statistically significant differences in diet between males and females exist only at the high-status Muslim burial ground of São Jorge Castle in Lisbon (Toso et al., 2019). The lack of dietary differentiation between sexes at all the other sites suggests a similar access to food. Since no specific indication of high status is known for these burials, these sites are likely to represent the general urban medieval population of a middle-low status. Smaller households and spaces of multifunctional purpose might have hindered the physical separation of space that is typical for higher status households and fostered a similarity in dietary practices and food access (Diez Jorge, 2002). Sex differences in diet are a rare occurrence among published datasets of Muslim populations from Spain (Fuller et al., 2010;Salazar-García et al., 2016;Alexander et al., 2015). Significant differences between sexes have been reported for a number of Christian medieval sites in the UK such as Newark Bay, Orkney , York (Müldner & Richards, 2007), Trino Vercellese, Italy (Reitsema & Vercellotti, 2012), Giecz, Poland (Reitsema et al., 2010), Solt-Tételhegy, Hungary (Gugora et al., 2018) and the Czech Republic (Kaupov a et al., 2018); however no similar trend was found in medieval Christian populations in the Iberian Peninsula so far (Jordana et al., 2019;Lubritto et al., 2017;MacKinnon et al., 2019). Although historical sources may allude to the different role of males and females in the medieval society and their different participation in food processing and consumption, it may be that the difference is too subtle to be detected using bulk collagen isotopic analysis (Müldner & Richards, 2006). Despite the obvious limitations of the technique, further research in microhistory might prove particularly useful to explore themes of identity, gender and family organization in past societies.
Most significantly, the isotopic data, supported by BSIMMs indicates that marine fish were not a major part of the diet for any of the early medieval populations analyzed here, despite the coastal proximity of Lisbon and Silves. This is similar to what has been observed in the Balearic Islands (Fuller et al., 2010;Pickard et al., 2017), where, despite the easy access to marine resources, the diet of the Muslim population tended to be mainly terrestrial; and in zooarchaeological assemblages in Southern Portugal where little in the way of fish remains were recovered regardless of their coastal proximity (Valente & Marques, 2017;Valente & Martins, 2015). Cultural preference played a key role in food choice and cookery books and medical treatises from the time (12th-14th) pay little attention to fish and its preparations (García-S anchez, 1986). Early medieval diet, however, is characterized by high variability which is apparent both in historical (i.e., people in Cairo did eat fish) (Lewicka, 2011) (Figure 3). This suggests a mixed diet in which animal protein and potentially freshwater fish were widely consumed with marine resources only complementing the diet of a few individuals (Curto et al., 2018). However, the data from Tomar represents a wide chronology of seven centuries, which may not be a wholly representative snapshot of the diet over this time period. By the early modern period, a decline of the fishing industry was documented for Galicia (L opez-Costas & Müldner, 2018). After the 16th century in Portugal, the Crown became more interested in the expansion of trade routes in the South Atlantic and Indian Oceans as opposed to fishing, as supported by a documented increase in 'sailors' as opposed to 'fishermen' (Amorim, 2009

| Fish consumption: Disentangling faith and chronology
Christian fasting is often directly linked to the increased consumption of marine protein in late medieval Christian populations (Woolgar, 2003). In this case however, both Muslims and Christians from late medieval Lisbon show a major reliance on marine resources, which is in stark contrast to the preceding Islamic period. This suggests that fish was widely accessible in the markets of the late medieval period, not only in the northern cities, as documented by historical sources, but also in the south of Portugal where fish was widely consumed by the population, including low status religious minorities living under Christian rule (Barros, 2015). The chronological trend indicates that faith was not the only factor that shaped the diet of urban dwellers in medieval Portugal. The political shift brought about by the conquest will have played an equally important role in defining food access and resource availability, especially for urban centers.
This calls us to move beyond interpretations driven by differences in cultural preferences and faith identities and to consider wider political and economic developments occurred after the Christian conquest. Historical sources indicate that in the 13th century English rulers were among the first to grant safe conduct to Portuguese vessels to fish in their waters. Strong relationships continued through the medieval period (Childs, 1992) and in 1226, Henry III (1216-72) allowed 106 merchants from Portugal to trade freely in England for one year (Miranda, 2013). In 1353, King Edward III (1312-1377) signed a commercial treaty with Portuguese ambassadors to allow Portuguese fishermen to carry on their industry off the coasts of England and Brittany (Childs, 1992). The reasons behind this rapid expansion of the Portuguese fishing industry are multifaceted. Similar to what has been observed in England, where fish catchment areas were dramatically extended offshore to meet demand (Barrett et al., 2011), an overexploitation of in-shore resources could have initiated this phenomenon in Portugal as well. There are historical records of a number of petitions to the King and local regulations in the 15th century to stop exporting fish from Porto, Vila do Conde and Lisbon for fear of fish shortages, and speculation on fish (especially sardines and cod) became a concern for those of higher and lower status alike (Amorim, 2009). What prompted this demand-driven intensification of local fishing is difficult to pinpoint. It is undoubtable that after the Christian conquest and conversion, Christian fasting will have been a factor in higher demand for fish supply, however it was notthe only reason behind this new interest in fish (McCleery, 2018).
Fishermen were held responsible for sustaining such eating habits by royal and municipal laws, especially during Lent (Moreira, 1987). However, population growth and the shortage of other foodstuffs such as grain, will have also contributed. Historical records report a population explosion in the 16th century, coinciding with an economic growth of coastal towns thanks to rising levels of Atlantic trade. For example, the population of Lisbon went from 70,000 inhabitants in 1528, to 100,000 in 1550 and 165,000 in 1619 (Amorim, 2009). Fish was an affordable source of protein and the demand to feed growing urban centers will have increased the pressure on food resources. Some species remained expensive and accessible only to the few, both marine (croaker, red porgy, red bream, and hake) and freshwater (including lamprey and eel). However, less expensive fish (sardines, sole and allis shad) were in abundance and would be easily accessible by those of lower status (Martins, 2016), like the Muslim community living under Christian rule in Lisbon in this study. The cost and the availability of fish coupled with the easy access to salt along the whole Portuguese coastline coincided to expand fishing activities into the creation of a wider fishing industry (Antunes, 2008;Newstead, 2014).

| CONCLUSIONS
In this study, isotopic data supported by BSIMMs has provided unique insight into the contribution of different food sources to individual diet in combination with archeological and historical data. Results indicate that the late medieval population obtained more of their dietary intake from marine resources, equally across both sexes and faiths indicating that this resource was widely available in significant quantities to all strata of society, including religious minorities. The change from a predominantly terrestrial diet to one with a heavy marine component provides unprecedented archeological evidence for the prelude to the early-modern Portuguese marine economy and its global expansion.

DATA AVAILABILITY STATEMENT
The data that supports the findings of this study are available in the supplementary material of this article and will be stored in the IsoArcH repository (https://www.re3data.org/repository/ r3d100013383).