The elemental composition of chocolates is related to cacao content and origin: A multi-element fingerprinting analysis of single origin chocolates
Introduction
There are several health benefits (such as on cardiovascular health) associated with moderate consumption of cacao and cacao products mostly due to its high polyphenol content. However, cacao-derived products may also contain potentially toxic trace elements such as cadmium (Cd), nickel (Ni), and lead (Pb).
In 2014, the EU Commission set cadmium limits on chocolate enforced since January 2019. In chocolate with less than 30% cocoa the limit is 0.10 mg/kg, and in cocoa powder or chocolates over or equal to 50% cocoa the limit is 0.80 mg/kg. The Codex Alimentarius adopted similar limits, ranging from 0.80 mg/kg for chocolates between 50-70% cacao content to 0.90 mg/kg for chocolates with 70% or more.
For nickel, the European Food Safety Authority concluded that daily intake nickel level should be 2.8 ug /kg bodyweight. As well, they highlighted that cacao based products are among the main food sources with high levels of nickel such as some types of chocolate (with an average of 3.8 mg/kg) and cacao beans and derived products (with an average of 9.5 mg/kg). The European Commission recommended states monitor the nickel content of foods, and limits for nickel in chocolate may result in the future.
The EU regulation does not currently state limits for lead in chocolate, but the European Commission does have a maximum level for contaminants in food stuffs, and that fats and oils should not have more than 0.10 mg/kg. This level is also endorsed by the Codex Alimentarius for edible fats and the US FDA for candy destined for young children.
Acronyms used:
LOQ - Limit of quantification, LOQ stands for the smallest amount or the lowest concentration of a substance that is possible to be determined by means of a given analytical procedure with the established accuracy, precision, and uncertainty.
Where trace metals in chocolate come from
There are many factors which may influence the level of trace metals in chocolate. The concentrations of different trace metals in chocolate increase with increasing the cacao content, which indicates that these elements mainly originate from the cacao and not from contamination during processing or other ingredients.
The mineral content of cacao is affected by geographical origin of where the cacao was grown, and these have been reported (Bertoldi et al., 2016). Most chocolate is made using a mix of cacao beans from different origins, but single origin chocolates are becoming popular. The composition of single origin chocolates may allow to trace back the geographical origin of the cacao. This has already been proven with cacao beans using multi element finger-printing analysis followed by PCA (principal component analysis) and DA (discriminant analysis). Junior et al. (2018) used discriminant analysis on elemental composition to differentiate between organic and conventional chocolates. Bertoldi et al. (2016) used 13 chocolates as a validation set to test the efficiency of their discriminant analysis model for the classification of cacao beans by geographical origin.
The focus of this study
This study was set up to assess the elemental composition of single origin chocolates and to identify the potential for tracing cacao origin by element fingerprinting chocolates.
Materials & Methods
Sample collection and preparation
In 2018, 139 single origin chocolates were purchased online and in retail shops from 47 different manufacturers in several countries. This excluded white chocolate and chocolates with inclusions (fruits or nuts) or noticeable food colouring on them. The countries were as follows:
Belgium (n = 86)
Colombia (n = 3)
Ecuador (n = 25)
France (n = 10)
Italy (n = 3)
Madagascar (n = 6)
Mexico (n = 2)
New Zealand (n = 2)
Peru (n = 1)
Switzerland (n = 1)
The samples were divided into four regions (see Table 1) based on the cacao origin mentioned on the packaging which included:
Africa (n = 34)
Asia Pacific (n = 14)
Central America (n = 22)
South America (n = 69)
Chocolates were grated and 100mg of each sample was digested with 3.0 mL of concentrated Suprapur nitric acid (HNO3, 65% w/w; Merck, Darmstadt, Germany). The total elemental composition of macro (Ca, K, Mg, Na, P, S) and trace elements (Al, As, B, Ba, Be, Cd, Co, Cr, Cs, Cu, Fe, Ga, In, Li, Mn, Mo, Ni, Sb, Se, Sn, Sr, Pb, Ti, Tl, U, V, Zn, Zr) were determined by inductively coupled plasma mass spectrometry (ICP-MS).
Two chocolate certified reference materials were included as quality assurance. NIST 2384 baking chocolate certified for Ca, Cd, Cu, Fe, K, Mg, Mn, P, Pb, and Zn, as well as ERM-BD512 dark chocolate certified for Cd, Cu, Mn, and Ni. Due to poor recovery from the reference material, Fe was excluded from further analysis.
Results & Discussion
The elemental composition of chocolates and its relation to origin and cacao content
The samples originated from 27 countries, which were categorized into 4 regions:
South America - 50%
Africa - 24%
Central America - 16%
Asia Pacific - 10%
The fractions here do not truly represent the fractions of cacao grown globally. Africa grows nearly 73% of the worlds cacao, and both Central and South America combined grow about 17%. The samples here appear to be biased towards South America, however, in regards to fine single origin chocolate South American cacao is considered some of the highest quality in the world. Manufactures often select many cacao origins that exist within South America. The mean cacao content (% listed on bar label of the chocolate) was unaffected by origin (P > 0.05).
Single correlation analysis between element concentrations and cacao content of the chocolates were significant for all analyzed elements except for Cr and V. This suggests that cacao is the main origin of these minerals in the chocolate (not other ingredients or contamination from equipment.
Macro Elements
Those present in the largest concentrations were:
Calcium (Ca) - median 720 mg/kg
Potassium (K) - median 6437 mg/kg
Magnesium (Mg) - median 1887 mg/kg
Phosphorus (P) - median 2748 mg/kg
Sulfur (S) - median 925 mg/kg
Sodium (Na) - was below the LOQ in 84% of samples
Micro Elements
The most prevalent micro elements were:
Zinc (Zn) - median 27 mg/kg
Manganese (Mn) - median 14 mg/kg
Copper (Cu) - median 13 mg/kg
The concentrations of Al, As, Be, In, Li, Na, Pb, Sb, Se, Sn, Ti, Tl, U and Zr were below LOQ in more than 15% of samples, and therefore were excluded from statistical analysis
The average elemental concentrations for each region can be observed in Table 3. The concentrations of of Ba, Cd, Mo, and Sr were significantly different between regions, but no significant difference was observed for Ca, K, Mg, P, and S between different regions.
Presence of potentially toxic trace elements in chocolate and its relation to cacao content and origin
Arsenic
There are no regulations regarding concentrations of Arsenic in chocolate, but measured values in all samples were below the LOQ (0.08 mg/kg).
Nickel
This ranged from 0.3 to 12.5 mg/kg. Nickel concentrations were strongly correlated with cacao content (%) of the chocolate bar (Table 3), which suggests that at least some part originates from the raw material. the European Food Safety Authority reported higher Ni concentrations in cacao beans (9.5 mg/kg) compared to the final chocolate products (3.8 mg/kg).
Lead
The Pb limit of 0.10 mg/kg was exceeded in six samples originating from India (0.11 mg/kg), Trinidad (0.12 mg/kg), Peru (0.68 mg/kg), and Ecuador (0.17 and 0.36 mg/kg). Pb levels were blow LOQ levels (0.02 mg/kg) in 60% of samples, with most being lower than the LOQ.
In regards to Pb correlating with cacao origin, no significant relationship was found between the two. This indicates that lead in chocolates likely originates from the production process rather than the raw material (the cocoa beans). Abt et al. (2018) observed that there were much lower lead concentrations in the raw product (cocoa nibs) at a level of 0.003 mg/kg versus the chocolate itself (0.01 to 0.03 in chocolate or 0.11 in cocoa powder), which again suggestions that the lead in these products originated during the production process, not the raw material. It should be mentioned though that Villa et al. (2014) did in fact find a significant correlation between cacao content and lead concentration in 30 chocolates purchased in Brazil.
Cadmium
Of all the samples of chocolate, 16 exceeded the new EU regulation limit for cadmium (0.80 mg/kg) for chocolates with 50% cocoa content or more. However, far more samples were below the EU regulation limit. There was also found to be a correlation between cacao content and cadmium concentrations (Table 3). Abt et al. (2018), Lo Dico et al. (2018), and Villa et al. (2014) all found a linear relationship between cacao content of the products and Cd concentrations.
Samples originating in South America contained significantly higher concentrations of cadmium than those from Africa and Asia Pacific (Table 2) and the chocolates that exceeded the new EU limits were all from Central and South America (Fig. 1):
Haiti (1.44 mg/kg)
Peru (0.81, 1.07, 1.34 mg/kg)
Colombia (0.94, 0.97, 1.22, 1.68, 2.3, and 2.61 mg/kg)
Ecuador (0.81, 0.85, 0.91, 0.93, 1.66, and 2.00 mg/kg)
The Cd concentration was much larger in the South American sample set than compared to other regions. This supports the hypothesis of cadmium hot spots as opposed to homogenous geographical presence, where elevated cadmium uptake by the tree and consecutive accumulation in the beans can result in high cadmium concentrations in the chocolate. Bertoldi et al. (2016) found Cd concentrations in beans from South America were on average 4 times higher than those from Asia, and about 10 times higher than cacao beans from West Africa. Mounicou et al. reported high average Cd concentrations in cacao powders from Venezuela and different provinces in Ecuador (0.533 to 1.833 mg/kg) compared to Ivory Coast (0.094 mg/kg) and Ghana (0.133 mg/kg).
One recommendation when using South American cacao high in cadmium would be to blend it with cacao from different regions to comply with new Cd regulations. Single origin chocolates would be most affected by the new regulation since they can only be produced from beans of the specific country. The large variation with the regions indicates that blending can be done with various cacao from the same regions.
Classification and regression tree (CART) analysis
Since there are a large number of unknowns, variables, and complexities of the chocolate recipes, fingerprinting the final chocolate product requires robust methods. A CART analysis is a technique used to verify the origin of food products based on their elemental compositions. For example, it can be used to verify the origin of wine. CART was chosen over other other methods such as PCA and DA because it is more robust to the influences of the unknown variables in the chocolate recipe. See Figure 2.
At the top of Figure 2, the full data set of A was divided into samples by cadmium levels into two subgroups. The subgroup on the left contained data from samples with high levels of cadmium (made up mostly of South American & Central American samples) versus on the right that contained samples of lower cadmium concentrations with a mix of various origins contained within it. Cadmium was the first variable capable of discriminating South American chocolates from other origins, which is in agreement with the higher Cd levels measured in South American chocolates (Table 2).
Following The Left Node 1 (High Cadmium Group)
This group contained a mixture of mostly South American and a few Central American origin chocolates. This group was then divided by node 2 according to the concentration of Mo (Molybdenum). Mo levels could discriminate this group into South American vs Central American origin chocolate.
The Node 1 (high Cd group) was then was then split according to their concentration of Mo (Molybdenum) at Node 2. Mo was able to discriminate the high cadmium group into South American versus Central American depending on the Mo level. Samples with a higher levels of Mo were all South American (leaf b), where samples with lower levels were a mix of Central American and a few South American samples (leaf a).
Following The Right Node 1 (Low Cadmium Group)
Mo levels were also able to separate the group of samples with lower cadmium levels as well (Node 3), with higher concentrations measured in South American chocolate (0.25 mg/kg) on the right versus samples from Central America (0.10 mg/kg) and Asia Pacific (0.10 mg/kg) as well as some African and South American samples on the left.
Node 4 used Barium (Ba) to separate African samples from the rest of the origins. Node 5 used Strontium (Sr) to separate African chocolates from Central American and Asia Pacific (but still mixed with South American), and then Node 7 was used to separate this African chocolate sample set from the South American samples using Zinc (Zn).
High probabilities of the leaves (end points) were obtained for South American (0.99 in leaf b) and African (0.83 in leaf g) samples. Both South American and African samples had a low misclassification rate (13% and 10% respectfully).
Asia Pacific sample classification was unsuccessful as all Asia Pacific samples were misclassified. This is likely due to very high geographical heterogeneity and differences in soil composition and climate with the Asia Pacific sample group (a combination of islands and mainland countries). This likely resulted in larger variety of element compositions and added complications to separating them successfully. As well, the small sample size of this group may also have hindered the separation of samples. The CART method minimizes the overall error rate, which may cause it to poorly predict an underrepresented group. Other parameters may be used instead such as stable isotope ratios.
Conclusion
In this study, we observed that 12% of the 139 chocolate samples contained higher levels of cadmium which exceeded the new EU regulation limits on chocolate and cacao products. There was a strong correlation with Cd concentration and cacao content, suggesting that Cd levels likely come from the cocoa beans themselves. South American samples saw higher levels of Cd concentration, but South American samples also made up 50% of the 139 samples used, and biased towards South American chocolate, compared to chocolate with cacao originated in Africa and Asia Pacific. The CART model validated these findings, by using Cd at the first node to separate mostly South American samples from other regions.
The concentration of lead was exceeded in 6 samples originated in India, Ecuador, Peru, and Trinidad. However, the data suggests that the lead concentrations were unrelated to the cacao itself, but rather a result of post-harvest processing.
Classification and regression tree analysis seems to be a suitable technique to trace the origin of chocolates based on their elemental composition. It offered a way to classify unknown samples and information regarding the elements that most characterize chocolates from each origin. The total misclassifcation rate was 23%, most of samples from Asia Pacific due to reasons mentioned above.