Cadmium reaches chocolate through the roots of the cacao tree. These studies follow it from soil to bar, and show a problem of particular soils and particular trees more than of countries.
19 studies2003 to 2023Each checked against its published abstract
What these studies add up to
Bean cadmium follows the soil. Surveys in four countries agree that it rises with the soil's total cadmium and with acidity, and the soils concerned were mostly in the range of unpolluted soils.
It varies sharply inside each country: twentyfold between farms in Costa Rica, and concentrated in particular departments in Peru. No survey supports a statement about a whole origin.
Studies disagree about the source. Two surveys read the pattern as natural, from geology and young soils; one read a surface concentration as a sign of human activity.
The tree matters as much as the soil. A hundred kinds of cacao on the same soil differed thirteenfold in bean cadmium, and the difference appears in the bean and not in the wood or leaves.
Mitigation in the field is unproven. In the one multi-year field trial read, lime changed the leaves on an acid soil and changed nothing in the beans within two and a half years.
Processing lowers the figure mainly by removing shell and diluting cocoa. Fermentation moved some cadmium out of the nib in two studies and none in a third.
Chocolate is a modest part of most people's cadmium intake: 7 to 9% in one Belgian study, and a fraction of it is absorbed.
In a laboratory digestion, 10 to 50% of the cadmium in cocoa powder was available
What are the concentrations and potential bioavailability of cadmium and lead in cocoa powders and related products of different origins?
How
Cocoa powders and related products (beans, liquor, butter) of different geographical origins analysed for total cadmium and lead, with fractionation by selective extraction and a sequential enzymolysis approach simulating digestion. The abstract gives no sample number.
Finding
Cadmium in cocoa powders varied from 94 to 1833 micrograms per kg, of which 10-50% was potentially bioavailable. Lead in cocoa powders was in the 11-769 micrograms per kg range and its bioavailability was generally below 10%. Virtually all the cadmium and most of the lead were found in the cocoa powder after pressing of the liquor, not in the butter.
Does not establish: It does not measure absorption in humans. A later Ecuadorian study reported a gastric bioaccessible fraction of cadmium of up to 90% in beans and liquor, so the two in vitro estimates differ.
Limits: An in vitro enzymatic simulation; sample numbers and origins are not given in the abstract.
What are the cadmium levels in soils and cacao plants in southern Ecuador, and how are they related?
How
Soil samples from 19 farms in southern Ecuador at four depth intervals down to 50 cm, with plant samples from four nearby trees; total recoverable and extractable cadmium measured.
Finding
Bean cadmium ranged from 0.02 to 3.00 mg/kg and averaged 0.94 mg/kg; 12 out of 19 sites had bean cadmium above the 0.6 mg/kg level the authors used. Bean cadmium correlated with M3- or HCl-extractable soil cadmium at the 0-5 and 5-15 cm depths (r = 0.80 and 0.82 for M3; r = 0.78 and 0.82 for HCl). Cadmium in tissues decreased in the order beans, shell, leaves. Because cadmium decreased with depth, the authors judged human activity the more likely source.
Does not establish: It does not identify which human activity, if any, added the cadmium; a later Honduran survey (Gramlich and colleagues) and the Ecuadorian national survey (Argüello and colleagues) attributed soil cadmium to natural variation and young soils instead.
Limits: Nineteen farms in one region. The inference about source rests on the depth profile alone.
Which soil and site factors govern cadmium uptake by cocoa in Honduras?
How
Survey of 55 cocoa farms in Honduras measuring cadmium in leaves, pod husks and beans and relating it to soil and site factors, including DGT-available soil cadmium.
Finding
Bean cadmium averaged 1.1 mg/kg and leaf cadmium 2.6 mg/kg, while average total soil cadmium (0.25 mg/kg) was in the range of uncontaminated soils. Bean cadmium differed strongly between geological substrates and was highest on alluvial substrates. No influence of fertiliser application or vicinity to industrial sites was found, and the authors concluded the differences were due to natural variation. DGT-available soil cadmium was the best single predictor of bean cadmium (R2 = 0.5).
Does not establish: It does not show that high bean cadmium requires contaminated soil: the soils here were in the uncontaminated range.
Limits: A one-country survey; half of the variation in bean cadmium was not explained by the best predictor.
How much do cacao genotypes differ in cadmium uptake and in how it is partitioned between leaves and beans?
How
One hundred accessions from the International Cocoa Genebank, Trinidad, representing the genetic groups and hybrid populations, measured for leaf and bean cadmium with three tree replications; bioavailable soil cadmium measured around each tree.
Finding
There was a 13-fold variation in bean cadmium and a 7-fold variation in leaf cadmium between accessions although bioavailable soil cadmium was uniform. Bean cadmium as a share of combined leaf and bean cadmium concentration ranged from 15 to 52% between accessions. Cadmium was in general higher in the testa than in the cotyledon, with considerable genetic variation. Genetic groups differed significantly, but much of the variation was between accessions.
Does not establish: It does not show that any named variety or genetic group is reliably low in cadmium, or that grafting onto a low-uptake rootstock lowers cadmium in the beans of the scion.
Limits: One site and one soil; trees were ungrafted genebank accessions, so the result does not show how a low-uptake type behaves as a rootstock or on other soils.
What cadmium and lead concentrations occur in cocoa powder and chocolate sold in the United States, and how do they relate to cocoa content and origin?
How
A convenience sample of cocoa powder, dark chocolate, milk chocolate and cocoa nib products bought at retail in the US and analysed by ICP-MS. The abstract gives no sample number.
Finding
Cadmium ranged from 0.004 to 3.15 mg/kg and lead from below the limit of detection to 0.38 mg/kg. Both were significantly correlated with percent cocoa, the correlations varying by product type and geographic origin. Cadmium was higher in products reported as originating from Latin America than from Africa.
Does not establish: It does not show that any origin's products exceed a health-based limit, and it reports no origin pattern for lead.
Limits: A convenience sample, so the ranges do not describe the market as a whole; origin was as reported on the product.
Japan: 8 of 180 chocolates and 26 of 140 cocoa powders above the EU cadmium levels
Kataoka and colleagues, 2018. Shokuhin Eiseigaku Zasshi (Journal of the Food Hygienic Society of Japan). doi:10.3358/shokueishi.59.269
Question
What cadmium concentrations occur in chocolate and cocoa powder on the Japanese market?
How
Retail survey in Japan of dark, milk and white chocolate and cocoa powder products analysed by ICP-MS: 180 chocolate products and 140 cocoa powder products.
Finding
Cadmium ranged from 0.00021 to 2.3 mg/kg in chocolate products and from 0.015 to 1.8 mg/kg in cocoa powder products. A weak positive correlation was found between cadmium and the cocoa solids content stated on the label. Compared with the EU maximum levels, eight of the 180 chocolate products and 26 of the 140 cocoa powder products exceeded them.
Does not establish: It does not show a strong link between declared cocoa percentage and cadmium: the correlation was weak, which implies origin or other factors matter as well.
Limits: The EU levels do not apply in Japan and were used only as a yardstick; the abstract does not say how products were selected.
How is cadmium in cacao beans distributed across Ecuador, and which soil and agronomic factors explain it?
How
Nationwide survey in Ecuador with paired soil and plant samples (pods and leaves) collected at 560 locations, plus a farmer questionnaire on agronomic practices; multivariate regression.
Finding
Total soil cadmium averaged 0.44 mg/kg, described as typical for young, non-polluted soils. Mean cadmium in peeled beans was 0.90 mg/kg and 45% of samples exceeded 0.60 mg/kg. Bean cadmium increased with total soil cadmium and decreased with soil pH, oxalate-extractable manganese and organic carbon (R2 = 0.65). Bean cadmium fell by a factor of 1.4 as orchard age rose from 4 to 40 years. Effects of genotype (CCN-51 versus Nacional), pruning and fertiliser were inconsistent.
Does not establish: It does not show that cadmium in Ecuadorian soils is high by world standards (the mean was typical of unpolluted soils), nor that fertiliser or industrial pollution is the source.
Limits: A cross-sectional survey: the regression describes association across sites in one country, not the effect of changing any one soil property on a given farm.
Do cacao cultivars differ in cadmium uptake and in its movement into the beans?
How
Sampling of rootstocks, scions, leaves, beans and surrounding soil from three replicate trees of each of 11 cultivars in the long-term CEDEC-JAS trial in northern Honduras.
Finding
Available soil cadmium was more closely correlated with cadmium in rootstocks (R2 = 0.56), scions (R2 = 0.59) and leaves (R2 = 0.46) than with bean cadmium (R2 = 0.26). Vegetative parts showed no significant differences between cultivars, whereas bean cadmium varied significantly among cultivars. In three cultivars examined in detail, mature beans differed in cadmium but immature beans did not.
Does not establish: It does not show that choosing a low-cadmium cultivar brings beans below any given limit on a high-cadmium soil.
Limits: One trial site with three trees per cultivar; the proposed mechanism (transfer from xylem to phloem during bean maturation) is an interpretation, not a direct measurement.
How effective are biochar and lime at reducing cadmium availability and uptake by cacao as the test moves from laboratory to field?
How
In vitro, greenhouse and field experiments with four, three and three replications respectively, testing biochar and lime at a range of multiples of the recommended application rate.
Finding
Lower rates were effective in vitro, but moving from in vitro to greenhouse to field the application rates and frequency had to be increased, because the effectiveness and longevity of the treatments were compromised by environmental factors. The authors describe the two amendments as complementary in their action. The abstract reports no reduction figures.
Does not establish: It does not establish how much, or for how long, biochar or lime lowers bean cadmium on a working farm.
Limits: The abstract gives no effect sizes, and the authors themselves call for further work on placement of amendments under field conditions.
How does cadmium uptake by cacao vary within the root zone when only the surface soil is limed?
How
Pot experiment with cacao seedlings in 22-cm-height pots with top and bottom layers of surface and subsurface soil from a cacao field, limed at the surface, fully, or not at all, with layers spiked with the stable isotope 108Cd to trace where plant cadmium came from.
Finding
70% of the root biomass was in the top layer. Plants on fully limed surface soil had 1.7 times lower leaf cadmium than unlimed treatments; the reduction was 1.2 times when only the top layer was limed. Isotope data showed that surface liming increased cadmium uptake from the unlimed bottom layer.
Does not establish: It does not show the size of any liming effect on beans in an orchard, where lime cannot be mixed through the root zone.
Limits: Seedlings in pots and leaf concentrations, not mature trees and beans.
Where does cadmium sit within the cacao fruit and bean, and does fermentation redistribute it?
How
Cacao fruits from four locations analysed by tissue; cadmium mapped within beans by laser ablation ICP-MS and its chemical form examined by X-ray absorption spectroscopy; cadmium followed during fermentations.
Finding
Cadmium concentrations decreased in the order testa, then nib, placenta and pod husk at similar levels, then mucilage. Fermentation induced an outward migration of cadmium from nib to testa, but only if fermentation was extensive enough to lower nib pH to below 5.0. The average change in dry-weight nib cadmium during fermentation was a factor 1.3 decrease.
Does not establish: It does not show that fermentation removes cadmium from the bean: the metal moves into the testa, which is removed later as shell, and the reduction in the nib was modest.
Limits: Fruits from four locations; the abstract does not give the number of fermentations.
Across the cacao belt, which climatic, soil and plant factors best predict cadmium in cacao beans?
How
A database of climatic, soil and plant data compiled systematically from the scientific literature and analysed with boosted regression trees. The abstract gives no number of studies or samples.
Finding
Total soil cadmium concentration, soil pH and leaf cadmium were the best predictors of bean cadmium. Available soil cadmium and soil organic carbon had negligible effects on bean cadmium itself, although soil pH and organic carbon reduced how strongly total soil cadmium was concentrated into the bean. The authors suggest plant-based approaches such as breeding and rootstocks are more likely to succeed than soil remediation.
Does not establish: It does not test any mitigation method. The preference for breeding over soil treatment is the authors' inference from the predictors, not a trial result, and its finding on available soil cadmium differs from the Honduran survey in which DGT-available cadmium was the best predictor.
Limits: A synthesis of heterogeneous published studies that used different extraction methods and sampling designs; the abstract does not report how many.
How does cadmium concentration change from soil through post-harvest steps to the chocolate bar?
How
Case study following cadmium on a single farm in Santander, Colombia, measured in soil, unfermented seeds, fermented and dried beans, shell, nibs and chocolate.
Finding
Post-harvest treatments did not affect bean cadmium (4.17 mg/kg on average). The shell held 6.57 mg/kg and the nibs 3.28 mg/kg, and the chocolate bar 1.60 mg/kg; shell removal and the percentage of cocoa mass used accounted for the reduction. The authors state that analysing the nib or cacao mass instead of the whole bean lowers the measured concentration by up to 40%.
Does not establish: It does not provide a general conversion factor from bean cadmium to chocolate cadmium, and here fermentation and drying made no difference, which differs from the fermentation studies that found a fall in nib cadmium.
Limits: A single farm with unusually high cadmium; the abstract does not give the recipe of the chocolate.
Where in Peru are cadmium concentrations in cacao soils and beans expected to be elevated?
How
Random forest models, national and regional, built on more than 2000 samples of cacao beans and soils from Peru, used to produce predictive maps across the area suitable for cacao.
Finding
Model projections place elevated cadmium in soils and beans largely in the northern departments of Tumbes, Piura, Amazonas and Loreto, with localised pockets in Huánuco and San Martin. Soil cadmium was by far the most important predictor of bean cadmium; geology, rainfall seasonality, soil pH and rainfall followed. The authors estimate that fewer than 20 % of cacao farming households nationally might be affected by cadmium regulations, but as many as 89 % in Piura.
Does not establish: It does not show that Peruvian cacao as a whole is high in cadmium: the modelled problem is concentrated in particular departments.
Limits: The maps are model predictions, not measurements at every farm, and the household estimates depend on the regulatory threshold applied.
Where are cadmium hotspots in Costa Rican cacao, and which soil properties and practices influence bean cadmium?
How
Survey with soil, leaf and bean samples from 150 farms in the three principal cacao-producing regions of Costa Rica.
Finding
Bean cadmium ranged from 0.12 to 3.23 mg/kg (average 0.56 mg/kg) and 22% of samples exceeded the threshold of 0.80 mg/kg the authors selected, mostly in the Huetar Caribe and Huetar Norte regions. Total soil cadmium and soil organic carbon best explained the variation (R2 = 0.62). Farm altitude and orchard age were negatively correlated with bean cadmium.
Does not establish: It does not show that a bean above 0.80 mg/kg produces chocolate above a legal limit; that depends on shell removal and the recipe.
Limits: Cross-sectional survey in one country; the threshold is one chosen by the authors for whole beans, where EU limits apply to finished products.
By which pathway is cadmium loaded into cacao nibs: directly from the roots through the xylem, or by redistribution through the phloem?
How
Field pulse-chase experiment: topsoil beneath two mature cacao trees was enriched with the stable isotope 108Cd by surface irrigation, and the isotope's abundance in plant tissues was followed for 548 days.
Finding
The tracer was taken up in the order immature leaves, mature leaves, nibs in both trees. Half of the equilibrium isotope abundance was reached in the nibs at 191 days after spiking, later than in mature leaves (151 days) and immature leaves (117 days). The authors conclude that nib cadmium likely comes from phloem redistribution from stem, branches or mature leaves and not from direct root-to-nib transport.
Does not establish: It does not show how quickly a change in soil cadmium availability, for example after liming, would appear in harvested beans, although the slow tracer arrival is consistent with a long delay.
Limits: Two trees at one site, followed for one season of growth.
Do surface-applied lime, gypsum or compost lower cadmium in established cacao orchards?
How
Field trial in two established orchards in Ecuador (soil pH 6.6 and 5.1) with agricultural limestone and gypsum each at 2.0 and 4.0 Mg per hectare per year and compost at 12.5 and 25 Mg per hectare per year, applied at the surface in two successive years; supporting soil column experiments in the laboratory.
Finding
Lime raised soil pH by one unit down to 20 cm depth. On the acid soil, lime lowered leaf cadmium, the reduction factor rising gradually to 1.5 after 30 months. Liming and gypsum had no effect on leaf cadmium in the pH-neutral soil. Compost reduced leaf cadmium there by a factor of 1.2 at 22 months, an effect absent at 30 months. Bean cadmium was unaffected by any treatment at 22 months (acid soil) or 30 months (pH-neutral soil).
Does not establish: It does not show that liming lowers cadmium in beans. It showed an effect in leaves on one acid soil and none in beans in the period observed.
Limits: Two orchards and about two and a half years of follow-up; any effect on beans may take longer than the trial ran.
Which fermentation conditions drive the fall in nib cadmium, and by what mechanism?
How
Laboratory incubation of cacao beans in media of different temperature, acetic acid and ethanol concentrations, mimicking fermentation. The abstract gives no sample numbers.
Finding
Incubation under typical fermentation conditions (45 °C and 20.0 g acetic acid per litre) reduced nib cadmium by a factor 1.3; under more extreme conditions (65 °C and 40 g acetic acid per litre) the factor was 1.6. Ethanol had no statistically significant effect. Final nib cadmium correlated with nib phytate concentration (R2 = 0.56).
Does not establish: It does not show that such reductions are achievable in commercial fermentation without damaging quality.
Limits: Incubation in a liquid medium, not a real fermentation heap or box; flavour consequences were not part of the abstract.
How much does cacao consumption contribute to dietary cadmium exposure in Belgium once bioaccessibility is taken into account?
How
Food frequency questionnaire and 24-hour recall in the Belgian population (N = 2055) combined with ICP-MS analysis of cacao-containing products (N = 349) and bioaccessibility measurements.
Finding
Average chocolate consumption was 28 g per day and chocolate contributed 7-9% of total dietary cadmium exposure, both higher than earlier Belgian estimates, which the authors attribute probably to some selection bias towards chocolate consumers. Cadmium bioaccessibility was lower than in wheat flour by a factor of 5 for cacao powder and 2 for dark chocolate.
Does not establish: It does not show that chocolate is a major source of cadmium for the average person, nor does it measure how much cadmium from chocolate is absorbed by the body.
Limits: The cohort over-represented chocolate consumers by the authors' own account, and bioaccessibility was measured in the laboratory, not as absorption in people.
We would like to set two Google Analytics cookies to count page views, and to count them in our own cookie-free measurement as well. They do not identify you and nothing is shared with advertisers. Decline and the site works exactly the same — see cookies for what each one does.