Introduction

Hydrogen (H), oxygen (O), carbon (C), nitrogen (N) and sulphur (S) occur in nature as mixtures of stable isotopes with variable isotope ratios. Isotopic signatures of the local environment are transported into plants and animals via soil, water and air, and are passed along the food chain. Multi-element isotope analysis is used to determine the origin and authenticity of dairy products, wine and beef [1–6] and to implicate human palaeodiets and residence mobility [7].

Isotope ratios of light elements in human tissue provide information about nutrition (δ13C, δ15N, δ34S), and the geographic and climatic conditions (δ2H, δ18O) of the habitat [8–11].

Over the past years the Institute of Forensic Medicine in Munich, the Department of Geo- and Environmental Science, the Bavarian State Collection in Munich and Isolab GmbH, Schweitenkirchen, Germany, have cooperated in investigation of tissues of recent human bodies to assign the geographical origin of persons by determining stable isotope ratios of light and heavy elements, and not only to the country of birth, such as DNA investigations could do [12–17], but also to other countries where a person has lived during their lifetime.

Origin assignment of unidentified bodies using stable isotopes is based on the comparison of analytical data with reference samples from humans of known provenance. Therefore a reference database of human tissues and environmental samples was established at the Institute of Forensic Medicine, Munich, in cooperation with and through technical and financial support of the Munich Police Department.

The aim of this paper is to demonstrate the potential of provenance determination of humans by investigating hair samples for isotope ratios of light elements (H, C, N, S).

Materials and methods

Scalp hair samples from 111 residents (people permanently living in the relevant region during the growth of their hair) of 13 different countries from around the world were collected and analysed. From each hair sample 50 mg was homogenised by cutting into small fragments and incubated overnight with petrol ether. For determination of δ13C, δ15N and δ34S, 2 mg dry mass was placed into tin capsules, for determination of δ2H, 3 mg dry mass of hair sample.

Determination of stable isotope ratios of H, C, N, and S followed internationally accepted methods described in detail elsewhere [2, 18]. Measurement of the light elements was carried out using an elemental analyzer isotope ratio mass spectrometer (EA-IRMS). All samples were run in triplicate. Results of H-, C-, N- and S-isotopic ratios are referred to international standards: V-SMOW, V-PDB, air-nitrogen (ATM), and CDT. The analytical uncertainties (precision and accuracy) were: δ2H ±3‰/V-SMOW, δ13C ±0.1‰/V-PDB, δ15N ±0.2‰/ATM and δ34S ±0.3‰/CDT.

Stable isotopes occur with relatively mean abundance in nature. The isotope content of samples is described as an isotope ratio, the relationship of the heavy to the light isotopes from one element. For instance, C has two stable isotopes, 12C and 13C. In practice the relationship is given as δ13C value, which is calculated from the difference of the isotopic ratio (R) of the sample and an international reference material, divided by the ratio of the reference material and multiplied by 1,000 according to the formula:

Isotope ratios of light elements (δ values) in human hair samples from different origins were statistically analysed and the data illustrated as box-plots (Figs. 1, 2, 3 and 4).

The number of samples from each country is described in Table 1.

Table 1 Countries of origin for scalp hair samples
Full size table

Additionally, data were evaluated by multivariate statistical analysis and discriminant analysis was performed by SPSS program 13.0 for Windows.

Results

The δ2H values show a considerable overlap with the exception of the Australian data (Fig. 1). Hair samples from Brazil and Costa Rica presented the highest δ13C values and the mean values were approximately 4‰ higher compared to the other countries (Fig. 2), where the δ13C values showed overlap. More discrimination between the countries is demonstrated by the δ15N values (Fig. 3). For instance, Russia can be differentiated from Pakistan or southern Germany, or Lithuania from Iran. δ34S values of the origins however showed many overlaps, except for Australia (Fig. 4).

Fig. 1
figure 1

δ2H-values of hair samples from different geographical origins, illustrated as box plots with minimum, maximum and mean values. An outlier exists for Denmark

Full size image
Fig. 2
figure 2

δ13C-values of hair samples from different geographical origins, illustrated as box plots with minimum, maximum and mean values

Full size image
Fig. 3
figure 3

δ15N-values of hair samples from different geographical origins, illustrated as box plots with minimum, maximum and mean values. An outlier exists for Pakistan and Brazil

Full size image
Fig. 4
figure 4

δ34S-values of hair samples from different geographical origins, illustrated as box plots with minimum, maximum and mean values

Full size image

Multivariate analysis of the isotope data was applied to allocate the hair samples from different countries of origin into the correct group using isotopic parameters. For this, the four variables (δ2H, δ13C, δ15N, δ34S) were evaluated by discriminant analysis. The structure matrix shows that δ13C and δ34S have the highest absolute correlation coefficient of the four elements, thus accounting for the discrepancy between the countries. Figure 5 describes the assignment of the light elements relating to the countries. For instance Australia, Brazil and Costa Rica are separated completely from the other groups.

Fig. 5
figure 5

Isotope ratio data of the light elements (δ2H, δ13C, δ15N and δ34S) of hair samples, evaluated by discriminant analysis. Function 1 and Function 2 describe the individual discriminant scores for the two discriminant functions. Group means are centroids, demonstrated as black squares (centre of the group)

Full size image

To qualify the criterion of the evaluated discriminant analysis, the hair samples were reassigned to groups via the determined discriminance function. Nevertheless, about 91% of the originally grouped hair samples were correctly assigned to the country of origin by consideration of the numbers of the group samples.

Discussion

The H isotopic signatures of human tissues are mainly determined by the isotopic ratio of drinking water and by water and organic hydrogen in the diet. The δ2H values of the single countries correlate with different climate zones and altitude. For instance very low mean δ2H values of hair samples from southern Germany, Russia, Poland and the Czech Republic may be due to a common moderate continental climate, whereas the higher mean δ2H value for Denmark reflects different conditions due to the place of origin near the sea (hair donors are from Copenhagen and Vejle). In contrast, Mediterranean and tropical climates show rather high δ2H values (e.g. Brazil, Sao Paulo, Greece/Cyprus; www.naweb.iaea.org). If the climate is dry and hot, but the δ2H value is relatively low, then the altitude of the region of origin could be high above sea level (Teheran, Costa Rica, San Josè; www.waterisotopes.org).

δ13C values reflect the composition of the food and indicate whether an individual eats C4 and/or C3 plants. European nutrition is based primarily on C3 crops (range −24‰ to −32‰), for example potatoes, rice and cereals [19]. High proportions of C4 plants (range −10‰ to −16‰) like maize and/or cane products prevail, for instance, in Costa Rica and Brazil.

δ15N values of the investigated hair samples differed, depending on the origin of the samples and known trophic fractionation (∼+3‰ from a trophic level to the next higher level) [8]. The higher the δ15N values of hair samples are, the more animal products were consumed and conversely, low δ15N values are due to a diet containing less animal products, consumption of legumes or poultry fed with Soya protein (Brazil, Costa Rica, Pakistan, China, Iran). δ15N values of hair samples from southern Germany varied widely, because the relatively high number of samples may reflect various individual nutritional habits (eg. inclusion of vegetarians). The variety of values from the European countries is lower, suggesting similar nutritional habits with regard to protein (nitrogen) sources.

Together with high δ15N values, high δ34S values are the result of fish and seafood consumption. Particularly high δ34S values have also been detected in meat and milk samples from the same area, which may be due to sea spray and special geological conditions [20, 21]. High δ34S values have also been observed both for Brazilian fruit juices and Brazilian hair samples [22]. Low δ34S values in hair samples are basically unspecific, but may be related, for instance, to the influence of volcanic sulphur (Costa Rica; private communication from S. Rummel, Bavarian State Collection of Paleontology and Geology, Munich).

It is usually impossible to discriminate between different countries of origin when applying only stable isotope data from one single element. Therefore it is necessary to take into account the information of four elements using multivariate analysis. Based on their structure matrix, δ13C and δ34S add the highest contribution for origin discrimination of hair samples.

Conclusion

As only of a limited number of samples have been investigated from several countries, and the highest number of samples originates from southern Germany, the present results can be used only as a feasibility test. The large number of hair samples from Germany probably represents a wide spectrum of nutritional habits and different geographical conditions. Since the number of collected and analysed hair samples from other countries of origin is very small, they probably represent only a portion of individual nutritional habits and in a very small part of the country.

We expect that when more regions of a country are included in the reference data base for stable isotope values, the better the possibility to assign a sample to its origin. In addition it enables a better understanding of the main factors that influence multi-element stable isotope patterns of human hair to be obtained. Therefore, our current results should be regarded as a first indication of the potential of this method.