Biting into the truth: Connecting oral pathology and stable isotopes through the paradigmatic example of a hyper-specialized marine diet in Medieval Pontevedra (NW Iberia)

Abstract

Human feeding patterns have been reconstructed in Archaeology by analysing either oral pathology or stable isotope ratios in human skeletal remains. However, no clear agreement has been developed between these two methodologies. The main objective of this study is to determine if we can establish a link between them when analysing a population with a hyper-specialized diet, in this case marine resources (and millet/maize). To reach this goal we developed a conjoined multi-isotope analysis using collagen and bioapatite (δ¹³C_col, δ¹⁵N_col and δ¹³C_car) and a detailed study of oral health (caries, antemortem loss, periapical lesions, periodontal disease, calculus, and wear). All available skeletons with at least one preserved tooth from two cemeteries of the medieval town of Pontevedra (n = 34) were studied. The buried individuals belonged to the guild of fishers and artisans, professions which were dominant among the families of medieval Pontevedra. A detailed FTIR-ATR study of extracted bone bioapatite showed a high correlation between bioapatite carbonate content, carbonate typical vibrations, and FTIR-ATR indices related to bone diagenesis, which is in line with previous research. No significant correlations were found with bioapatite yield and isotopic composition (δ¹³C_car and Δ¹³C), ruling out possible diagenetic effects. The diet was based on marine fish protein with contributions of millets (e.g., δ¹³C_car -11.9 ± 1.8‰) that seems to be slightly higher in individuals linked to artisanal guilds. The oral pathology study shows severe dental wear from an early age (Grade 2–4 in permanent dentition for 20% of infants and 60% of juveniles in M1), as well as moderate-high presence of caries in permanent dentition (64%, 22/34) and dental calculus (72%, 24/33). Both the oral pathology and the isotopic signal differ from that observed in other areas of the Iberian Peninsula. This study points to the existence of connections between findings of the two methodologies, and specifically an association between intense dental wear and high consumption of marine resources and millet. At the same time, this analysis implies the necessity of caution in estimation of age by dental wear in populations linked to the sea.

Introduction

Reconstructing diet through the study of human remains has always been a prominent goal for researchers working in Biological Anthropology and Osteoarchaeology. Skeletons can provide direct information about an individual diet or population consumption patterns—in contrast to archaeological materials and animal/plant remains (which provide indirect information). This information can be linked to other features inferred from the burial and skeleton, for example, funerary ritual, health, social status, and biological profile. Traditional methods include oral pathology, which has been used for dietary reconstruction since the advent of paleopathology. There is a link between the type of diet and the pathological features observed in the teeth and mouth (Hillson 1996; Bonsall and Pickard 2015). Teeth are the only part of the digestive system in direct contact with the external environment and, in consequence, cultural practices can modify oral pathology too (Cheung et al. 2019; Yanko et al. 2021). Individual physiology can also change the development of oral pathology; for example, a given individual can be more/less reactive to oral biofilm and this sensitivity can evolve with age or specific circumstances such as pregnancy (Hillson 2001; Bonsall and Pickard 2015; Kubehl and Temple 2020; Schats et al. 2022). Since the 1970s, geochemical approaches to diet have increased in importance, with stable isotope ratios being the most widely used today. Research on light stable isotope ratios in bone and tooth collagen and bioapatite (e.g., δ¹³C_col, δ¹⁵N_col, δ¹³C_car) has allowed us to reconstruct the type of food consumed during different periods of life. However, even though the reconstruction of diet through oral pathology and stable isotopes have coexisted for more than forty years and pursue a common goal, specialists are yet to find a clear relationship between them. In recent years, interest in linking stable isotope fingerprints with paleopathological markers has grown. Some agreement has been achieved for fractures (Curto et al. 2020), metabolic markers (Mangas-Carrasco and López-Costas 2021), and infectious diseases (Curto et al. 2019; Katzenberg and Lovell 1999), but the relationship with oral pathology remains elusive (see for example Murphy et al. 2013; Beck et al. 2018; Petersone-Gordina et al. 2018; Jílková et al. 2019; Dlamini et al. 2019). Different authors have found slight and inconsistent relationships of δ¹³C_col with caries and dental wear (Toso et al. 2019; Bonsall and Pickard 2015; Valentin et al. 2006), or no clear pattern (Petersone-Gordina et al. 2018; Lillie et al. 2003). A high frequency of dental caries has often been linked to specific inputs in diet such as consumption of carbohydrates (low δ¹³C_col and δ¹⁵N_col), but the results obtained so far do not show consistency (Toso et al. 2019; Inskip et al. 2019). The disagreement can be explained in part as a reflection of the fact that the two methods reflect different periods of the human diet: oral pathology normally reflects diet throughout life, but it can change quickly to respond to physiological stresses, while stable isotopes in bone reflect an average diet that can go from ten years to multiple decades and can also be modified by stress but in an opposite way than caries (high δ¹⁵N_col). In addition, both isotopic ratios and oral diseases are affected by cultural, social, and physiological aspects that may not necessarily be expressed in similar ways.

Most of the published comparisons between oral pathology and stable isotopes have been made in populations with an ‘average’ diet, which implies a vegetable base with a variable intake of terrestrial animal protein. When comparing average diets, the other confounding factors can be overexpressed easily. Due to this reason, we think that research in populations with an extreme or hyper-specialized diet can provide a clearer view of the connection between the two methods. In addition, widening the spectra of the recorded oral pathology, rather than caries, and light stable isotopes, instead of δ¹³C_col and δ¹⁵N_col, can also help to unveil associations between the two most widely used methods for diet reconstruction in osteoarchaeology. To pursue that goal, we have selected Pontevedra on the coast of Galicia, Northwest Iberia (Fig. 1), whose medieval and post-medieval skeletal collections have revealed one of the highest consumptions of marine resources in Europe (López-Costas and Müldner 2016, 2019). Textual sources, archaeological materials, and osteological remains all suggest that the medieval diet of the Pontevedra inhabitants was highly specialized in fish and shellfish with important intakes of C₄ plants (millets and later corn) (López-Costas and Müldner 2019; López Costas 2012), that contrast with other contemporary communities in Iberia. This is also one of the few regions in Iberia that remained under Christian domination for almost all the medieval period, a fact that may have helped to homogenise cultural preferences in diet over time. For all these reasons, the Pontevedra skeletal collections offer an excellent opportunity for our aims. Nevertheless, the area also has a disadvantage, as the highly acidic soils and high humidity of Northwest Iberia produce the alteration of the mineral part of the bone, causing intense diagenesis (López-Costas et al. 2016), and limiting the availability of large skeletal data sets.

Fig. 1

Modified from López-Costas and Müldner (2019)

Map of Pontevedra location including the necropolei of Santa María (nº1) and San Bartolomé (nº2), and the archaeological sites of Plaza España (nº3), Ponte do Burgo (nº4) and Ampliación do Museo de Pontevedra (nº5). The identifiers nº3, nº4 and nº5 correspond to medieval sites from which the animal samples of López-Costas and Müldner (2019) referenced in the isotopic comparison of this study were extracted.

In this work, we analyse all available skeletons from medieval (13th-15th) and post-medieval (16th-17th) times found at the Pontevedra necropoleis of Santa María a Mayor and San Bartolomeu churches. Stable isotopes in bone collagen (δ¹³C_col, δ¹⁵N_col) and bioapatite (δ¹³C_car) were performed on all skeletons with at least one preserved tooth. Finally, an oral pathological assessment was done, recording not only caries but also antemortem tooth loss (AMTL), periapical lesions, periodontal disease, dental calculus, dental wear, and hypercementosis. Our findings are compared with those of other Iberian collections. Our hypothesis is that a trend between oral pathology and stable isotopes can be found when analysing a hyper-specialized diet, and that once this trend is discovered it can be transferred to other types of diet and skeletal collections.

Material and methods

The archaeological collections of Pontevedra

During the High Middle Ages (eleventh-thirteenth centuries) there was relative economic prosperity in northern Iberia due to the richness of the harvests, derived from the Medieval Climatic Optimum (10th to thirteenth centuries) (Le Roy 1971; Oliva et al. 2018) as well as other social factors such as the relative economic importance of the Camiño de Santiago. In contrast, during the Late Middle Ages (14th to fifteenth centuries) there is evidence of a gradual worsening of climatic conditions – the Little Ice Age phenomenon-, which, alongside various other factors, significantly affected the modus vivendi and the cereal-based production system of the aforementioned societies (Priegue 1998; Oliva et al. 2018; Silva-Sánchez et al. 2022). The Late Middle Ages were characterised by a decrease in the production of cereals, with direct repercussions on the economy of medieval societies in the form of economic impoverishment. However, in the transition to the Late Middle Ages, a few towns, whose economy was based on hyper-specialised commerce, and with the support of the emerging royal families, became independent of religious power. Among them, Pontevedra stands out for its specialization in the capture, processing and distribution of fish and other marine products. Its economy was primarily based on fishing, processing and commercially-derived products; it constituted, together with Padrón and Noia, one of the three coastal towns that were authorised to trade fish internationally by sea (Armas Castro 1992; Priegue 1998). Pontevedra’s seafaring families acquired different royal privileges as well as commercial advantages such as the commercial monopoly on three fish species (sardine, conger, and hake) and one marine mollusk (octopus) (Armas Castro 1992). Those species were easily preserved by salting or smoking and were in high demand abroad due to the religious precepts regarding fasting. The Ría de Pontevedra (a shallow firth or “fjord”) is one of several of the Rías Baixas which are known for their seafood and abundance of fish (Armas Castro 1992; Mangas-Carrasco and López-Costas 2021). The settlement grew around the church of Santa María a Maior dos pescadores (that is, the church of ‘the fishermen’), which is visible from the port and which dominates the bridge at the mouth of the navigable Lérez River (López-Costas and Müldner 2019; Armas Castro 1992). This parish comprised sea-connected families and guilds. In the thirteenth century, Pontevedra’s urban area increased, and the church and parish of San Bartolomeu appeared at this stage, encompassing the most important artisan guilds. Both churches and their respective cemeteries are located inside the medieval wall of Pontevedra (Filgueira Valverde 1991 in López-Costas and Müldner 2019) (Fig. 1). The inhumations here comprise a chronological window that spans four hundred years (13th to seventeenth centuries). The chronological ascription was made on the basis of field-work records, stratigraphy and radiocarbon dating (explained in López-Costas and Müldner 2019). Textual sources from this period suggest that the entire community consumed fish; these products formed the main base of the diet (Armas Castro 1992; Puig 2012). However, seafood consumption is associated in historical sources with poorer social classes, showing a cultural bias against the consumption of this type of food among the middle class in Pontevedra (Priegue 1998). Small vegetable gardens for fresh products as well as the introduction of cereals (including millet) from neighbouring farmlands provided the vegetable component of the diet. Terrestrial meat appears to have been scarce, as only two butcher shops were registered (see López-Costas and Müldner 2019), but eggs could be consumed from domestic fowl. Exported fish and octopus were either smoked or salted in the basement of family houses, since the humid and rainy environment of this area of Iberia made this unviable outside.

Skeletal collections

All human remains were excavated between the years 2007–2009 during two different archaeological rescue campaigns in the centre of Pontevedra, areas that are today the church of Santa María and the Teatro Principal (built after the demolition of the church of San Bartolomeu). They were found in layers dated to the late Medieval and Early Modern period, 13th to seventeenth centuries (for a detailed explanation about necropolis layers and dating see López-Costas and Müldner 2019 and Chao Álvarez 2008). The churches of Santa María and San Bartolomeu were separated by ~ 100 m and belonged to the same urban complex. Their churchyards can be considered typical of Christian burial spaces, with substantial reuse of the space and W-E oriented primary supine position burials (see Mangas-Carrasco and López-Costas 2021; López Costas 2012; Chao Álvarez 2008). Inhumations were simple fossae excavated directly from the soil, with an absence of coffins or other evidence that suggest the previous presence of coffins (e.g., nails). The individuals belonged to a non-diverse urban middle class, whose activities were related to seafaring and various craft activities. This is consistent with the differences between Pontevedra burials and ad sanctos burial type, which were reserved for social elites (Armas Castro 1992; López-Costas and Müldner 2019; MacKinnon et al. 2019) and which were not found here. Many burials had been previously disturbed by construction and infrastructure works. In the Santa María churchyard, remodelling of the water pipes system affected specifically the upper part of many skeletons, and in consequence most crania were lost. In San Bartolomeu, Survey nº16 produced a high number of skeletons, but high physical alteration due to pressure disturbances caused the collapse and fragmentation of several skulls. A previous study of these collections (López Costas 2012) indicates that unfortunately the cranium was the most altered and missing part of the skeleton in both sites.

Sex and age estimation, and demographic profile

Markers for anthropometry, sex, and age were analysed using international standards suitable for Iberian populations. Sex was estimated for adults and older adolescents (> 16 years old for females and > 18 years old for males). Sex determination followed internationally established markers on the innominate and cranial bones (see Buikstra and Ubelaker 1994), aided by metrical analyses of long bones (Aguilera 1997). Age of non-adults was estimated using international (Scheuer and Black 2004) and Iberian (Lopez-Costas et al. 2012) postcranial developmental standards for growth and maturity, as well as dental development (Schour and Massler 1941; Gustafson and Koch 1974; Van der Linden and Duterloo 1976; Ubelaker and Ripley 1999; Alqahtani et al. 2010). Age for juvenile individuals was also aided by using postcranial fusion of the centres of ossification (Brothwell 1981; Gray et al. 1988; Krogman 1986) and checking them with western European datasets for the sacrum and long bones (Ríos et al. 2008; Cardoso 2008a, 2008b; Rissech 2008). Adult age was assessed using methods developed for the 4th rib, pubic symphysis morphology and auricular surface in the innominate bones, but also considering other features such as cranial suture closure or changes in the clavicle (see Buikstra and Ubelaker 1994). For a better explanation of the selected methods see López-Costas (2012). The sample was divided into five categories according to age estimation: infants/children (1–12 years old), juveniles (13–19 years old), young adult (20–39 years), middle adult (39–60 years), old adult (60 + years). Further information about age methods and categories can be found in López-Costas (2013). The sample comprises no individuals under 1 year old. We are conscious about the problems regarding the artificial group of Infant/Children. We decided to include both categories together in tables due to the small sample size (n = 7; four individuals up to six years of age; three individuals up to twelve years of age); however, discussion regarding these individuals will be done considering their specific age.

Of the total skeletal assemblage (n = 82), only 34 skeletons (25 Santa María: 9 San Bartolomeu. 14 females: 9 males; 11 indeterminate) were included in this study since they have part of their maxillae and/or at least one tooth (in situ and in avulsion). The skeletal collections are curated at the Universidade de Santiago de Compostela, in the facilities of the research group EcoPast. The demographic distribution can be observed in Table 1. The most represented category is young adult, ~ 44% of the sample; of those ~ 17% females and ~ 21% males. The ~ 35% of the sample were non-adults: ~ 20% of them were less than 12 years old and ~ 15% juveniles. Note that the number of females over 40 years old is slightly higher (n = 4) than the males (n = 2), and it is the same with juvenile females (n = 4) compared to males (n = 0). This means that the female age distribution is wider while most males are young adults (20–39 years old; see Table 1).

Table 1 Age and sex distribution of the analyzed assemblage considering together Santa María and San Bartolomeu churches

Isotopic and FTIR-ATR analyses

The isotopic study was designed to complement a previously published work exclusively focussed on adult paleodietary reconstruction through δ¹³C_col and δ¹⁵N_col in bone collagen (> 12 years old) from the two collections (López-Costas and Müldner 2019). Here we performed two types of isotopic analyses: 1) to address protein diet with stable isotopes (δ¹³C_col δ¹⁵N_col n = 34) by completing the previous bone collagen analyses including those individuals (n = 7) that were not included previously; 6 infants/children and one juvenile (Table 4). 2) To reconstruct the total diet by knowing the bioapatite isotopic fingerprints (δ¹³C_car n = 30 up to 34; 4 samples failed during the extraction) with the bioapatite extraction in the same subsample used previously for collagen extraction. All analyses were performed at the EcoPast research group’s clean-lab at the Universidade de Santiago de Compostela. Before processing, all bone samples were cleaned for at least 5 × 5 min ultrasonic cleaning with ultrapure water type-1 at the same facility.

Collagen was extracted in 7 samples following Longin procedure (1971) with modifications recommended by Collins and Galley (1998). The protocol described by Müldner and Jay (2008) and described in López-Costas (2012) explains each step followed in the laboratory. Carbon and nitrogen isotope analysis in collagen was undertaken by Elemental Analysis—Isotope Ratio Mass Spectrometry (EA-IRMS) at Iso-Analytic Inc. using a Europa Scientific elemental analyser for the samples. Reference materials used for the δ¹³C_col and δ¹⁵N_col analysis are listed in Table 1 of supporting materials.

Bioapatite was extracted following Garvie-Lok et al. (2004) at the Universidade de Santiago de Compostela in the clean-lab facilities of the research group EcoPast (for more details see López-Costas et al. 2021, where the protocol is explained step by step). Carbon isotope ratios in carbonate from bioapatite were measured using a Europa Scientific 20–20 IRMS by adding phosphoric acid and measuring CO₂ by continuous flow-isotope ratio mass spectrometry (CF-IRMS) at Iso-Analytic Inc. Reference materials used for this analysis are listed in Table 2 and NBS-18, NBS-19 are distributed by the international Atomic Energy Agency (IAEA) as inter-laboratory comparison standard materials as well as internal or external standards. Analytical error was calculated by repeated analyses of internal standards and was ± 0.2‰ or less for all elements.

Finely milled bioapatite samples (27) were analyzed by Fourier-transform infrared spectroscopy in attenuated total reflectance mode (FTIR-ATR), using an Agilent Technologies Cary-630 spectrometer, equipped with a diamond crystal, and hosted at the EcoPast laboratory of Universidade de Santiago de Compostela, Spain. Spectra were acquired in absorbance mode, in the mid-infrared region (4000 to 400 cm⁻¹), using 4 cm⁻¹ resolution and averaging 100 scans per sample. The equipment was thoroughly cleaned, and a background was collected before each measurement. The R package {baseline} (Liland et al. 2010) was used for baseline correction of the spectra, to avoid bias in the spectroscopic signal due to scattering, reflection, temperature, concentration, or instrument anomalies; and the R {andurinha} package (Álvarez Fernández and Martínez Cortizas 2020) was applied to determine peak position (i.e., wave numbers of the peaks) based on the second derivative spectra.

Five indices usually applied in bone research to evaluate bone diagenesis and composition were applied, the IR splitting factor (IRSF), B-type and A-type carbonate substitutions in phosphate (BPI, API), the carbonate/phosphate ratio (C/P), and carbonate/carbonate ratio (C/C) (France et al. 2020). Following the minimum standard criteria proposed by Smith et al. (2023), the highest and lowest wavenumbers of peaks and troughs were calculated for each single spectrum. Smith et al. (2023) also reviewed the application of the C/P ratio and found that different authors propose different ratios involving the use of the vibration at peak position of the phosphate stretching (around 1019 cm⁻¹) or at the slope of the peak (at 1035 cm⁻¹), arguing that the second is more stable than the former. Thus, we calculated the C/P ratio in both ways: C/P_1 using the peak absorbance of phosphate and C/P_2 using the absorbance at 1035 cm⁻¹.

Oral pathology

Oral pathology analysis was performed by different macroscopic studies on 458 teeth (belonging to the 34 individuals analyzed in this study, including both permanent and deciduous teeth), analysing 723 lesions. Dental caries, antemortem tooth loss (AMTL), periapical lesions (PL), periodontal disease (PD), dental calculus, dental wear and hypercementosis were recorded. Tooth type, laterality and arch in cases of avulsed teeth were determined according to Hillson (1996). The presence, type and location of carious lesions was identified according to Chimenos (2003). For the severity of periodontal disease, the standards of Ogden (2007) in Nelson (2015) were followed. Periapical lesions were studied based on the recommendation proposed by Chimenos (2003). The presence and severity of calculus was recorded using the standard proposed by Buikstra and Ubelaker (1994). Tooth wear was considered following Smith (1984) for the anterior dentition and premolars and Brothwell (1981) for the posterior dentition (M1, M2). The eight classical categories of dental wear were merged into four to enhance the robustness of the analysis due to the small sample size (premolars and anterior dentition: Grade 0–1, Grade 2–4, Grade 5–7, Grade 8; posterior dentition: Grade 0–1, Grade 2–4, Grade 5–5 +  + , Grade 6). The first group Grade 0–1 corresponds to the absence of wear and may present small facets; Grade 2–4 corresponds to moderate wear of the crown including exposure of the dentine on the cusps; Grade 5–7/Grade 5–5 +  + (anterior dentition and molars/posterior dentition) corresponds to exposure of large areas of dentine, and Grade 8/Grade 6 (anterior dentition and molars/posterior dentition) corresponds to severe reduction of the crown height and may also include exposure of the roots. An average dental wear score of anterior and posterior dentition was performed for each individual. First and second molars in both maxillae and mandibles were selected to enhance the comparability with other studies, excluding third molars due to the variable age of its eruption. Anterior dentition was excluded from the dental wear comparison analysis as few Iberian collections address them as well as in the cases that are referenced, methodologies diverge. Hypercementosis was recorded following the descriptions in Tang et al. (2015).

Statistical analysis

Statistical analyses were performed with Rstudio calculating basic descriptive statistics for each group of samples. Shapiro–Wilk tests were used to test normality. We have applied non-parametric (Wilcoxon Mann–Whitney U-test, Kruskal–Wallis tests) and parametric (Student’s-t test) tests as necessary to explore comparisons within groups. Correlation between variables was explored through Fisher exact tests due to the small sample size. The p-values were considered significant when p < 0.05.

Results

FTIR-ATR evaluation of bioapatite integrity

The FTIR-ATR spectra of the bioapatite samples show the characteristic vibrations of its main mineral constituents, phosphate and carbonate (Fig. 2). The most intense absorbance vibration occurs at 1019 cm⁻¹ corresponding to the asymmetric stretching of PO₄³⁻. It is complemented by PO₄³⁻ symmetric stretching at 958 cm⁻¹, and the doublet of bending vibrations at 602 and 559 cm⁻¹ (Figueiredo et al. 2012). The shoulder at 1094 cm⁻¹ is attributed to acidic phosphate anion (HPO₄²⁻). Carbonates are also present, as attested by their stretching vibrations and 1455 and 1415 cm⁻¹, and the bending vibration at 874 cm⁻¹. A minor absorbance band is also observed at 475 cm⁻¹ and attributed to PO₄³⁻ (Jastrzebski et al. 2011). No vibrations of secondary products formed upon diagenesis (as secondary carbonates) were found, as for example the peak at 710 cm⁻¹ that is usually related to secondary carbonates formation (Lee-Thorp and Van der Merwe 1987; Smith et al. 2023).

As for the IR indices, we use here the range values suggested by France et al. (2020) for well-preserved archaeological bone as a reference for the values obtained in the samples we analysed (Fig. 3). Nevertheless, the comparison must be considered with care as we calculated the indices on extracted bioapatite and those of France et al. (2020) are based on whole bone measurements. Overall, values for all indices fall in narrow ranges (IRSF 4.34 ± 0.35 avg ± std; C/P_1 0.10 ± 0.02; C/P_2 0.12 ± 0.02; C/C 1.00 ± 0.06; BPI 0.26 ± 0.05; API 0.26 ± 0.03).

Fig. 2

FTIR-ATR spectra of the bioapatite samples studied in this investigation, with indication of the position (wavenumbers) of the most characteristic vibrations. The region 4000–1600 cm-1 is not represented because it showed background absorbance

Fig. 3

IR calculated indices. Lines correspond to the lower and upper limits of well-preserved bone bioapatite (after France et al. 2020). Blue lines in the BPI-API panel correspond to API and orange lines to BPI limits. C/P: black circles C/P_1, gray circles C/P_2; BPI black circles, API gray circles

The IRSF values are above the range for well-preserved archaeological bone bioapatite for all except three samples; for C/P, both C/P_1 and C/P_2, almost half of the samples fall within the range, but are close to the lower limit; C/C values are well within the range, except for six samples; for BPI all samples are close to the lower limit, but for API all samples fall within the range of well-preserved bioapatite.

In terms of information content, the calculated indices are largely redundant. Very high correlations (> 0.90 or < -0.90) were found between IRSF, C/P indices, and C/C; C/P_1 and C/P_2; BPI and C/P_1 and C/P_2, and API and C/P_2 (Table 2). The lowest correlations were found for API, with the IRSF and C/C (Table 2).

Table 2 Correlation between the IR indices calculated for the samples of this study

We also calculated the correlation between the indices and the characteristic absorbances of the bioapatite components (i.e., phosphate and carbonate vibrations) to bioapatite yield, carbonate content, δ¹³C_car in bioapatite carbonate, and Δ¹³C (Table 3). Strong positive and negative correlations were found between the indices and characteristic vibrations of carbonates with bioapatite carbonate content. C/P_1, C/P_2, API, BPI, and vibrations at 1455, 1415 cm⁻¹ are positively correlated (r 0.77–0.93), while the IRSF, and C/C are negatively correlated. Vibrations at 1019 cm⁻¹ (maximum absorbance of phosphate) and 874 cm⁻¹ (bending vibration of carbonates), show moderate negative and positive correlation, respectively. A strong linear correlation between C/P and carbonate content of the bioapatite has been previously noted (Featherstone et al. 1984; Smith et al. 2023).

Table 3 Correlation between the IR indices, characteristic vibrations (in cm⁻¹) of bone phosphate and carbonate, bioapatite yield (%), carbonate content (%) of the bioapatite and the isotopic composition

In the highly acidic (pH < 5) environment of the necropoleis studied here, carbonates are expected to be more susceptible to chemical alteration and leaching (i.e., diagenesis) than phosphates. This agrees with our findings, since increases in IRSF (i.e., phosphate crystallinity) and C/C (i.e., increase in B-type over A-type carbonate substitution), which are usually interpreted as evidence of diagenesis, are accompanied by decreases in bioapatite carbonate content – reflecting carbonate depletion, possibly by leaching of the weathering products.

Interestingly enough, no index or vibration is even moderately correlated to bioapatite yield or bioapatite isotopic composition. Our data suggest that, for the studied samples, diagenesis may have resulted in mass loss of the carbonates of the bioapatite, and possibly also of the phosphate to a lower extent, but the remaining carbonate preserves the original isotopic composition. This is consistent with the lack of evidence of exogenous carbonates incorporation to bone. As already indicated, the geochemical environment of the studied necropoleis is compatible with strong carbonate leaching but not with the formation of secondary carbonates.

Isotopic analyses

The isotopic study for non-adult and adult bone collagen is summarized in Table 4. The non-adult results are plotted against the total isotopic sample of López-Costas and Müldner (2019) in Fig. 4. The seven samples analyzed fulfilled the established quality criteria for good collagen preservation (see Van Klinken 1999). The C/N ratios of all our samples range from 3.1–3.4, which is considered a sign of no aberrant results and no presence of organic contaminant molecules (following DeNiro 1985; Van Klinken 1999). We calculate the Pontevedra site sample’s average (including previous results from López-Costas and Müldner 2019) for δ¹³C_col is -16.4 ± 2.1‰, n = 68 (Santa María -16.1 ± 2.5‰, n = 49; San Bartolomeu -16.7 ± 0.8‰, n = 19) and for δ¹⁵N_col is 12.6 ± 0.9‰ n = 68 (Santa María 12.8 ± 0.9‰, n = 49; San Bartolomeu 12.2 ± 0.8‰, n = 19).

When considering just the 34 skeletons with well-preserved maxillae and in agreement with previous results from (López-Costas and Müldner 2019), when comparing Santa Marías and San Bartolomeus individuals (n = 34) no significant difference is found for δ¹³C_col (U-test 538.5, p = 0.40), while Santa Maria individuals have consistently more elevated δ¹⁵N_col values than those from San Bartolomeu (U-test 699.5, p = 0.0026). Regarding δ¹³C_col including previous results from López-Costas and Müldner 2019) of adult individuals two groups were identified, where most of the individuals display values between -18.5‰ and -14.6‰, while a small group shows values > -12.5‰ (STM824, STM828, STM830, STM833, STM883) (see Fig. 4) (López-Costas and Müldner 2019). This pattern of two separate groups is also observed for the 34 individuals analysed in this study (Fig. 4) presenting three individuals with values > -12.5‰ (STM823, STM824, STM883). Further, one individual (STM825) presents a δ¹³C_col value lower than the rest of the sample reaching -20.3‰.

Table 4 Collagen stable isotope ratios and quality indicators for San Bartolomeu and Santa María churches. In this table we included the new isotopic values for the 7 non-adults as well as those obtained in López-Costas and Müldner (2019)

Fig. 4

(Left) Bone collagen δ¹³C_col and δ¹⁵N_col values of animal and human sample from Santa María and San Bartolomeu of Pontevedra including isotopic values from López-Costas and Müldner (2019). (Right) Bone collagen δ¹³C_col and δ¹⁵N_col values of the 34 human samples from skeletons with at least one preserved tooth from Santa María and San Bartolomeu of Pontevedra analyzed in this study. Created with RStudio

Collagen data also suggests that the Infant/Child (1–12 years old) assemblage is distributed in two groups independent of their cemetery – Santa María and San Bartolomeu, provenance, see Fig. 4. Individuals SB519, SB520 and STM875 presented high δ¹⁵N_col values (> 13.0‰), and a wider distribution in δ¹³C_col (-15.3‰ to -17.3‰); whereas individuals SB518 and STM877 showed lower δ¹⁵N_col values (< 12.0‰) and a narrower δ¹³C_col distribution (-17.1‰ to -17.4‰). The juvenile STM876 is also included into the low δ¹⁵N_col non-adult group. Individual STM883 belongs to the aforementioned group due to the high δ¹⁵N_col (13.6‰) and δ¹³C_col (-11.7‰), but her/his values are among the adult group with high δ¹³C_col. When the whole non-adult assemblage was compared to the adult assemblage (Fig. 4) (López-Costas and Müldner 2019) no significant differences were found in δ¹³C_col (U-test 300.5, p = 0.52) or δ¹⁵N_col (U-test 311.5, p = 0.63). Individual STM825 showed δ¹³C_col values reaching -20.3‰ while in terms of δ¹⁵N_col showed the minimal value of 10.8‰. When the comparisons between non-adults and adults were done considering each cemetery separately, again no significant differences were found (San Bartolomeu, non-adult SB518, SB519, SB520 to adults; δ¹³C_col U-test 13.00, p = 0.318; δ¹⁵N_col U-test 16.00 p = 0.09), (Santa María, non-adults STM875, STM876, STM877, STM883 to adults; δ¹³C_col U-test 38.0 p = 0.91; δ¹⁵N_col U-test 33.0, p = 0.63). However, Infants/Children belonging to the first group are among the most elevated in δ¹⁵N_col of San Bartolomeu and Santa María.

Isotopic data for bone bioapatite are presented in Table 5. Established quality criteria for well-preserved bioapatite were reached by all samples (n = 29). There were no significant FTIR-ATR bands that indicate the presence of amide A, B, I, II and III, that is, collagen; or aliphatic compounds (for more information about collagen bands see Martínez Cortizas and López-Costas 2020) – only carbonate and phosphate bands were detected.

Table 5 Carbon stable isotope ratios in bioapatite and quality indicators for Santa María and San Bartolomeu churchyards

For bone bioapatite isotope analysis, average and standard deviation for δ¹³C_car is -11.9 ± 1.8‰, n = 29 (Santa María -11.9 ± 2.1‰ n = 22; San Bartolomeu -11.6 ± 0.8‰, n = 8) with values ranging from -14.2‰ to -4.7‰. For δ¹³C_car, no significant difference was found between San Bartolomeu and Santa María for δ¹³C_car (U-test 50, p = 0.08), but Santa María’s δ¹³C_car values tend to be higher than those from San Bartolomeu. There are no significant differences between non-adults and adults from Santa María (δ¹³C_car U-test 73, p = 0.16) or for the San Bartolomeu site (δ¹³C_car U-test 26, p = 0.12). The total assemblage shows a smooth distribution, but the non-adult individual SB520 separates from the group due to his/her high δ¹³C_car value (-10.6‰). There are also no significant differences between males and females (San Bartolomeu δ¹³C_car U-test 5, p = 0.64; Santa María δ¹³C_car U-test 13, p = 0.18).

For Δ¹³C_car-col, the average of the total sample is 4.71 ± 1.12, ranging from 1.95 (STM882) to 7.87 (STM825); Santa María 4.46 ± 1.24; San Bartolomeu 5.32 ± 0.71. Although no significant differences were found between both sites when considering the whole sample; Santa María adult samples (n = 20) are significantly lower in Δ¹³C_car-col than San Bartolomeu adult (n = 9) samples (U-test 75; p = 0.047), a trend that despite the small sample size can be seen in Fig. 4. Individual STM825 showed the maximum value of Δ¹³C_car-col (7.87) and the minimum value of δ¹⁵N_col (10.8‰) with respect to the total set. Individual STM882 showed the minimum value of Δ¹³C_car-col (1.95) while its δ¹⁵N_col is over the average (13.6‰). No significant differences were found among age groups or sex.

Oral pathology

Dental caries

With regard to the study of caries, 64% (22/34) of the individuals showed some type of lesion. Carious lesions were displayed in seventeen individuals from Santa María’s churchyard (n = 17/25; 68%) and five from San Bartolomeu (n = 5/9; 59%), showing similar prevalence and with no significant differences between them. Within the non-adult sample, 33% (4/12; Santa María, n = 2; San Bartolomeu, n = 2) of the individuals showed carious lesions in the deciduous dentition, and 33% (4/12; Santa María, n = 3; San Bartolomeu, n = 1) in the permanent dentition. At least one carious lesion is observed in 60% (12/21) of the adult sample (n = 6 males, n = 6 females). Regarding age, 53% (n = 8/15) of the young adults displayed some type of carious lesion (1:1 male:female individuals), middle adults showed 60% (n = 3/5; 40% female individuals, 20% male individuals) and the only old adult showed carious lesions. No differences were found between churchyards (U-test 79, p = 0.47) or sexes (Kruskal–Wallis chi-squared = 4.9, p = 0.08), adult/non-adult, and age groups. Note that four individuals presented a greater number of carious lesions (STM876, STM824, STM825, STM833), three of them being female. As expected, there are significant differences between adult and non-adult groups (U-test = 28.5, p = 0.003) for permanent teeth. There are no differences between adult age groups (U-test = 36, p = 0.83), nor between juveniles and adult age groups (U-test = 28, p = 0.69); there are differences between the Infant/Child age group and the Juveniles (U-test = 0, p = 0.003) as the Infant/Child group did not show any carious lesions in the permanent dentition (0/82). Therefore, a significant pattern of increased lesions with age is not observed, except among the Infant/child group and the rest of the sample.

Table 6 displays the prevalence of carious lesions in the teeth. For the total tooth sample, 10% of the teeth presented some type of carious lesions. Among the adult age group, 18% of permanent teeth were affected by at least one carious lesion, 11% of the permanent juvenile teeth also presented lesions. In contrast, the infant/child group (< 12 years) showed 0% in the permanent teeth, while 24% of the deciduous teeth were affected. They did show significant differences between non-adults and adults (U-test 51.5, p = 0.007), and with the adult group (U-test 0, p = 0.001) without observing significant differences between juveniles and adults (U-test 13, p = 0.51), or adult groups (U-test 42, p = 0.83). No significant differences between biological sexes for the number of teeth affected were found (U-test 80, p = 0.87). Analysing by type of teeth (deciduous/permanent dentition) a high prevalence of caries was found in all dentitions. The prevalence for permanent teeth is 22.1% for incisors, 28.3% in canines, and 24.5% in premolars, while first, second, and third molars showed 32.7%; 27.3%, and 54.3% respectively.

Analysis of the total number of lesions for permanent teeth (n = 130) by position in the tooth, coronal lesions are the most prevalent in the anterior dentition as well as in premolars and first molars (M1); while occlusal lesions are the most common in the second molars and the third molars. In the anterior dentition after coronal lesions (62%), cementoenamel junction caries reach 38%. Prevalence on premolars and first molars is distributed among coronal caries (46%, 45%), lesions in the cementoenamel junction (21%, 35%), root caries (18%, 5%), and occlusal lesions (1%, 5%). The main difference between M1 and premolars is that in the former 10% of the lesions correspond to total destruction of the crown (severe caries), where the origin of the lesion could not be determined. In the second and third molars, after lesions in the occlusal area (48%, 41%), the most common are the lesions in the cementoenamel line (38%, 22%) and coronal area (10%, 30%); in both cases, we found that 5% were severe caries with indeterminate origin.

Table 6 Oral pathology prevalence in deciduous and permanent teeth in the Pontevedra (together Santa Maria and San Bartolomeu) sample for carious lesions (CL), antemortem tooth loss (AMTL), dental calculus, and hypercementosis

Antemortem tooth loss (AMTL), periapical lesions (PL) and periodontal disease (PD)

Since presence of alveolar bone from upper or lower maxilla is necessary to evaluate AMTL and periapical lesions, only 21 individuals could be studied (see Table 6). From them, a total of 14 (67%) individuals showed AMTL: 62% adults (n = 13; Santa María, n = 11/19; San Bartolomeu, n = 2/2; no significant differences between sub-samples) and 9% non-adults (n = 1) (Table 6). Within individuals with AMTL, 52% showed also at least one carious lesion (48% adults, 4% non-adults). By number of preserved alveoli, 13% were reabsorbed (alveoli completely closed). When age groups are taken into account, 0% of the Infant/Child alveoli (0/25), 4% of Juvenile non-adult alveoli (3/85) and 16% of adult alveoli were reabsorbed; of them 9% (20/225) belonged to females and 22% (30/139) to males. No significant differences could be found regarding sex (U-test 84, p = 0.7), although males tend to be more affected, despite their age distribution being narrower than females. Significant differences exist between non-adults/adults (U-test 55, p = 0.003) as older individuals are more affected. Although no differences can be determined between adult age groups (U-test 52.5, p = 0.12), significant differences are shown between < 12 years and adult groups (U-test 3, p = 0.006) as well as juvenile and adult groups (U-test 2.5, p = 0.011). Regarding periapical lesions 23% (5/21; Santa María, n = 4/19; San Bartolomeu, n = 1/2) of the individuals showed at least one. Results are displayed in Table 6. Among individuals in which PD was recognisable (n = 25) significant differences were observed between non-adults and adults (U-test 125, p = 0.0001) with regard to severity as older individuals presented more intense PD. Non-adults did not show significant differences among age groups (U-test 1.5, p = 0.19) nor did the adult groups display differences between them (Kruskal–Wallis chi-squared = 3.0, df = 1, p = 0.08).

Dental calculus

Individuals with dental calculus on permanent dentition constitutes 72% (24/33; Santa María, n = 20/24; San Bartolomeu, n = 4/9; no significant differences between sub-samples) of the sample. For those individuals who presented deciduous dentition, 40% (2/8) of the sample exhibited dental calculus. In Table 6, data about prevalence per tooth are displayed. Among the total sample 27% (112/413) of permanent teeth showed dental calculus. For the non-adult sample Infant/Child (1–12 years old), infants and children, 1% (1/82) of the permanent teeth and 11% (5/46) of the deciduous teeth had dental calculus (see Table 6). For non-adults > 12 years old, juveniles, dental calculus was observed in 50% (62/125) (see Table 6), and 33% (64/195) of adults. There is a significant difference (U-test 170, p = 0.05) between adults and non-adults, indicating that adults had more calculus. Infants have significantly less calculus than juveniles (U-test 0, p = 0.004), young adults (U-test 3, p = 0.001) and middle adults (U-test 3, p = 0.009). Differences between sexes were obtained (U-test 95.5, p = 0.04), with female individuals more affected than males. Since the adult sample includes more middle age females than males (n = 5; females = 4; males = 1) (Table 1), we repeated the analyses only considering young adults (n = 13; females = 6; males = 7), and found no significant differences in this age category (U-test 98, p = 1).

Dental wear and hypercementosis

Dental wear results for posterior permanent dentition are described in Table 7, as Table 8 lists anterior permanent dentition and premolar results. The San Bartolomeu and Santa Maria sub-samples were analysed together since no significant differences in their values were detected. The majority of teeth display dental wear of Grade 2–4 and Grade 5–7, including anterior permanent dentition. As expected, there is a relationship between dental wear and age. Infant/children have none or very little dental wear in anterior and posterior teeth (Grades 0–1), while in juveniles most of teeth show a moderate intense wear (Grade 2–4). However, we found cases where M1 dental wear is moderately severe (Grade 2–4) in non-adult individuals (20% of infants/children, and 60% of juveniles), or even highly severe (Grade 5–7/Grade 5–5 +) in some of the oldest non-adult females (20% in juveniles).

Table 7 Prevalence of hypercementosis and dental wear severity in permanent posterior teeth (M1 and M2) in the Pontevedra samples (together Santa Maria and San Bartolomeu individuals)

Table 8 Prevalence of hypercementosis and dental wear severity in permanent anterior teeth (incisors, canines) and premolars in the Pontevedra samples (together Santa Maria and San Bartolomeu individuals)

Adult individuals´ dental wear can be considered high or very high. Anterior dentition and premolars displayed Grade 5–7 dental wear in 56% (9/16) and 57% (8/14) of the individuals, respectively; and Grade 8 in 19% (3/16) of anterior teeth and 14% (2/14) of premolars. For the posterior dentition, Grade 5–5 +  + was reached in 56% (9/16) of first molars and 40% (6/15) of second molars; while Grade 6 appeared in one individual (6%, 1/16) of M1 and none of M2 (0/15). Another interesting feature related to the severe dental wear is that none of the adult teeth have dental wear of Grade 0–1. In addition, young adults present a remarkable percentage of teeth with Grade 5–5 +  + , with 50% (5) of the M1 and 33% (3) for M2. Both percentages increase in middle adults, being 60% for both M1 and M2. There is no significant difference between biological sexes in terms of dental wear for the first molar (U-test 28, p = 0.46), but there is between adults and non-adults (U-test 132, p = 0.0002).

Within the total sample, eight individuals (24%, 8/34; Santa María, n = 5/25; San Bartolomeu, n = 3/9) presented hypercementosis. Five (15%, 5/34) showed at least one case of hypercementosis in the anterior dentition. Hypercementosis in premolars and M1/M2 was also found in a 9% (3/34) and 6% (2/34) of individuals, respectively. When considering the adult cohort from Pontevedra (this condition rarely appeared in non-adults), 23% (5/22) showed hypercementosis in the anterior dentition, 9% (2/22) in premolars and 18% (4/22) in first and second molars. In summary, 36% (8/22) of the analysed adult individuals showed at least one case of hypercementosis in the permanent dentition, presenting 23% of those more than one affected tooth. Within the group of affected individuals 63% belong to the young adult age group (25% female n = 2: 38% male n = 3), 25% to Middle adult (100% female) and the only Old adult makes up 13%. Among the eight individuals that present hypercementosis, 50% of them (n = 4) have this lesion in several teeth.

Discussion

Fish vs millet

The diet reconstructed by the analyses of the Pontevedra collections was interpreted previously as rich in marine resources with a variable input of C₄ plants (millets or later maize), but the cereal intake was C₃-dominated (López-Costas and Müldner 2019). Our analyses for collagen are consistent with previous results and we agree that the protein food fraction was dominated by fish/seafood consumption, mainly fish. The significant differences in δ¹⁵N_col between San Bartolomeu and Santa María suggest different access to high-level trophic position fish or salty fish, as was previously published (see López-Costas and Müldner 2019). The δ¹³C_col results are consistent with the consumption of C₃ and C₄ plants and are parallel to those obtained in other populations in the north of the Iberian Peninsula (see Guede et al. 2020; MacKinnon et al. 2019) where C₄ plants (i.e., millets, and less likely sorghum) had been consumed. However, it has been shown in the literature that both the analysis of δ¹³C_col and δ¹³C_car on their own does not provide a diagnostic tool for diet reconstruction, while the combined analysis of both provides models that are more representative of actual diet (Kellner and Schoeninger 2007), since δ¹³C_col and δ¹⁵N_col analyzed in extracted collagen only reflect protein intake and δ¹³C_car reflects the total diet (Ambrose and Norr 1993; Price and Hopkins 2015). The Pontevedra sample shows higher δ¹³C_col values (> -18‰) compatible with the consumption of some fish and/or C₄ plants. However, after comparison of carbon isotopes in carbonates from bioapatite, we found that although the whole population shows values compatible with a marine origin of diet, eleven individuals (Santa María = 7/20; San Bartolomé = 4/9) show elevated δ¹³C_car values (> -12‰) pointing to a substantial intake of energy from C₄ plants.

The δ¹³C_col and δ¹³C_car values were plotted against the linear regressions proposed by Kellner and Schoeninger (2007) in Fig. 5 for best visualization, which indicate energy source (vertical axis) and protein origin (horizontal axis). Note that the study by Kellner and Schoeninger (2007) is based on maize as a C₄ plant, while for medieval Pontevedra we also need to consider other C₄ plants that can cause a bias in the model proposed by these authors. After considering the results from Fig. 5, the present study refines previous results suggesting that, although fish consumption is prevalent in Medieval Pontevedra and it is even favoured in those individuals belonging to the fishing guild, which had easier access to salt fish (sardine) and high trophic level marine food (hake, conger, octopus), C₄ plants were also consumed in both parishes and may have had a greater role in the diet of the town than was expected. The next step is to refine the type of C₄ plant consumed.

Fig. 5

Carbon apatite and carbon collagen plotted against Kellner and Schoeninger (see Kellner and Schoeninger 2007) regression lines indicating main protein and energy sources of diet. Note that C4 plants for the regression lines are mainly maize, while millet regression line would be different. Created with RStudio

In the Late Middle Ages, the available C₄ plants in Iberia were millets (introduced in the Bronze Age; see López-Costas et al. 2016) and sugar cane (introduced in al-Andalus; see Inskip et al. 2019). Climatic conditions in NW Iberia did not allow the optimal growth of sugar cane, but it was cultivated in Southern Iberia. The port of Pontevedra had a prominent commercial connection with Valencia although its main area of maritime operation was the Atlantic coast (Armas Castro 1992; Priegue 1998); sugar cane may therefore have arrived by boat. Molasses, in which carbohydrates predominate, would have an impact on the bioapatite signal but not on collagen. Any individual from Fig. 5 displays this relationship between δ¹³C_col and δ¹³C_car. However, we could not find any historical reference of sugar cane import, and in case it arrived it would not have been a staple for the middle class.

The other option for C₄ plants is millets, which were widely used in Europe due to their high tolerance to climate stress as well as for being a versatile and nutritious species (Peña-Chocarro et al. 2019). There are isotopic markers of the use of this plant as human food in the NW Iberian Peninsula since Roman times (López-Costas and Müldner 2016), or even earlier if we consider carpological remains (Peña-Chocarro et al. 2019; Teira-Brión 2022). The use of C₄ plants, probably millet, is also documented isotopically in medieval Asturias (also in NW Spain); these plants were replaced by maize in the mid-1600 s (a pest-resistant plant in Europe suitable for depleted soils) (see MacKinnon et al. 2019). In the Basque Country, millet was used for human consumption and may have supplemented the shortages of wheat which are historically documented (Martínez Martínez et al. 2015). Historical records for medieval Pontevedra (Armas Castro 1992) indicate that this cereal was transported to Pontevedra from neighbouring farms to be consumed by humans, while in the rest of the Peninsula millets were used as animal fodder (Moreno-Larrazabal et al. 2015). Faunal remains analysed in Galicia do not yet show evidence of having been fed with C₄ plants (López-Costas 2012; López-Costas and Müldner 2016, 2019). Shortages of wheat during the Middle Ages were common in the northern Iberian Peninsula due to the climate and topography, which favoured the cultivation of rye, barley, and millet (Martínez Martínez et al. 2015). Millets were consumed here as bread, porridge, and soups, especially among the lower social classes, who devoted their oats and wheat harvests as fora payment (these payments are the product of long-term agrarian usufruct transfers between a person/family and an institution—the nobility or the Church—very common in medieval Galicia) (Portela Silva and Pallares Méndez 1987; Peña-Chocarro et al. 2019). ​Millets, with their climatic resilience, were also well suited to the subsistence strategies of the Galician model, which relied on smallholdings, crop rotation, and diversification as part of a non-intensive agriculture philosophy (Teira-Brión 2022). Furthermore, the middle classes had limited accessibility to wheat imported from Castilian and Andalusian granaries, in contrast to the privileged upper echelons of society (Armas Castro 1992). Therefore, millets could have formed a significant part of the diet of the population of Pontevedra.

Millets have lower δ¹³C signatures (~ -13 to -12‰) (An et al. 2015) than other C₄ plants such as maize (~ -10‰) (Tieszen and Fagre 1993), which were introduced at a later historical stage. From the seventeenth century onwards, the area relied heavily on maize for human consumption (Puig 2012), so we do not rule out the intake of maize in some later individuals. Individuals STM824, STM833, and STM883 have been stratigraphically linked to the Early Modern period and point to a slightly different human feeding pattern according to their isotopic values, with much higher δ¹³C_col and δ¹³C_car values. Their protein input seems also to have come from fish, but there is also evidence of enrichment from C₄ plants with elevated δ¹³C_col and δ¹³C_car values. In Fig. 5, they fit in the regression line for maize as indicated in Kellner and Schoeninger (2007). Indeed, we have interpreted this anomalous isotopic data as belonging of a set of chronologically later individuals who may have relied on the consumption of maize.

The isotopic data of medieval Pontevedra’s non-adults are similar to the adult pattern, but their δ¹⁵N_col tends to be higher. The enrichment in ¹⁵N_col in infants/children not older than 4 years old is normally interpreted as the result of breastfeeding (Katzenberg and Lovell 1999; Katzenberg and Pfeiffer 1995; Schurr and Powell 2005; Haydock et al. 2013), while absence of this enrichment is normally interpreted as a reflection of the non-existence or short duration of breastfeeding. However, this assumption is currently under debate, as it may also be a reflection of stunting processes rather than of nursing experiences, that is, due to stress relating to individual interruptions of new bone formation (Beaumont et al. 2018). However, since dentine collagen seems to reflect better the nursing effect (Beaumont et al. 2018) in addition to our very small sample for individuals under four years old, we would not want to infer aspects related to breastfeeding practices. The slightly higher δ¹⁵N_col of some infants could have been connected to the consumption of breast milk or physiological stress (e.g., famile disease), although they are not correlated with δ¹³C. Delta¹³C_col is less susceptible to physiological processes and has been shown to be more reliable as evidence of non-adult diets (Beaumont et al. 2018). Since we found no differences in δ¹³C_col in this sample, we may conclude that non-adults shared the general diet of adult individuals. Interestingly, the δ¹³C_col and δ¹³C_car suggest that five non-adult individuals had a moderate-high consumption of C₄ plants. The consumption of C₄ plants in the Infant/Child individuals of Santa Maria and San Bartolomeu (Santa María infant/child = 2; Santa María juveniles = 2; San Bartolomeu infant/child = 1) could be connected to porridges and similar foods based on millets. In other parts of medieval Europe that phenomenon was identified and related with weaning foods (Adamson 2004). However, adult individuals also showed values compatible with C₄ plants consumption, as well as four of those five aforementioned non-adult individuals are above the considered weaning stage (although no information about Galician weaning process is available on the literature). In summary, the non-adult group seems to have had a diet similar to that of the adults (i.e., high marine resources), with a non-negligible element of C₄ plants.

Few studies have analyzed δ¹³C_car in populations of the Iberian Peninsula (Table 9), most of them in enamel that reflect childhood diet. All indicate a diet mainly based on C₃ plant energy, although enamel δ¹³C_car from the Southern Iberian city of Écija (Inskip et al. 2019) suggests an energetic enrichment categorized as originating from sugar cane during childhood. In Évora (MacRoberts et al. 2020), also a Southern Iberian site, enamel δ¹³C_car is slightly higher and shows important intra-site variation, in a way similar to the Pontevedra adult values. Millet in the form of porridge as a baby/infancy food or during weaning could be consumed in the village. The isotopic signal of individuals buried in the Lugo Cathedral (López-Costas et al. 2021) and Santiago de Compostela Cathedral (Pérez-Ramallo 2022) -both in Medieval Galicia and belonging to the same period and geographical location- present more negative δ¹³C_car values than the Pontevedra ones (Table 9), which may be interpreted as a diet with a high consumption of C₃-based terrestrial origin protein, possible contributions of freshwater fish, and minor intake of C₄ plants. Both sites contained skeletons belonging to elite social groups, a fact that would explain the higher access to meat, eggs, and dairy products. Textual sources mention rye and millet as cereals that were consumed by the middle and lower classes from Pontevedra, while wheat was reserved for the upper classes (Armas Castro 1992). The Lugo Cathedral books of payments indicate that millet was also available for the social elites, but according to the skeletal analyses, it was not a staple for them (López-Costas et al. 2021). In contrast, our study suggests a staple consumption of millet, but minor in comparison to fish, by the middle-class people from medieval Pontevedra. Considering the textual information, millet in diet could be considered an indicator of social class, at least in Northwestern Iberia.

Table 9 δ¹³C_car average and standard deviation for the populations used in the comparisons

Oral health and diet in hyperspecialized community

When the medieval sample of Pontevedra is compared with 18 other medieval sites from the Iberian Peninsula (see distribution in Fig. 6, all using similar recording methods to this study), we found poorer dental health of the former (Table 9) despite being a relatively young-age assemblage. In terms of age, oral health decreases with age. The lack of sex differences, despite the link between pregnancies and caries (Kubehl and Temple 2020; Watson et al. 2010), could be biased by the wider age distribution of the female cohort compared to the male cohort. When the oral pathology results are compared with those from other archaeological sites (Table 10), the prevalence of dental caries in the analysed sample is slightly above the average of medieval cohorts, while the calculus prevalence in Pontevedra is average. Carbohydrate-rich diets have been linked to a high prevalence of dental caries and dental calculus in contrast with low-carbohydrate and high-protein/fat diets which have been related to lower dental caries rates but high dental calculus prevalence (Hillson 2008; Bonsall and Pickard 2015; Larsen 2015; Temple 2015; Kubehl and Temple 2020; Bertilsson et al. 2022). There is no agreement in the scholarly literature about caries and calculus rates that can be connected to intense consumption of fish. Marine diets have been associated with low prevalence of dental caries and high rates of dental calculus due to the enrichment in fluoride and silica (Larsen et al. 1999; Polet and Katzenberg 2003; Ramezani et al. 2015; Yanko et al. 2021), precisely the opposite of what we observe in Pontevedra. A high prevalence of dental caries has been also associated in the literature with seafood consumption (Delgado-Darias et al. 2006; Arnay-De-La-Rosa et al. 2010). In medieval Pontevedra, occasional consumption of shellfish may have increased the caries rate; however, its consumption may have been avoided in medieval Galicia as it was seen as a sign of poverty (Priegue 1998), which is in line with the isotopic values obtained in López-Costas and Müldner (2019) that target fish consumption (plus octopus) rather than shellfish consumption. Individuals STM824 and STM833 presented the highest number of caries episodes, which coincides with their possible consumption of maize.

Table 10 Comparison of Medieval Pontevedra and eighteen medieval populations of the Iberian Peninsula regarding the prevalence of four oral pathologies. Acronym code: PM = Pontevedra Medieval (this study). VA = Valença (Marado and Silva 2018). AMM = Mesquita de Mértola (Silva 2011). CES = Capela do Espírito Santo (Ferreira 2015). INSA = Igreja da N.Sra. Anunciada (Ferreira 2015). MS = Maqbara de Shantarîn (Rodrigues 2013). CNSV = Capela da Nossa Senhora da Vitória (Marques 2007). SMH = Santa María de Hito (Galera 1989). SPT = Sant Pere de Terrassa (Jordana Comin 2007). HCL = Poblaciones históricas de Castilla y León (López Martinez, 2000). BA = Bragança (Márquez-Grant 2008). PR = Palat del Rey (Prada Marcos 1993). LT = La Torrecilla (López-Morago 2020). SLSN = San Lorenzo y San Nicolás (López-Morago 2020). VNS = Villanueva de Soportilla (López-Morago 2020). SBB = San Baudelio de Berlanga (López-Morago 2020). MSO = Monasterio de Suso (López-Morago 2020). EC = El Castillo (López-Morago 2020). SBM = Sahl ben Malik (López-Morago 2020)

Fig. 6

Political map of the current countries of Spain and Portugal showing the distribution of the samples used as comparison for oral pathology. Pontevedra is marked with a black dot. The map follows the same system of acronyms as Table 10. Created with QGis map

Despite these other features, dental wear is the most remarkable aspect of oral pathology. Pontevedra shows severe dental wear, with high severity reached by younger age groups (both in the anterior and posterior dentition). The great severity of dental wear defines the oral pathology of Pontevedra, with intense incidence from an early age. A hypothesis is that this severe dental wear could have also been a determining factor in the impoverishment of their global oral health since it is possible that the exposure of the pulp caused by high attrition of the enamel allows the process of compensatory eruption, exposing tooth nerves to caries and making them susceptible to colonisation by the oral microbiota, accelerating processes such as periodontal disease, plaque accumulation, and demineralization of the teeth (Hillson 1996. Ogden 2007). Therefore, the above-average rate of caries, periapical lesions and periodontal disease could be a consequence of the intense dental wear. However, more studies related to the causes/consequences of hypercementosis are needed to obtain a final answer.

The hyperspecialized fish diet observed and demonstrated by stable isotope analyses in López-Costas and Müldner (2019) would have an impact on the high rates of dental wear that also conditions the whole oral health. Once systemic conditions such as Paget’s disease, Gardner’s syndrome, and acromegaly (among others) have been ruled out (Le Cabec 2013; Bonsall and Pickard 2015; Tang et al. 2015; Massé et al. 2022), the possible causes of high dental wear include the consumption of poorly processed cereals, flours with abrasive particles implemented during processing, diet based on animal protein, and on marine products (Arnay-De-La-Rosa et al. 2010). Regarding poorly processed cereals or flours with abrasive particles, cereal milling in the NW Iberian Peninsula was mainly carried out in mills of the predominant geological material in the area, which is very abrasive (biotite and two mica granitoids) (Centro Nacional de Información Geográfica Website (n.d.). Instituto Geográfico Nacional: Geology and tectonic of Galicia map). This fact has been already suggested by Novak (2015) for granite areas in medieval Ireland, but also in areas with greater dependence on fish. In this case, most populations from Galicia and the north of Portugal might have suffered from intense dental wear since they shared geology. We have not found other sites where dental wear was studied and were located in granite areas. Nevertheless, it is expected that grain sieving and grinding in late medieval times was quite efficient. In all cases, knowing the dental wear rates from other Galician sites would be useful in the future, to address the impact of abrasive flours as an aggravating factor for dental wear. High consumption of fats and protein without vegetable intake could have also had an impact on dental wear; however, the literature is not clear (Kieser et al. 2001; Eshed et al. 2006; Watson et al. 2013). Novotná (2017) proposed that high or exclusively animal protein diets generate greater dental wear, but this idea has been criticized by Jílková et al. (2019). Regardless of the inconsistencies, the people of Pontevedra had access to vegetables by trade or self-sufficiency (Armas Castro 1992), a fact that isotopic values support, so a diet based only in animal protein is not applicable here.

Eating fish and shellfish in general, and a life connected to the sea, implies several factors that can enhance severe dental wear, as explained by Littleton and Frohlich (1993). Diets rich in marine products result in high wear (Littleton and Frohlich 1993; Watson et al. 2013; Eshed et al. 2006; Mickleburgh 2016). Fish, especially when it is dried, has abrasive particles. Pontevedra was a centre of production of salted/smoked fish in medieval times (Armas Castro 1992; Mangas-Carrasco and López-Costas 2021). The salt could also increase the abrasive nature of fish. Contributory factors can be also windblown sand (Maat et al 1990 in Littleton and Frohlich 1993) in populations living by the sea, as was the case of Pontevedra. The church of Santa María and the corresponding neighbourhood are located in a sand deposit area. The sand contained in shellfish can also increase wear. The occasional consumption of shellfish cannot be ruled out in Pontevedra, although the isotopic values of the sample suggests that the consumed protein comes from fish rather than shellfish (López-Costas and Müldner 2019). This is consistent with the avoidance of shellfish noted in the literature, arising from the fact that it was associated with the stigma of poverty (Priegue 1998). However, consumption of abundant shellfish has been recorded in the area of Pontevedra since prehistory, as shell middens are normally present in Iron Age settlements by the sea (Rodríguez Martínez et al. 2011). It is certain that the inhabitants of Pontevedra, especially those belonging to fisher families, had access to abundant shellfish since the area is rich in these animals. Finally, fibre work in relation with fishing gear and other artisanal processes could also be producing wear; however, no marks compatible with extramasticatory wear or distinctive patterns of anterior wear were observed.

In summary, the connection between stable isotopes and oral pathology seems clear in Pontevedra, as it is supported by the results and several previous works that have connected marine resources consumption and dental wear. Interestingly, the Asturian population of Veranés (placed in a non granite area) that consumed fish in high quantities, also displays a significant rate of dental wear (21%) (Rascón Pérez et al. 2013).

Conclusion

Multi-isotopic analysis has been crucial in determining. the energy origin of the diet. We profile a local population, of middle social class with a common dietary pattern characterised by high consumption of marine protein and some input of C₄ plants (probably millet).

In the Pontevedra sites, a diet rich in marine resources, mainly fish, and C₄ plants (millet/corn) was connected to special features in oral pathology such as severe dental wear. Dental wear is shown to be very severe from a very early age and above the average of medieval Iberian populations. This can only be connected to a special feature in the diet, such as the intense dependence on fish that is revealed by stable isotopes and textual sources. Another factor for explaining the poor oral health is the consumption of flours with abrasive particles due to granite mills. The associated medium–high rates of caries and calculus could be explained by millet and other cereal consumption, but also due to the dental wear that makes dental pulp available for bacteria. The hypothesis that there is a relationship between oral pathology and stable isotopes that can be determined in hyper-specialised diets is confirmed in the Pontevedra case. We encourage other researchers to analyse those populations with hyper-specialised diets by both pathology and chemical-isotopic analysis.

We think that the connection between high dental wear (and also possibly hypercementosis) with high δ¹³C_col and δ¹⁵N_col, as well as medium δ¹³C_car and Δ¹³C_car-col, which has been traditionally interpreted as consumption of marine resources, can be applicable to other types of diets and skeletal collections. It is worth the effort to discuss in the future the consequences in oral health when the consumption of fish is a staple but not so high. However, and despite other factors such as the age of the population, fish and seafood can have an impact on attrition, even if it is minor. Another conclusion of our work is the usefulness of combining isotopic signals rather than exclusively discussing the over-studied δ¹³C_col and δ¹⁵N_col, which can provide a preliminary approach to diet in order to later refine the results by taking into account other isotopic values.

A final conclusion of our work relates to the use of dental wear to estimate age. Traditionally, tooth wear has been used to determine the age of individuals on the premise that as enamel does not remodel, greater severity will be evident in direct proportion to age (Smith 1984; Brothwell 1981; Miles 2001; Burnett 2016). Although researchers are aware that factors such as hygiene, extra-masticatory uses of the mouth, or consumption of abrasive foods can directly influence the severity of dental wear (Watson and Schmidt 2020; Molnar 1971 in Faillace et al. 2017), it is used when no other proxy for age estimation is available on the premise that “it is better than nothing”. However, we encourage our colleagues to be especially cautious when determining age with dental wear in populations with a diet rich in marine products since there is a risk of overestimating the age of individuals.