Ethics statement

This research was approved by the Institutional Review Board (IRB) of the University of Utah under protocol number 10249.

Sample collection and processing

Human scalp hair was collected as trash from the floors of barbershops and donated by anonymous volunteers in thirteen Western European (WE) countries (Figure 1; Table S1). Although we use the term ‘Western Europe’ to group all thirteen countries sampled in our study, we recognize that some (e.g., Greece, Italy and Malta) are not always classified as Western Europe but rather as Southern Europe. For the United States of America (USA) we used the same dataset of hair samples described by Valenzuela et al. [33], with the addition of 28 new samples. These 28 new samples did not represent new collection sites and thus the sampling locations are the same as those presented in Figure 1 in Valenzuela et al. [33]. All hair samples were placed in paper envelopes at the time of collection. No information was recorded regarding the age, gender, diet, and health or travel history of the donors. We assumed that the hair samples represented individuals local to the collection site. Prior to analysis, hair samples consisting of 20–40 strands of hair were washed twice in a 2∶1 chloroform:methanol mixture at room temperature to remove lipids and other surface contaminants. In the case of dyed hair, the washes were repeated until the solvent mixture was clear and no additional color was leached from the hair. The volume of solvent mixture used in each wash was sufficient to completely submerge all hair strands. The solvent mixture was gently agitated during the washes. After the washes were completed the samples were placed in paper filters and left to dry inside a fume hood. Once the samples were dried they were ground to a fine powder using a ball mill (Retsch; Haan, Germany) and placed in capped 1-dram glass vials for storage until analysis. For δ¹³C and δ¹⁵N analysis, 500 µg (±10%) of ground material was loaded into tin capsules (3.5×5 mm, Costech Analytical; Valencia, CA, USA); for δ³⁴S analysis, 900 µg (±10%) was loaded into tin capsules.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Distribution of sample locations in Western Europe.

Countries where samples were collected are highlighted in grey; black circles represent cities where sampled were obtained. For the names of the cities see Table S1. For the distribution of sample locations in the USA see [33].

https://doi.org/10.1371/journal.pone.0034234.g001

Stable isotope analyses

Hair samples were analyzed using an isotope ratio mass spectrometer (IRMS) operated in continuous flow mode (Finnigan-MAT Delta S; Bremen, Germany). Analyses of hair samples for combined carbon and nitrogen as well as for sulfur isotopes were conducted using different instruments, but the analytical procedures were similar. Tin capsules were loaded into a zero-blank autosampler (Costech Analytical, Valencia, CA, USA) interfaced with an elemental analyzer (Carlo Erba, Milan, Italy) where they were flash combusted to produce CO₂, N₂, and SO₂ for carbon, nitrogen, and sulfur isotope analysis, respectively. To minimize the need for oxygen isotope corrections of the SO₂ gas we followed the method developed by Fry et al. [42]. The resulting gases were chromatographically purified and carried to the IRMS via He stream. For δ¹³C and δ¹⁵N analysis a 3-m Poraplot-Q packed gas chromatography column (80 mL min⁻¹, 40°C) was used, and for δ³⁴S analysis a 0.8-m Hayes Sep packed gas chromatography column (80 mL min⁻¹, 90°C) was used. Hair samples were analyzed alongside a set of internal laboratory reference materials. For carbon and nitrogen analyses, the isotope ratios of our laboratory reference material, Powdered Queratin (Voigt Global Distribution Inc., Lawrence, KS, USA), had values of −24.04‰ for δ¹³C, +5.93‰ for δ¹⁵N when calibrated against the international standards USGS40 (−26.389‰ for δ¹³C, −4.5‰ for δ¹⁵N) and USGS41 (+37.626‰ for δ¹³C, +47.6‰ for δ¹⁵N). For sulfur analyses, the isotope ratios of our laboratory reference materials, silver sulfide (Salt Lake Metals, Salt Lake City, UT, USA), zinc sulfide (Mass Spectrometry Laboratory, Institute of Physics, Marie Curie-Sklodowska University, Poland) and powdered feathers (International Down and Feather Testing Laboratory, Salt Lake City, UT, USA), had values of +17.94, −31.94 and +16.9‰, respectively, when calibrated against the international standards IAEA-S-1, 2 and 3 (−0.30, +22.7 and −32.3‰, respectively). Results for δ¹³C values are presented on the Vienna Pee Dee Belemnite (VPDB) scale; for δ¹⁵N values, on the atmospheric AIR scale; and for δ³⁴S values, on the Vienna Canyon Diablo Troilite (VCDT) scale. The analytical precision (1σ) based on long term measurements of internal laboratory reference materials for δ¹³C, δ¹⁵N and δ³⁴S values was 0.1‰, 0.2‰ and 0.4‰, respectively. Stable isotope ratios are reported following new guidelines by Coplen [43] and using the standard δ-notation relative to an international standard in units per thousand (‰) as follows: δ = (R_sample/R_standard − 1), where R_sample and R_standard are the molar ratios of the heavy to light isotopes (e.g., ¹³C/¹²C) of the sample and standard, respectively.

Statistical analyses

Data are presented as means ±1 standard deviation unless stated otherwise. Normality and homoscedasticity were tested with Shapiro-Wilk and Bartlett's tests, respectively. Unpaired t-tests were used to compare stable isotope values of samples from Europe and from the USA. Single-classification Analyses of Variance (ANOVA) was used to compare isotope ratios among European countries. Due to the covariation of latitude with longitude in the European dataset (Figure 1), we used partial correlation analysis to assess the correlation of isotope ratios with latitude, or longitude, while controlling for the effect of longitude, or latitude respectively; for example, the correlation coefficient of δ¹³C values and latitude controlling for longitude is expressed as r_CLat.Long. Data analysis and statistical calculations were conducted using R [44]. Statistical significance was set at α = 0.05.

Isotope variability of human hair from industrialized nations reflects dietary heterogeneity

As predicted, we observed isotopic differences in human scalp hair among thirteen western industrialized countries of Europe and the USA. We have assumed that the hair samples belonged to healthy individuals without metabolic or nutritional diseases that might influence their stable isotope ratios of hair protein [45]–[48]. Thus, we propose that the isotopic heterogeneity detected in this study is a result of dietary heterogeneity among the sampled regions because the stable carbon, nitrogen and sulfur isotope ratios in hair from healthy individuals are primarily determined by the isotopic composition of the protein sources and the relative consumption of those proteins [23].

The stable carbon, nitrogen and sulfur isotope ratios presented in our study are consistent with values previously published in smaller datasets of hair and fingernail isotope ratios for a few European countries [20]–[22], [24], [35], [38]. The absolute ranges of carbon (4.5‰), nitrogen (2.5‰) and sulfur (3.5‰) values calculated for the European dataset were smaller than for USA hair samples (6.9‰, 3.2‰ and 11.1‰ for δ¹³C, δ¹⁵N and δ³⁴S values, respectively; Figure 2). However, the USA dataset contains one particular sample that can be considered an outlier (Figure 2; a single sample had the lowest δ¹³C and δ¹⁵N values, and the highest δ³⁴S value). That outlier was from a sample obtained in Chicago, IL and it is possible that this individual could have been a visitor. When this outlier was removed the ranges for carbon (4.3‰) and nitrogen (2.2‰) isotopes in the USA dataset were similar to the European dataset. This suggested that the carbon and nitrogen isotopic ranges of consumed protein were likely similar in the two regions. The range in δ³⁴S values in the USA remained approximately twice (6.5‰) that of the European samples when the outlier was removed, although the sample size for Western Europe was substantially smaller than for the USA (n = 26 vs. 228).

Overall, most samples from Western European countries had lower δ¹³C values than samples from the USA. The overlap in the distributions (between −19.3 and −18.3‰; Figure 2A) was represented by only eight samples from southwestern Europe (Blois, Bordeaux, Laleu, and Mirambeau from France; Cangas de Onis, Hostalric, and Valdepenas from Spain; and Almada from Portugal). This segregation in carbon isotope values of modern human hair between Western Europe and the USA is consistent with limited datasets presented in previously published work [20], [24], [35], [36]. Here we have shown with a larger dataset (total n = 360) that the isotopic separation remained significant, including observations from thirteen different European countries and 25 states across the USA. This stable isotope ratio difference has been attributed to the consumption of animal protein raised primarily with corn silage in the USA and with C₃ silage and pasture in Europe [24], [35]. The widespread use of corn fodder in the USA contrasted with a much lower use of this crop in the European farms is evident in the carbon isotope ratio of meats from these regions published to date [5], [49]–[54].

Within Europe, a negative correlation between hair δ¹³C values and latitude exists and the trend agrees with our prediction. That is, the pattern is consistent with the regional decreases in the proportion of C₄ grasses from the tropics towards higher latitudes [39]. If consumed animal proteins were of local origin and grown on local food stocks, then we would expect that the carbon isotope ratios of the consumer eating these proteins would also follow this latitudinal trend. Accordingly, this suggests consumption of locally or regionally produced foodstuff throughout Western Europe. A similar pattern was observed for the beef used in the Big Mac® worldwide by Martinelli et al. [5] where higher δ¹³C values were associated with lower latitudes and lower δ¹³C values towards higher latitudes. In addition, the higher δ¹³C values observed in the European south (particularly in the Iberian Peninsula) could be influenced by higher consumption of marine resources with high δ¹³C values in these countries. This would also explain the weak but significant correlation of δ¹³C values and longitude observed in Western European hair. However, this trend seems to be primarily driven by the low δ¹³C values of Finnish samples (Figure 3B).

The stable nitrogen isotope distributions of Western Europe and USA hair samples are characterized by extensive overlap; all samples from the USA (except the outlier at 6.7‰) were contained within the European range. The similarities in these continental distributions and the lack of geographic trends within Western European countries suggested 1) that residents of these countries consume similar amounts of animal protein, including dietary styles from vegans with no animal protein intake to omnivores with high animal protein intake [21], [22], and 2) that the baseline δ¹⁵N value of crops and animal feed may be similar [54]–[58]. The small number of higher δ¹⁵N values detected in Western Europe did not originate from a single geographic region; rather, they came from six different countries (Heinola, Finland; Mirambeau, France; Aosta, Italy; Dingli, Malta; Almada, Portugal; Medina del Campo, Cangas de Onis, Hostalric, Bilbao, and Toledo, Spain; and Sheffield, England). These samples might represent individuals with diets with unusually high animal protein consumption, consumption of elevated amounts of marine fish, or consumption of foodstuff derived from organic farming where manure is used as fertilizer [59].

The difference in the δ³⁴S values between hair samples from Western Europe and the USA was interesting. Previous work with modern European hair samples is limited to a few samples from England [35]; accordingly, we had no prior expectations. The factors affecting δ³⁴S values in humans are dependent upon the source of proteins being consumed. Marine proteins (e.g., fish, shellfish) have on average higher δ³⁴S values than terrestrial proteins [60]. Within terrestrial agro-ecosystems, geographic location and farming practices may further influence the δ³⁴S values of foods produced. Plants, and the animals consuming them, grown near the coastline tend to have high δ³⁴S values due to the deposition and assimilation of sulfur molecules carried from the ocean by the wind (sea spray effect; [30]). However, a recent study showed that the influence of marine sulfur inputs on terrestrial animals was very localized (<100 km from the coastline; [34]). Furthermore, the soils where plants grow may have different δ³⁴S values according to geologic origin [30]. Similarly, the type of fertilizers used will impact the isotope values of the crops. Large ranges in δ³⁴S values (up to 30‰) of fertilizers exist due to their varied geochemical origins and manufacturing methods [59]. Given that all these processes could be acting at different geographic scales alone or in conjunction, it is very difficult to ascribe a single factor or combination of factors that would explain the isotopic difference in δ³⁴S values of hair from Western Europe and the USA. Although the sample sizes for δ³⁴S values of European countries are too small for rigorous interpretation in such a large geographic area, the differences among countries and the potential longitudinal trend were consistent with two hypotheses: 1) higher consumption of marine resources in Spain and France than in Germany and Italy [41], and 2) higher marine sulfur deposition on the western European coasts [34]. These differences are well reflected in the δ¹³C and δ³⁴S values for the two areas presented in Figures 2A and 2C.

Hair from Western European citizens reflects the existing regional dietary heterogeneity

Western Europeans and Americans consume large amounts of protein, in most cases in excess of the average nutritional requirements [41], [61]–[63]. However, agrofood industry practices and individual dietary choices based on cultural differences and food availability can have dramatic effects on the type of protein that is consumed in these regions, and therefore on hair isotope ratios, as observed in our study. The two main differences between Western Europe and the USA are 1) the pervasive use of corn in the American diet [64], and 2) the higher proportion of marine resources in the European diet [41], [65].

Dietary heterogeneity within Western Europe has been previously assessed through the use of dietary surveys such as food frequency questionnaires and 24-hour recall as well as the use of data from the Food and Agriculture Organization of the United Nations Statistical Division (FAOSTAT) [7]–[13], [40], [41], [63], [66]–[68]. The European Prospective Investigation into Cancer and Nutrition (EPIC) project in particular has provided great insights into the nutrition and dietary patterns of Western Europe in recent years [69] (http://epic.iarc.fr). The EPIC studies describe large levels of heterogeneity in foodstuff, beverages and overall dietary patterns between and within countries [8], [40], [63], [66], [68]. Generally, eastern Mediterranean countries such as Italy and Greece show the lowest numbers of food groups consumed, while France and Spain show the most heterogeneous diet [8], [66]. Western European citizens are characterized as high animal-protein consumers, with animal sources contributing more than 50% of total protein intake (in some cases more than 70%) [63]. Although red meat is the most prominent animal protein consumed across Europe, the relative contributions from other meats differ among countries [63]. For example, poultry is an important source of protein in England (∼20% of mean animal protein) but a very small one in Nordic countries (< than 5%) [8], [63]. Consumption of processed meats (e.g., sausages) is relatively high in Germany (25–30%) but very low in Greece [8], [63]. Spain, Greece and the Nordic countries are characterized by relatively high (∼20%) consumption of fish and other marine resources, while Germany and the Netherlands report very low relative intakes [8], [40], [41], [63]. In the case of Finland (sampled in our study but not part of the EPIC project), there is evidence that freshwater fish rather than marine fish represent a very large proportion of the everyday diet [65], [70].

The proportional contribution of different plant proteins to the average Western European diet also presents marked geographic differences [8], [66]. Although most plant protein sources consumed by Europeans are C₃ plants (potatoes, vegetables, fruits and most cereals and legumes), the geographic differences in their relative intake are depicted in the heterogeneity of European dietary patterns [8], [66]. As a proportion of total plant protein, cereal (the overall highest contributor to plant protein) intake is higher in Mediterranean countries and in Scandinavian countries, while Spain and Germany report lower intake [63], [66]. Furthermore, there are clear south-north increasing trends in the relative consumption of legumes, fruits and vegetables [63], [66].

The dietary heterogeneity discussed above for Western Europe represents a complex set of regional and cultural differences affecting the consumption patterns of local residents. Thus, it is difficult to compare directly these dietary patterns and the isotopic data. We present two reasons for this difficulty: 1) The number of samples in our study is small in comparison to the thousands of individuals interviewed in efforts such as the EPIC project [69]; 2) The stable isotopic compositions of a consumer's tissues reflect the proportional contribution of all dietary sources to that tissue and different combinations of sources could produce similar isotope results. However, it is possible to observe some dietary trends reflected in both dietary patterns and isotopic data. The measured isotopic compositions of hair tissues from Western Europe tend to reflect dietary differences in cases where a “food group” is overwhelmingly represented in some regions and is characterized by very low intake in other regions. For example, residents of the Iberian Peninsula (and some regions of France) typically have a higher consumption of animal protein, particularly a much higher intake of seafood and marine fish, than other European countries, which coincides well with higher δ¹³C and δ³⁴S values (and slightly higher – but not statistically different – δ¹⁵N values) observed for hair samples collected in these regions. It is also interesting to note that some of the lowest δ¹³C and δ³⁴S values and some of the highest δ¹⁵N values were observed for Finnish samples, which coincides well with the idea that freshwater fish represents a large proportion of the protein consumed in Finland.

Even though globalization processes are acting at every level of the modern human food web separating people from their local resources (e.g., homogenization and industrialization of farming practices, large-scale distribution of food products and raw materials), here we have shown that regional dietary signatures are reflected in the tissues of consumers. Despite homogenization trends already detected in Europe [7], [11]–[14], isotopic data of modern human hair from Western Europe support the existence of dietary heterogeneity, which has characterized European observations for many years [7], [8], [69]. While isotope data cannot fully resolve whether these patterns reflect consumers eating different locally produced foods or eating the same food items produced with local resources (glocal food), they capture a geographically structured heterogeneous diet among the sampled countries.

Here we showed the potential that isotope analyses have for large-scale, rapid and efficient dietary screening. Stable isotope analyses will certainly not replace the level of detail generated with well-designed dietary questionnaires; rather they will complement those surveys and provide additional information on assimilated macronutrients. Stable isotope analyses could even provide independent verification of data obtained by self-reporting instruments. Furthermore, stable isotope analysis of hair could be accompanied with measurements of carbon isotopes in breath CO₂ (as biomarker of sweetener consumption) to provide a broader view of dietary patterns across large regions [51], [71].