Kalicharan et al. 2025 — Heavy metal contamination in extruded pet food, South Africa

Kalicharan and colleagues quantified nine metals (As, Cd, Co, Cu, Fe, Ni, Pb, Zn by ICP-OES; Hg by HGAAS) in 31 extruded dry kibble products available in South Africa, split between 14 premium brands (PB) and 17 supermarket brands (SB) and across protein formulations (red meat, poultry, fish, meat medley, “other”). Concentrations are reported as mean ± SD with min-max ranges in mg/kg on an as-purchased basis and compared against FDA/NRC and FEDIAF maximum tolerable limits (MTLs). The paper is held in this wiki as a methodological and supply-chain cross-reference: the matrix is pet food, outside HMT&C’s human-food scope.

Scope classification: This source does not contribute occurrence data to any HMT&C-scope product page. It is routed only to pet-food-dry-kibble (HMTc Cat 17 Row 1, pet-food taxonomy) and not to human-poultry, human-fish, or human-meat product pages, even though the kibble formulations are described by their human-style protein source.

Key numbers

Non-essential metals (Table 5, mg/kg). Maximum tolerable limits (MTLs): As 12.5 (NRC, most-sensitive-mammal reference), Cd 10 (FDA), Co 2.5 (FDA), Ni 50 (FDA), Pb 10 (FDA), Hg 0.27 (FDA).

Essential metals (Table 6, mg/kg). FEDIAF minimum recommended limits (MRL) cats/dogs and EU MTL: Cu 5/7.5 MRL, 28 MTL; Fe 80/36 MRL, 681.8 MTL; Zn 75/72 MRL, 227 MTL.

All 16 cat food samples and all 15 dog food samples met or exceeded the FEDIAF minimum recommended limits for Cu, Fe, and Zn (essential-nutrient floor), while simultaneously breaching MTLs at the rates shown above.

[†] Source-level data anomaly. The dog-food Zn mean of 25.25 ± 68.21 mg/kg as printed in Table 6 (and reproduced in the abstract) is internally inconsistent with the same row’s reported range (79.86–353.55) and exceedance count (13 of 15 samples above the 227 mg/kg MTL). A mean of 25.25 cannot coexist with a minimum of 79.86 and 13/15 samples exceeding 227. The mean as printed is likely a typographical error in the published paper; the range and exceedance count are mutually consistent and consistent with the Conclusion text. Wiki preserves the printed value for fidelity but flags the inconsistency here so downstream synthesis does not naively rely on 25.25 as a usable central tendency.

Total heavy-metal load by protein source (Tables 7 and 8, mg/kg, dry-matter THM sum).

An outlier of 1034.54 mg/kg total heavy metals was observed in the canine supermarket range (Discussion p. 4–5).

Aggregate findings (Conclusion, p. 6). Fourteen cat food brands exceeded the MTL for Cu; five exceeded for Zn. Nine dog food brands exceeded the MTL for Cu, one for Fe, 13 for Zn. Ten dog food brands and two cat food brands exceeded the MTL for Hg. Ten dog food and seven cat food brands exceeded MTLs for Co. Mercury exceedances reached up to 26.6× the MTL in one dog food sample.

Methods (brief)

Thirty-one kibble samples were quartered (Campos and Campos 2017 quartering method), milled individually to prevent cross-contamination, and stored in re-sealable bags. Aliquots of ~0.3 g were weighed in triplicate into Teflon vials, pre-digested in 7 mL of 1 M nitric acid for 1 h at room temperature, then microwave-digested (CEM MARS 6) with a 15-min ramp to 180 °C, 10-min hold, and 15-min cooldown. Digests were filtered (0.45 µm nylon) and diluted to 25 mL with double-distilled water. ICP-OES (Perkin Elmer Optima 5300DV; plasma 15 L/min; carrier 0.24 L/min; 1–5 min/sample throughput) quantified Cu, Cd, Fe, Co, Ni, Pb, As, Zn against a five-point calibration from 1000 ppm mono-element stock standards (LabChem, Polychem). Mercury used a Perkin Elmer A Analyst 200 AAS with a multi-mode hydride system (MHS 15), 3% NaBH4/1% NaOH reductant, 1.5% nitric-acid diluent, and detection at 253.7 nm. LODs (mg/kg) ranged from 0.00012 (Zn) to 0.46262 (Pb); LOQs from 0.00005 (Cu) to 1.40188 (Pb); linearity R² ≥ 0.9918 for all eight ICP-OES elements (Table 2). All analyses were performed in triplicate with procedural blanks. No certified reference material (CRM) is reported; quality control rested on procedural blanks and emission-line correlation > 0.99.

Limitations

The matrix is pet food, not human food; regulatory comparators are FDA/NRC/FEDIAF pet-food MTLs, which are not equivalent to human-food limits. Sample size is small (n = 31). Brand-level data are not reported in the paper (aggregate tier-level summaries only). No CRM is documented, so traceability of accuracy depends on calibration linearity and procedural blanks alone. The paper aggregates over per-brand replicate triplicates without reporting within-brand variability separately from across-brand variability. Speciation is not performed for arsenic (tAs) or mercury (HGAAS reductive method measures total Hg, not MeHg).

Implications

Cross-reference value for the wiki: methodological corroboration of multi-element ICP-OES + HGAAS workflows on protein-rich matrices, and documentation of a regional (South African) pet-food contamination snapshot that may inform supply-chain discussions of animal by-product as a vehicle for metal accumulation. The paper does not contribute occurrence data to any human-food HMT&C product-page row and is not eligible for synthesis into ingredient contamination_profile blocks.

Verification notes

• 2026-05-18 enhancement (Claude Opus 4.7): corrected truncated DOI (10.1002/vms3.703 → 10.1002/vms3.70319 per p. 1 footer); added missing metals Co and Fe to the metals: array (the paper measures nine elements; the prior page listed seven); removed the products/poultry-product slug from the products: array because that slug routes to HMTc Cat 11 Row 4 (human poultry meat) and this paper measures pet kibble with poultry-derived protein, not human-grade poultry — mis-routing would falsely contribute pet-kibble values to the human poultry contamination evidence base; collapsed matrices: [pet-food, meat-based, poultry-based, fish-based] to matrices: [pet-food] because pet-food is the matrix and the others are formulation descriptors that belong in the prose, not the matrix vocabulary; renamed “Key concentration data” → “Key numbers” and “Methodological relevance” → “Methods (brief)” to align with Part 6 source-page template; expanded Key numbers to include exact MTLs, per-row exceedance counts, ranges, and the Tables 7/8 protein-source breakdowns; removed the prior “Supply-chain cross-relevance” paragraph because its assertion “consistent with the known biology: large muscle and organ tissue in ruminants concentrates Cd, Pb, and As through longer bioaccumulation pathways than poultry” was a Part 2 synthesis claim across sources; removed the legacy “Wiki pages updated on ingest” trailing list per Part 6 template guidance.
• Matrix vocabulary note: pet-food is not in the standard matrices list documented at docs/gpt-collaboration/system-prompt.md. Used here because no listed matrix slug accurately describes commercially extruded companion-animal kibble; flagged for Karen’s review.
• 2026-05-18 audit-subagent application (Claude Opus 4.7): Subagent flagged Table 7 DF-SB Red meat SD as 44.27, claiming 44.55 in wiki was a transcription error — verified false positive: PDF p. 5 Table 7 prints “812.01 ± 44.55” for DF-SB Red meat (and a min-max range “846.94–762.23” that is itself impossibly inverted but is also faithfully reproduced from the published table). No change applied. Subagent flagged Table 6 DF Zn mean (25.25 ± 68.21) as inconsistent with the same row’s range (79.86–353.55) and exceedance count (13 of 15 above 227 mg/kg MTL) — verified correct: the printed mean cannot coexist with the printed range, indicating a typo in the source paper itself; added an inline footnote and explanatory paragraph beneath the essential-metals table preserving the printed value while flagging the source-level anomaly so downstream synthesis does not naively rely on 25.25. Subagent also noted absence of a ## Wiki pages this source may touch section — intentional given the pet-food scope classification (paper does not contribute to any human-food product, ingredient, or regulation page), no change applied.

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.