As You Sow 2022 — Expert Investigation Related to Cocoa and Chocolate Products (Final Report)

This is the Final Report of a four-member Expert Committee charged under a California Proposition 65 Consent Judgment with investigating the predominant sources of cadmium (Cd) and lead (Pb) in cocoa and chocolate products and recommending feasible measures to reduce Cd and Pb levels in those products. The Committee operated from March 2018 through April 2022. The Final Report (March 28, 2022) consolidates three earlier technical reports — the Root Cause Report (submitted March 27, 2020), the Reductions Recommendations Report (submitted April 12, 2021), and the Warning Triggers Report (submitted November 15, 2021) — and adds individually authored “affirmative conclusions” from each of the four Experts. The full document is 381 pages including the three technical attachments and the alphabetized cited-references list.

Source role

For Heavy Metal Index, this document is the most consequential single piece of grey-literature evidence in the corpus on root-cause attribution and feasibility-of-reduction for Pb and Cd in cocoa-based products. Its standing derives from four facts: (1) it is the consensus product of a four-member multi-disciplinary expert committee (food-safety industry practitioner, USDA soil chemist with H-index 105, California toxicologist with Cal/OEHHA Prop 65 experience, and West Indies cocoa-Cd researcher); (2) the Committee reviewed >400 scientific publications and conducted >250 outside subject-matter expert consultations across a three-year evaluation; (3) the Committee performed its own primary studies (the Bean Cleaning and Winnowing Study at a major North American cocoa processing facility, and the Bean Abrasion Study) plus two field trips (three U.S. industrial facilities and an end-to-end Ecuador supply-chain trip including farms, collection points, and exporter facilities); and (4) the Committee had access to a large non-public sampling dataset shared by the Settling Defendants under confidentiality agreements, plus the complete As You Sow market-test dataset for nearly 600 chocolate products tested for Pb and Cd from 2014 through 2021.

The document supplies four kinds of evidence that recur across this wiki’s chocolate/cocoa work. First, it supplies the strongest available root-cause attribution for Pb in cocoa products to post-harvest fermentation, drying, and bean-handling contact with soil/dust rather than pre-harvest uptake from soil. Second, it supplies the strongest available root-cause attribution for Cd in cocoa products to pre-harvest soil uptake by the cocoa tree, with soil pH, Cd:Zn ratio, parent-rock geochemistry, and historical anthropogenic enrichment as the dominant determinants. Third, it supplies a quantitative bean-cleaning-and-winnowing mass balance showing that approximately 70% of nib Pb concentration is acquired during the bean-breaking/winnowing step at processing facilities, with mechanical cleaning reducing mean cocoa-bean Pb from 95.2 ppb to 39.7 ppb (a 58% reduction) in the Committee’s own facility study. Fourth, it supplies the feasibility-rated list of 30 candidate Cd/Pb reduction strategies (15 rated High confidence, 5 Medium, 10 Low) and individual-expert Pb and Cd drop-down trigger-level proposals for the three Consent Judgment cacao-content product groups.

Legal character and scope

The document is the consensus output of an Expert Committee constituted under a private Proposition 65 Consent Judgment between As You Sow (AYS), a California 501(c)(3) shareholder-advocacy non-profit, and a set of chocolate-industry companies referred to in the report as the Settling Defendants (SDs). The Consent Judgment followed AYS’s 2014-2017 market testing of chocolate products for Cd and Pb, which identified products where typical-serving-size exposures plausibly exceeded the Prop 65 Maximum Allowable Dose Levels (MADLs) for Pb (0.5 µg/day, reproductive endpoint) and Cd (4.1 µg/day, reproductive endpoint), leading AYS to issue Notices of Violation. The SDs entered the Consent Judgment in lieu of litigation and jointly funded the Expert Committee. The Project Manager was Eastern Research Group, Inc. (ERG); ERG wrote the executive summary and background sections of the Final Report but the Experts drafted all technical content individually.

The report is neither a peer-reviewed publication nor a binding regulatory document. It is consent-judgment-mandated expert technical work-product. Its numerical findings (analytical results from the Bean Cleaning and Winnowing Study and the Bean Abrasion Study), its root-cause conclusions, and its reduction-strategy confidence scoring are scientifically grounded and the Committee’s affirmative conclusions are individually attributable. Its policy outputs (proposed Prop 65 drop-down trigger levels for Pb and Cd by cacao-content group) are individually authored and explicitly note that the four Experts did not reach consensus on the lowest feasibly-achievable concentrations — they “agreed to disagree” on the floor.

The Expert Committee bios establish the four Experts’ credentialing for HMI evidence-tier purposes: Tim Ahn (Sr. Manager Food Safety, Lloyd’s Register; 35 years in chocolate including 27 years as Mars Global Director of Chocolate Quality & Food Safety); Rufus L. Chaney (47-year USDA-ARS Senior Research Agronomist, retired Oct. 2016; 511 published papers, H-index 105 as of March 2022; Soil Science Society of America Fellow; American Association for the Advancement of Science Fellow; USDA Presidential Rank Meritorious Senior Professional Award 2003; Agricultural Research Service Science Hall of Fame 2013); Michael J. DiBartolomeis (PhD, DABT; 38 years in environmental and occupational health; 30 years at California Department of Public Health managing the California Environmental Contaminant Biomonitoring Program and Exposure Assessment Section; extensive Prop 65 / Safe Harbor Level / Cal/EPA experience); Gideon Ramtahal (PhD Analytical Chemistry 2012, University of the West Indies, St. Augustine; cocoa-Cd soil-amelioration research focus).

HMTc interpretation

The numerical thresholds proposed in this report — both the Consent Judgment-stated drop-down trigger ladders (for which the Experts evaluated feasibility) and the alternative MADL-based and individually-derived Expert ladders — are policy outputs of a California Prop 65 negotiation, not literature-derived occurrence percentiles. They are not Codex maximum levels; they are not EU 2023/915 ceilings; they are not FDA action levels. The 2018-2021 AYS market-test database underlying the feasibility analyses contains brand-identified product-level Pb and Cd concentrations, but the Final Report main body does not republish brand-by-brand values — it works at the level of Group 1 (≤65% cacao), Group 2 (>65% to 95% cacao), and Group 3 (>95% cacao) compliance percentages. HMTc workups for the chocolate row should treat the proposed numerical thresholds in this report as one Prop 65-jurisdictional input among several (Codex CXS 193 Cd MLs, EU 2023/915 Cd MLs, FDA Closer to Zero action levels where they exist) and should preserve the legal-status label “Prop 65 Consent Judgment proposal — not currently in effect” when displaying any of the per-Expert drop-down numbers alongside enacted maximum levels.

The root-cause attributions and the bean-cleaning-and-winnowing mass-balance findings are the parts of this report that HMTc workups should weight most heavily, because they speak to platform-level (manufacturing-controllable vs supply-chain-controllable) reduction levers. The Committee’s strong, consensus conclusion that Pb in commercial cocoa products comes overwhelmingly from post-harvest soil/dust contact rather than pre-harvest soil uptake is a load-bearing finding for the chocolate Levers section, because it implies that Pb reductions are achievable at the cocoa-bean processing platform (via cleaning and winnowing practice changes) and at the post-harvest farm platform (via covered drying surfaces and elimination of ground-contact bean drying) — neither of which require breeding, grafting, or soil amendment. By contrast, the Committee’s strong consensus conclusion that Cd in commercial cocoa products comes primarily from pre-harvest soil uptake at country- and farm-specific levels implies that near-term Cd reductions for fine-flavor and specialty (Latin American / Caribbean origin) products are not achievable on manufacturing or business-practice levers alone — blending of high-Cd with low-Cd beans is the only High-confidence near-term Cd reduction strategy at scale for bulk cocoa, and even that strategy is constrained for fine-flavor products where ~90% of world supply is from the LAC region and >60% of LAC chocolate-liquor samples already exceed the current Consent Judgment Cd trigger at the proposed drop-down level.

The report’s documentation that bean-Cd levels in some LAC cocoa-growing regions reach 10 mg/kg vs the 0.8 mg/kg EU limit (with one Colombian location documented at 27 mg/kg topsoil Cd, a 100-fold background enrichment) is important corpus context for any Cd-in-cocoa synthesis pass: it establishes the upper tail of natural-plus-anthropogenic Cd soil concentrations that the LAC cocoa supply chain has to manage, and is the underlying physical reason why EU 2023/915 cocoa Cd ceilings are biting more heavily on LAC-origin specialty products than on West African-origin bulk products.

Root Cause Report — key findings (verbatim where quoted)

The Root Cause Report (Attachment 1) was the first technical output, submitted March 27, 2020. Its conclusions, as compiled in §4.1 of the Final Report (which states explicitly that the text is “quoted verbatim from the executive summary of the Root Cause Report”), are:

Pre-harvest conclusion for Cd (Section 4.1 of Root Cause Report). Soil is one of the major sources of Cd in cocoa beans, which occurs through direct uptake of phytoavailable Cd by the cocoa tree during the pre-harvest stage. Cd in cocoa-growing soils is a mixture of natural and anthropogenic sources, is highly variable, and is country- and farm-specific. Many studies (mostly in Latin America and the Caribbean) report that anthropogenic sources of Cd are of greater concern than natural sources in areas where historical or current industrial activities have resulted in elevated levels of soil Cd. Data and evidence are still needed to fully understand why pre-harvest cocoa beans from the African region have lower Cd. Soil physicochemical properties (soil pH, clay and/or organic material content, zinc levels) and cocoa-tree genetics are also very important interactive factors.

Pre-harvest conclusion for Pb (Section 4.2 of Root Cause Report). Only two publications (Rankin et al., 2005; Manton, 2010) used clean-room preparation, appropriately low limits of quantification, and clean preparation techniques to separate nibs from shells to measure Pb concentrations directly from cocoa pods on trees. These publications report pre-harvest concentrations of Pb in cocoa nibs and shells less than 3 ppb dry weight. Thus, nearly all Pb in commercial cocoa beans is from post-harvest contamination.

Harvest conclusion for Cd and Pb (Section 5). Though there may be the potential for Cd or Pb contamination of cocoa beans at the harvest stage, the Expert Committee believes that compared to other stages, this is most likely not a source of contamination. The Expert Committee does not recommend further study to evaluate this issue.

Post-harvest conclusion for Cd (Sections 6.1 to 6.4). The Expert Committee finds that the most significant source of Cd in cocoa beans occurs pre-harvest. While post-harvest activities might slightly change Cd concentrations, these changes are believed to be minimal in comparison to pre-harvest contributions.

Post-harvest conclusion for Pb (Section 6.5). Based on information in the scientific literature regarding sources of Pb in cocoa products and observations made during a tour of a cocoa-growing region, the Expert Committee finds that the most significant source of Pb in cocoa beans occurs post-harvest. This includes fermentation and drying of harvested beans both on-farm and off-farm. Scientific literature demonstrates that cocoa beans on the tree contain very low concentrations of Pb (less than 3 ppb dry weight). Cocoa beans have been demonstrated in commercial markets as high as >1,000 ppb dry weight. Pb in cocoa beans has not accumulated by uptake from soils in which the cocoa tree roots are growing and translocated to the nib. The most likely source of “post-harvest” Pb contamination is believed to be from the outdoor fermentation and drying of beans. Fermentation has been observed to occur in bags, covered piles, and wooden boxes exposed to exterior elements. Outdoor drying has been observed being performed along roadsides, on concrete patios, on drying tables, on plastic tarps, and in direct contact with the ground. Soil and dust containing Pb from anthropogenic sources is believed to come into contact with the cocoa bean shell and serve as the primary source of Pb to the cocoa bean.

Manufacturing-environment conclusion for Cd and Pb (Section 7). Cocoa-bean processing and chocolate-manufacturing operations are not a likely source of Cd or Pb in chocolate products. Regulatory inspections and food-safety certification audits assure that controls are in place to avoid the introduction of Cd or Pb through processing water, production equipment, product packaging, and environmental dust. Non-cocoa-based ingredients used in typical chocolate formulations are not a likely source of Cd or Pb beyond that already in raw-material shipments of cocoa beans. While not an additional source of Cd or Pb into chocolate products, bean-cleaning and shell-removal processes play a key role in the redistribution of Cd- or Pb-containing particles from cocoa beans and waste material to nibs. These processes are not specifically controlled to manage Cd or Pb levels in nibs.

Bean Cleaning and Winnowing Study (Reductions Recommendations Report Appendix C)

The Expert Committee conducted a Bean Cleaning and Winnowing Study at a major North American cocoa processing plant operated by one of the largest companies in the industry. The study’s purpose was to quantify the impact of the mechanical bean-cleaning and shell-removal steps on Cd and Pb concentrations.

The mechanical cleaning process, “typical of the industry,” reduced mean cocoa-bean Pb by 58%, from 95.2 ppb to 39.7 ppb. When compared with wet cleaning beans using water and detergent, there was no significant difference between the reductions obtained by mechanical cleaning versus wet cleaning. With respect to winnowing, the study concluded that nib Pb concentration is a function of particle size: nib Pb concentration increases as nib particle size decreases. Smaller nib particles have a higher surface area to volume ratio, which provides more contact area with Pb-containing material (shell, soil, light, and fine material) during bean breaking and winnowing. The study concluded that approximately 70% of nib Pb concentration is a result of contact during bean-breaking and winnowing.

Fine material removed during bean cleaning via aspiration and sieving was found to have a Pb concentration approximately 10 times higher than that of the shell. Nib Pb concentration cannot be fully accounted for by the Pb concentration of shell remaining in the nib after winnowing. This suggests that a significant source of Pb is enriched fine shell material, most likely resulting from Pb adsorption by the wet cocoa-bean shell during post-harvest handling — a mechanism proposed in Manton (J. Agric. Food Chem. 2010, 58, 713–721).

Bean Abrasion Study (Reductions Recommendations Report Appendix D)

The Bean Abrasion Study tested whether bean-to-bean abrasion could release Pb-rich particles as a Pb-reducing treatment. Several “shaking” methods were applied to 200 g of dry beans in 500 mL polyethylene bottles: roller, side-to-side, and vigorous hand shaking just short of breaking beans. Because little or no fine particles (<2 mm) were released by any of these shaking methods, only a small amount of broken shell, the Experts concluded that the fine particles were not easily released or were generated during or after the heating step before breaking and winnowing at cocoa-bean processing facilities. The study supports the bean-cleaning-and-winnowing study’s identification of the bean-breaking step itself as the locus of Pb transfer from shell-and-fines to nib.

Reductions Recommendations Report — 30 strategies, confidence-scored

The Expert Committee identified 30 potential Cd and Pb reduction strategies from the available literature (CAOBISCO/ECA/FCC 2015; CODEX 2020; Meter et al. 2019) and from its own deliberations. The strategies were grouped into Better Agricultural Practices (BAPs), Better Manufacturing Practices (BMPs), and Better Business Practices. Each strategy was rated on six feasibility factors (public-health impact, environmental impact, social impact, economic impact, technological considerations, scalability) plus the Committee’s overall confidence that the strategy would result in demonstrable Cd or Pb reductions in cocoa beans or chocolate products. Scoring was conducted three times independently by individual Committee members during report development; the third scoring was final and was blinded. Scoring results were grouped into “High,” “Medium,” and “Low” confidence terciles based on the range of confidence scores (0 to 94 percent).

The High-confidence strategies (15 of 30) are:

• Better Agricultural Practices (Cd): Exporters to stop purchasing beans from regions with high Cd phytoavailability (§4.1.1)
• Better Agricultural Practices (Cd): Farmers to stop planting new orchards in regions with high Cd phytoavailability (§4.1.2)
• Better Agricultural Practices (Cd): Use soil amendments to increase soil pH (§4.1.3)
• Better Agricultural Practices (Pb): Prevent Pb contamination of beans during fermenting and drying (§4.4.1.1)
• Better Manufacturing Practices (Cd): Blend beans or liquor as a Cd control measure (§5.1.1)
• Better Agricultural Practices (Cd): Use scion grafts to reduce Cd uptake from soils (§4.1.5.1)
• Better Agricultural Practices (Cd): Develop and plant rootstocks that accumulate less Cd from soils (§4.1.5.2)
• Better Agricultural Practices (Cd): Use zinc sulfate soil amendments to reduce Cd uptake from soils (§4.1.4)
• Better Agricultural Practices (Pb): Prevent Pb contamination of whole wet beans during transport (§4.4.1.2)
• Better Manufacturing Practices (Pb): Establish bean cleaning/winnowing QA practices for Pb contamination (§5.1.2)
• Better Business Practices: Incorporate Better Agricultural Practices into cocoa sustainability/certification programs (§6.1.1)
• Better Business Practices: Provide education/training at the local level to implement reduction strategies (§6.1.2)
• Better Agricultural Practices (Cd): Use self-rooted cocoa to reduce Cd uptake from soils (§4.1.5.3)
• Better Agricultural Practices (Cd): Use molecular breeding techniques to identify genotypes that accumulate less Cd (§4.1.5.4)
• Better Business Practices: Test surfaces of cocoa beans for Pb contamination at point of purchase (§6.1.3)

The Committee notes that the three categories of strategy approaches (agricultural, manufacturing, business) should be considered as a matrix or menu rather than a ranked list, and that selecting based on direct-reduction potential alone would miss the indirect “foundational” strategies that enable peak effectiveness of the direct strategies.

Warning Triggers Report — drop-down level findings

The Warning Triggers Report (Attachment 3, submitted November 15, 2021) was the third technical output. The Experts could not reach consensus on the lowest Pb and Cd drop-down levels feasibly achievable; they “agreed to disagree.” The Final Report therefore presents each Expert’s individual assessment.

The Consent Judgment defined three cacao-content product groups:

• Group 1: chocolate products with up to 65% cacao content
• Group 2: chocolate products with greater than 65% and up to 95% cacao content
• Group 3: chocolate products with greater than 95% cacao content

The current Consent Judgment Pb warning trigger levels and the Experts’ feasibility assessments are reproduced in Tables 3 and 4 of the Final Report. The Committee evaluated whether each of three candidate Pb drop-down levels per group could be feasibly achieved, plus the lowest level each Expert individually identified as feasible.

Table 3 — Pb feasibility (verbatim from Final Report):

For Group 1 (≤65% cacao), candidate levels 0.100 ppm and 0.065 ppm: all four Experts answered “Yes.” For the open-ended “<0.065 ppm” column, the Experts’ individually-identified lowest feasible Pb levels were Tim Ahn 0.060 ppm; Rufus Chaney 0.050 ppm; Michael DiBartolomeis 0.015 ppm; Gideon Ramtahal 0.060 ppm.

For Group 2 (>65% to 95% cacao), candidate levels 0.150 ppm and 0.100 ppm: all four “Yes.” For “<0.100 ppm,” lowest feasible: Ahn 0.090 ppm; Chaney 0.075 ppm; DiBartolomeis 0.020 ppm; Ramtahal 0.090 ppm.

For Group 3 (>95% cacao), candidate levels 0.225 ppm and 0.200 ppm: all four “Yes.” For “<0.200 ppm,” lowest feasible: Ahn 0.180 ppm; Chaney 0.100 ppm; DiBartolomeis 0.040 ppm; Ramtahal 0.180 ppm.

Table 4 — Cd feasibility (verbatim from Final Report):

For Group 1 (≤65% cacao), candidate levels 0.400 ppm and 0.320 ppm: Ahn “No” / “No”; Chaney “Yes” / “No”; DiBartolomeis “Yes” / “Yes”; Ramtahal “Yes” / “No.” For “<0.320 ppm”: Ahn “No”; Chaney “No”; DiBartolomeis “Yes (0.120 ppm)”; Ramtahal “No.”

For Group 2 (>65% to 95% cacao), candidate levels 0.450 ppm and 0.400 ppm: Ahn “No” / “No”; Chaney “Yes” / “No”; DiBartolomeis “Yes” / “Yes”; Ramtahal “Yes” / “No.” For “<0.400 ppm”: Ahn “No”; Chaney “No”; DiBartolomeis “Yes (0.175 ppm)”; Ramtahal “No.”

For Group 3 (>95% cacao), candidate levels 0.960 ppm and 0.800 ppm: Ahn “No” / “No”; Chaney “Yes” / “Yes”; DiBartolomeis “Yes” / “Yes”; Ramtahal “Yes” / “No.” For “<0.800 ppm”: Ahn “No”; Chaney “No”; DiBartolomeis “Yes (0.400 ppm)”; Ramtahal “No.”

The disagreement between DiBartolomeis (who proposes the most aggressive drop-down levels based on his Cal/EPA-aligned Prop 65 MADL back-calculation) and the other three Experts (who treat 0.450 ppm as the floor for Group 2 Cd) is a direct artifact of the two different methodological frames invoked: DiBartolomeis derives his floors from the Prop 65 MADL of 4.1 µg Cd/day and 0.5 µg Pb/day plus consumption data and a 50%-reduction assumption from High-confidence feasibility strategies, while Ahn, Chaney, and Ramtahal derive their floors from observed AYS market-test compliance percentages adjusted by High-confidence-strategy reduction efficiency without a population-exposure recalculation.

Individual Expert affirmative conclusions

The Final Report includes 2-page individually-authored affirmative conclusions from each Expert (§3.1-§3.4). These are not consensus consensus content; they are individually attributable scientific positions. The Project Manager confirmed in advance that no edits other than minor formatting would be applied to the affirmative-conclusion text.

Ahn (§3.1). Cocoa-bean processing and chocolate-manufacturing operations are not the source of Cd in chocolate products. The significant source of Cd in chocolate products is chocolate liquor, produced from milled cocoa nibs, with Cd entering the bean pre-harvest by soil uptake. For Pb, scientific literature reports pre-harvest Pb in cocoa nibs and shells <3 ppb from cocoa beans sampled directly from the pod. Post-harvest contact with soil and dust during fermentation, drying, and transport is the source of post-harvest Pb. Ahn proposes Pb Product Warning Triggers feasibly achievable through bean cleaning and winnowing manufacturing practices of 0.060 ppm (Group 1), 0.090 ppm (Group 2), and 0.180 ppm (Group 3). Ahn explicitly states that for fine-flavor cocoa products (where ~90% of world fine-flavor cocoa supply comes from the LAC region), the only feasible reduction strategy is blending high-Cd with low-Cd beans, but at the drop-down Cd trigger levels approximately 70% of LAC chocolate liquor would exceed the trigger — making blending impractical and leaving no feasible drop-down strategy for fine-flavor cacao products.

Chaney (§3.2). Affirmative Conclusions for Pb: Research by Manton (2010) showed unequivocally that Pb in cocoa nibs above ~3 ng Pb/g (ppb) is contamination post-harvest. Careful separation of shells and nibs in a clean room confirmed low nib Pb for beans with as high as 2900 ng Pb/g shell. With uncontaminated soils, beans can gain 100 ng Pb/g; if soil is rich in Pb, higher bean Pb levels can occur. Some bean-processing/cleaning technologies are able to remove most of the Pb that arrived with the beans. Affirmative Conclusions for Cd: In contrast with Pb, Cd in cocoa beans is accumulated by the cocoa roots and transported to beans by natural processes. Excessive nib Cd levels occur due to: cocoa as a natural Cd-accumulating species; topsoil Cd increases from litter and harvest-debris deposition; difficulty incorporating soil amendments without root damage in perennial orchards; LAC soils with pH as low as 4.5 making soil Cd highly phytoavailable; historical mining-waste contamination in part of LAC cocoa land; and parent-rock-enriched soils (e.g., marine shale) with high Cd:Zn ratios and thus high Cd phytoavailability. In these soils, beans can reach 10 mg Cd/kg rather than the 0.8 mg Cd/kg EU limit. One location in Colombia has 27 mg Cd/kg topsoil, 100-fold background. Other high Cd:Zn problem soils growing cocoa identified in Ecuador, Honduras, Peru, and Trinidad & Tobago. Cocoa plants absorb Cd into roots on a ZIP-family Zn transporter, not the NRAMP5 transporter used by rice. Zn²⁺ in soil solution strongly inhibits Cd²⁺ uptake; higher soil Zn can reduce bean Cd. Genetic solutions (HMA3 promoter modification analogous to rice work, or grafting onto low-Cd cocoa rootstocks) are the most promising long-term direction; finding a low-Cd cocoa rootstock would be fastest.

DiBartolomeis (§3.3). Proposes, as feasible, health-protective Prop 65 warning trigger levels of 15 ppb (Group 1), 20 ppb (Group 2), and 40 ppb (Group 3) for Pb; and 120 ppb (Group 1), 175 ppb (Group 2), and 400 ppb (Group 3) for Cd. AYS market-based test data for nearly 600 chocolate products were the only California-specific data available; SD data had limited utility for Prop 65 trigger derivation due to technical flaws in testing/reporting, lack of traceability, and lack of California specificity. Under status quo: all products are >95% compliant with the current Consent Judgment Pb triggers and >81% compliant with the current Consent Judgment Cd triggers — current triggers are not health-based. Most-public-health-protective Pb triggers using MADL back-calculation: 14, 14, and 40 ppb for Groups 1, 2, 3 respectively. Under status quo, 20-56% of products are already compliant with these MADL-based Pb triggers. Most-public-health-protective Cd triggers using MADL: 111, 111, 325 ppb for Groups 1, 2, 3. Under status quo, 25-69% of products are compliant. DiBartolomeis proposes alternatives that are “protective of health but even more feasible to attain”: Pb at 15/20/40 ppb (status quo 35-57% compliant, projected 75%+ compliant with High-confidence feasibility Pb reduction strategies implemented at 50% reduction; potentially >95% compliant at 60-75% Pb reduction efficiency); Cd at 120/175/400 ppb (status quo 57-73% compliant, projected >80% compliant at 50% Cd reduction). DiBartolomeis notes that 27% of AYS-tested products were at or below 10 ppb Pb under status quo conditions, suggesting that in regions with low environmental Pb contamination and clean post-harvest processes, Pb in chocolate will be negligible. For specialty (single-origin and organic) chocolate, ~22% of AYS-tested products, Cd compliance is 25-34% under status quo and rises to ≥70% with 50% Cd reduction from High-confidence feasibility strategies; Pb compliance for specialty products is comparable to bulk.

Ramtahal (§3.4). Pb contamination occurs primarily during post-harvest, with cocoa-manufacturing-facility bean-cleaning and shell-removal as a secondary but significant influence. Pb-BAPs at the post-harvest stage are key but will vary country-to-country. The Committee’s bean-cleaning-and-winnowing study at a major North American facility demonstrated demonstrable Pb reductions at the bean-cleaning and winnowing phase. For Pb drop-down trigger feasibility, the majority of AYS and SD sample data could meet the current and proposed triggers, but cocoa-powder origin-country SD data showed that 25% of countries may have the potential to exceed the proposed 0.200 ppm trigger (for >95% cacao products) with a Pb concentration of 0.180 ppm — Ramtahal used 0.180 ppm to determine feasible drop-down levels. For Cd, pre-harvest contributions are predominant; soil pH, organic matter, and Zn levels influence Cd uptake; soil-amelioration approaches will be farm-specific. Until evidence-based Cd-BAPs are validated, blending beans or liquor is the short-term solution. Ramtahal’s view is that the original Cd warning triggers should not be modified for chocolate products with higher cacao content (>65%) because LAC beans destined for fine-flavor cocoa markets could be excluded by tighter triggers, affecting LAC farmer livelihoods.

Industry-context observations the Committee establishes

Several factual claims relevant to the chocolate product category emerge from the Final Report and are useful inputs to the wiki:

• The cocoa market consists of two product categories based on bean type: “fine flavor” and “bulk or common.” Fine-flavor accounts for approximately 10-15% of world cocoa exports. Latin America and the Caribbean account for approximately 90% of fine-flavor cocoa production (Ahn §3.1).
• Ahn (§3.1) reports that at Consent Judgment current Cd trigger levels, approximately 60% of chocolate liquor from the LAC region exceeds the trigger; at the drop-down Cd trigger levels, this rises to approximately 70%. This is an individually-authored affirmative-conclusion finding from Ahn, not a Committee consensus statement.
• The chocolate industry obtains certified analysis of whole-bean Cd and Pb prior to purchase, but cocoa-bean shell is significantly higher in Cd than the nib. Beans with lower nib-Cd may be rejected based on irrelevant whole-bean Cd measurements; the Nib-Cd/Shell-Cd ratio varies genetically. This is a market inefficiency the Committee documents.
• Pre-export bean cleaning and packaging facilities in Ecuador were observed during the Committee’s field trip; drying practices on roadsides, concrete patios, drying tables, and plastic tarps in direct contact with the ground are documented as observed real-world practices generating post-harvest Pb contamination.
• Fermentation occurs in bags, covered piles, and wooden boxes “exposed to exterior elements” — these are the venues at which post-harvest Pb exposure begins.

Methods (sampling, analytical, and reduction-strategy scoring)

The Committee’s evidence base draws on five lines: (1) sampling data, including a large non-public SD dataset, peer-reviewed literature, and the Committee’s own Ecuador field sampling; (2) >400 scientific publications including peer-reviewed literature, graduate theses, and unrefereed sources reviewed for relevance; (3) >250 outside subject-matter expert consultations made by individual Committee members; (4) direct observations from two field trips (three U.S. industrial facilities; an end-to-end Ecuador supply-chain trip including farms, collection points, and exporters); and (5) the Committee’s professional judgment.

The two Committee-conducted studies — the Bean Cleaning and Winnowing Study and the Bean Abrasion Study — are documented in Appendixes C and D of the Reductions Recommendations Report respectively (these appendixes are within Attachment 2 of the Final Report; full appendix text is not reproduced in the Final Report’s executive-summary section). The Bean Cleaning and Winnowing Study was conducted at “a major North American cocoa processing plant operated by one of the largest companies in the industry.”

The reduction-strategy scoring was conducted three times by individual Committee members across the report’s drafting period; the third scoring was final and was blinded so that scores submitted by individual members are not identifiable. The Committee acknowledges in the report that reaching consensus on each individual feasibility score was not possible given the project timeline, data limitations, and variation among Committee member backgrounds; the discussion sections in §§4-6 of the Reductions Recommendations Report (within Attachment 2) provide qualitative narrative complementing the quantitative scoring.

Key numbers

• Mean cocoa-bean Pb before mechanical cleaning, Bean Cleaning and Winnowing Study: 95.2 ppb.
• Mean cocoa-bean Pb after mechanical cleaning, same study: 39.7 ppb (58% reduction).
• Fraction of nib Pb concentration attributed to contact during bean-breaking and winnowing: approximately 70%.
• Pb concentration of fine material removed during bean cleaning via aspiration and sieving, relative to shell: approximately 10× higher.
• Pre-harvest Pb in cocoa nibs and shells sampled directly from pods (Rankin et al. 2005; Manton 2010): less than 3 ppb dry weight.
• Pb in cocoa beans demonstrated in commercial markets: as high as >1,000 ppb dry weight.
• Pb gain in beans grown on uncontaminated soils, Chaney’s estimate: ~100 ng Pb/g (~100 ppb).
• Highest documented bean Cd in LAC growing regions (Chaney §3.2): up to 10 mg Cd/kg vs the 0.8 mg Cd/kg EU limit.
• Documented Colombian topsoil Cd anomaly: 27 mg Cd/kg, 100-fold background.
• Other LAC countries with documented high-Cd cocoa-growing soils: Ecuador, Honduras, Peru, Trinidad & Tobago.
• Cocoa Cd uptake transporter: ZIP-family Zn transporter (cocoa) versus NRAMP5 (rice).
• Fine-flavor share of world cocoa exports: approximately 10-15%.
• LAC share of world fine-flavor cocoa production: approximately 90%.
• Number of chocolate products in AYS market-test database used by Committee: nearly 600.
• AYS data Group distribution: 40% Group 1 (≤65% cacao), 48% Group 2 (>65-95% cacao), 12% Group 3 (>95% cacao); specialty products ~22% of total.
• Status-quo compliance with current Consent Judgment Pb triggers (DiBartolomeis): >95%.
• Status-quo compliance with current Consent Judgment Cd triggers (DiBartolomeis): >81%.
• Current Consent Judgment proposed Pb trigger ladder (per Final Report §3.3 narrative): Group 1: 100 and 65 ppb; Group 2: 150 and 100 ppb; Group 3: 225 and 200 ppb.
• Current Consent Judgment proposed Cd trigger ladder: Group 1: 400 and 320 ppb; Group 2: 450 and 400 ppb; Group 3: 960 and 800 ppb.
• DiBartolomeis MADL-based Pb triggers: 14, 14, 40 ppb (Groups 1, 2, 3).
• DiBartolomeis MADL-based Cd triggers: 111, 111, 325 ppb (Groups 1, 2, 3).
• DiBartolomeis proposed feasible Pb drop-down: 15, 20, 40 ppb (Groups 1, 2, 3).
• DiBartolomeis proposed feasible Cd drop-down: 120, 175, 400 ppb (Groups 1, 2, 3).
• DiBartolomeis lowest feasible Pb (Table 3 “<lowest” column): 15 ppb (Group 1), 20 ppb (Group 2), 40 ppb (Group 3).
• DiBartolomeis lowest feasible Cd (Table 4 “<lowest” column): 120 ppb (Group 1), 175 ppb (Group 2), 400 ppb (Group 3).
• Ahn lowest feasible Pb: 60, 90, 180 ppb (Groups 1, 2, 3).
• Chaney lowest feasible Pb: 50, 75, 100 ppb (Groups 1, 2, 3).
• Ramtahal lowest feasible Pb: 60, 90, 180 ppb (Groups 1, 2, 3).
• Ahn/Ramtahal Cd verdict at all candidate drop-down levels for Groups 1 and 2: “No” (not feasible).
• Chaney Cd feasibility: “Yes” at top-of-range (0.400/0.450/0.960 ppm) for all three Groups, “No” at the lower candidate or open-ended floor for all three Groups.
• Pb total ranges in pre-harvest nibs (Rankin/Manton clean-room studies): <3 ppb dry weight.
• Number of candidate Cd and Pb reduction strategies evaluated: 30.
• High-confidence strategies: 15 of 30. Medium-confidence: 5 of 30. Low-confidence: 10 of 30.
• Range of Committee confidence scores across the 30 strategies: 0 to 94 percent.
• Outside subject-matter experts contacted by Committee members: >250.
• Scientific publications reviewed by Committee: >400.
• Project span: March 2018 through April 2022; three earlier technical reports plus the March 28, 2022 Final Report.

Implications

For the chocolate product-category page, this report supplies (a) the strongest available evidence that Pb in cocoa products is overwhelmingly post-harvest in origin and that 70% of nib Pb is acquired during the bean-breaking/winnowing step at processing facilities, which makes Pb reductions amenable to manufacturing-platform interventions; (b) the strongest available evidence that Cd in cocoa products is overwhelmingly pre-harvest in origin and that LAC origin is the dominant Cd risk axis; (c) the consent-judgment-proposed Prop 65 Pb and Cd trigger ladders by cacao-content group (Consent Judgment current proposed levels and the four Experts’ individually-derived drop-down levels), which are non-binding policy outputs of the Prop 65 negotiation but a useful comparator alongside Codex CXS 193 Cd MLs and EU 2023/915 Cd MLs; and (d) the AYS market-test compliance percentages by Group at each candidate threshold level.

For the cocoa ingredient page, this report establishes (a) that bean-Cd levels in LAC growing regions can reach 10 mg/kg, an order of magnitude above the EU 0.8 mg/kg bean-import limit, and that one Colombian location has 27 mg/kg topsoil Cd at 100× background; (b) that the Nib-Cd/Shell-Cd ratio varies genetically and the chocolate industry’s reliance on whole-bean Cd analyses introduces a sourcing inefficiency that may reject beans with acceptable nib-Cd levels; (c) that the Cd uptake transporter is ZIP-family (Zn-displaceable) rather than NRAMP5, making soil Zn amendment a mechanistically targeted Cd reduction strategy; and (d) that pre-harvest Pb in clean-room-sampled cocoa nibs is <3 ppb dry weight, with commercial bean Pb reaching >1,000 ppb dry weight, an enrichment of >300× attributable to post-harvest contamination.

For the Pb metal page, this report adds (a) the Manton (2010) mechanism of Pb adsorption by the wet acidic cocoa-bean shell during post-harvest contact with soil/dust, with subsequent redistribution to nibs during bean breaking and winnowing; (b) the quantitative observation that fine material removed during bean cleaning has Pb 10× higher than shell, consistent with adsorption-enrichment of shell fines; (c) the field observation that fermentation in bags/covered piles/wooden boxes exposed to exterior elements and drying on roadsides/concrete patios/drying tables/plastic tarps/ground contact are the dominant venues of post-harvest Pb exposure.

For the Cd metal page, this report adds (a) the soil pH dependence of Cd phytoavailability in cocoa-growing soils, with LAC soils as low as pH 4.5 making soil Cd highly phytoavailable; (b) the Zn-Cd antagonism on the ZIP transporter as the mechanistic basis for ZnSO₄ soil-amendment Cd reduction strategies; (c) the dominant role of soil Cd:Zn ratio over absolute Cd concentration in determining bean Cd accumulation.

For the California-Prop 65 regulatory-context surface, this report establishes the consent-judgment-driven negotiation that produced the proposed Pb and Cd trigger ladders for the three cacao-content Groups, with four-Expert feasibility evaluations and the individual lowest-feasible-floor proposals from each Expert. The proposed levels are not currently in effect; they are an input to ongoing Prop 65 process. The wiki entries cross-reference this report from oehha-cadmium-prop65 and oehha-lead-prop65 as the cocoa/chocolate-specific Prop 65 expert investigation.

Verification notes

• Ingested 2026-06-02 from the Kimi-agent download-corruption manual-fetch batch (condiments2_papers/04_Chocolate_Seasonings_Salt). The PDF is a 381-page consent-judgment-mandated Expert Committee report; this source page treats the Final Report front matter (pages i-iv of bios, abbreviations, and contents) plus the report body (pages 1-21 covering the executive summary, background and process, four individually-authored affirmative conclusions, and verbatim executive summaries of the three earlier technical reports) plus the cited-references list (pages 21-onward) as the primary readable scope. The three technical attachments (Root Cause Report; Reductions Recommendations Report; Warning Triggers Report) are referenced via their executive-summary content as compiled in §§4.1-4.3 of the Final Report. The Final Report states explicitly (§4.1, §4.2, §4.3) that the §4.x executive-summary text is “quoted verbatim from the executive summary of the [Root Cause | Reductions Recommendations | Warning Triggers] Report,” so the verbatim quotations used in this source page from §4.x are the technical-attachment text as presented in the consolidated Final Report.
• No DOI is assigned. The report is a consent-judgment-mandated work-product, not a peer-reviewed publication. no_doi_assigned: true is recorded in frontmatter. access_url is left null because the canonical hosting URL was not re-verified at ingest; related_urls records the As You Sow program-page pattern (asyousow.org/reports/2022-expert-investigation-cocoa-chocolate) as the most likely canonical-document index, but this URL pattern is not a re-verified live link.
• Evidence tier B reflects the source’s grey-literature status (not peer-reviewed) balanced against the depth of expertise on the Committee (four credentialed experts including a USDA-ARS Senior Research Agronomist with H-index 105), the breadth of its evidence base (>400 publications reviewed, >250 outside expert consultations, two primary studies conducted by the Committee, two field trips, three years of work), and the consent-judgment-mandated independence framing. The B tier is consistent with how the corpus treats other industry/government technical reports of comparable rigor (e.g., the CBI 2024 buyer-requirements article also tagged B; see cbi2024-cocoa-eu-market-requirements).
• Brand-firewall (Part 12) handling: The Final Report main body does NOT republish the underlying AYS market-test brand-level Pb and Cd values. AYS-test data are presented only at the Group 1/2/3 aggregate compliance-percentage level. The “Settling Defendants” are referenced only as a category of consent-judgment parties, not by individual company names in the report body. The Bean Cleaning and Winnowing Study is described as conducted at “a major North American cocoa processing plant operated by one of the largest companies in the industry” — the host company is not named in the Final Report. No brand-by-brand values are introduced into this source page. Where the report describes regulatory/consent-judgment events (the consent judgment between AYS and the SDs), these are characterized as enforcement-program events per Part 12 Exception 1 (recalls and enforcement actions are public-record and may be named on regulations and products pages framed as regulatory events). No HMTc-certified brands are mentioned anywhere on this page.
• Wiki/HMTc firewall (Part 2) handling: The proposed Pb and Cd Prop 65 drop-down trigger levels are explicitly labelled in this source page as “policy outputs of a California Prop 65 negotiation, not literature-derived occurrence percentiles” and “not currently in effect.” The Implications section frames the report as supplying input to ongoing Prop 65 process and as a Prop 65-jurisdictional comparator alongside Codex CXS 193 and EU 2023/915, not as HMTc threshold candidates. No HMTc within-row clean/dirty subcategory pre-decision is made on the chocolate row; that work belongs in the Step 0 Lock for chocolate, which is not addressed here. Part 19 percentile-selection arithmetic (P97/P45) is NOT performed; this is a source page, not a workbench output.
• Speciation labelling: The report consistently writes “Pb” and “Cd” as total Pb and total Cd; no inorganic-versus-organic distinction applies for Pb in this matrix (Pb in food is generally treated as total Pb). No total-Hg or methylmercury content; Hg is not within report scope. No Cr or As speciation; these metals are not within report scope.
• Units fidelity: The report mixes ppm (mg/kg) in the consent-judgment-trigger tables and ppb (ng/g, µg/kg) in the bean-cleaning-and-winnowing study results. The “Key numbers” section preserves the units as printed in each context; cross-table comparisons should convert 0.060 ppm = 60 ppb, 0.090 ppm = 90 ppb, 0.180 ppm = 180 ppb, etc. The bean Pb basis is “dry weight” where stated by the report (Rankin/Manton clean-room sampling; commercial bean Pb >1,000 ppb dry weight); the cleaning-study results (95.2 ppb → 39.7 ppb) are reported without an explicit dry/wet-weight basis but in the context of cleaning fresh-product input streams at a processing plant — corpus convention treats unspecified bean/nib bases as dry weight for cocoa unless stated otherwise.
• Geography and sampling-locations: The report’s evidence base spans California (AYS market testing of finished chocolate products sold in California), Ecuador (Committee field trip and bean sampling), Trinidad & Tobago (Ramtahal’s research base), Colombia (Chaney’s documented 27 mg/kg topsoil location), Peru, Honduras (Chaney’s high-Cd:Zn problem-soils list), Brazil (Ahn’s Mars experience), Indonesia (Ahn’s Mars experience), Ghana (Ahn’s Mars experience), and Côte d’Ivoire (Ahn’s Mars experience). sampling_year_range: [2014, 2021] reflects AYS market testing 2014-2017 plus the Committee’s evidence-gathering and field-work window 2018-2021. sampling_locations is intentionally left empty as the report does not present a single coherent sampling-location array — the Committee’s evidence is multi-source, multi-method, and not reducible to a discrete sampling-locations list.
• Provisional-scaffold posture for chocolate: The chocolate product page exists (chocolate). This source’s products field declares [[products/chocolate]] only. No new product page is created. Ramtahal’s report-internal “fine flavor” and “bulk or common” distinction is documented in the body and Key numbers but is NOT promoted to a separate product slug; that would be a Step 0 Lock decision outside this ingest’s scope.
• Ingredient routing: ingredients declares [[ingredients/cocoa]] and [[ingredients/chocolate]]. Both pages exist (verified via Bash ls at ingest time). The cocoa-nibs, cocoa-shell, cocoa-liquor, and cocoa-bean matrix values are recorded in the matrices array but no separate ingredient pages are proposed by this ingest.
• Audit subagent (2026-06-02, fresh-context general-purpose) verdict PROMOTE. The subagent independently re-verified ~25 numerical values against the PDF source — including the Bean Cleaning and Winnowing Study (95.2 → 39.7 ppb, 58% reduction, 70% nib Pb), the pre-harvest <3 ppb dry-weight clean-room result, the 10 mg Cd/kg LAC bean and 27 mg Cd/kg Colombian topsoil figures, the AYS ~600-product market-test count, the Group distribution (40%/48%/12%), the >95%/>81% status-quo Pb/Cd compliance figures, the full Pb Table 3 and Cd Table 4 verbatim values for all four Experts across all three cacao-content Groups, and the 15/5/10 confidence-tier strategy counts — all match exactly. Checks 3 (speciation/methods/units), 4 (Part 12 brand firewall), and 5 (Part 2 wiki/HMTc firewall) all clean. Two ⚠️ concerns raised on Check 1 plus one ⚠️ on Check 2, all three applied: (a) Industry-context observation regarding LAC chocolate-liquor exceedance percentages (60% current / 70% drop-down Cd triggers) re-attributed to Ahn §3.1 individually rather than to the Committee consensus, since this is an individually-authored affirmative-conclusion finding; (b) Key-numbers line on commercial-market bean Pb tightened from “market-collected samples” paraphrase to “demonstrated in commercial markets” to track the source language; (c) matrices vocabulary plural-form alignment: cocoa-shell → cocoa-shells, cocoa-bean → cocoa-beans, consistent with the corpus precedent in oliveira2021-cocoa-beans-cd-pb-brazil-processing (matrices: [cocoa-beans, cocoa-liquor, cocoa-butter, cocoa-powder, cocoa-shells, chocolate]). No false positives recorded.

Sources

• Ahn, T., R.L. Chaney, M.J. DiBartolomeis, and G. Ramtahal. 2022. Expert Investigation Related to Cocoa and Chocolate Products: Final Report. Submitted to As You Sow and the Settling Defendants under a California Proposition 65 Consent Judgment. Project Manager: Eastern Research Group, Inc. (ERG). March 28, 2022. 381 pp. including three technical attachments (Root Cause Report; Reductions Recommendations Report; Warning Triggers Report).
• Manton, W.I. 2010. Sources of lead and other metals in cocoa products. J. Agric. Food Chem. 58:713–721. [referenced by Chaney’s affirmative conclusion and by the Root Cause Report executive summary as the load-bearing clean-room study establishing pre-harvest nib Pb <3 ppb].
• Rankin, C.W., et al. 2005. [Pre-harvest cocoa bean Pb measurement]. [referenced by the Root Cause Report executive summary as the second load-bearing clean-room study].
• CAOBISCO/ECA/FCC. 2015. Cocoa Beans: Chocolate and Cocoa Industry Quality Requirements. Chocolate, Biscuits, and Confectionery of Europe; European Cocoa Association; Federation of Cocoa Commerce. [referenced as one of the Committee’s prior-art sources for candidate reduction strategies].
• CODEX. 2020. CX/CF 20/14/7. Proposed Draft Code of Practice for the Prevention and Reduction of Cadmium Contamination in Cocoa Beans. 48 pp. Joint FAO/WHO Food Standards Programme.
• Meter, A., et al. 2019. [referenced by the Committee as a prior-art source on Cd in cocoa].
• Ramtahal, G., et al. 2022. [Surface application of hydrated lime / biochar / ZnSO₄ on field-grown cocoa, referenced by Chaney’s affirmative conclusion].

Page history

The five most recent substantive edits to this page. The full version history lives in git; when DOI minting comes online (see schema docs), each entry below will also link to a version-pinned DataCite DOI.