Vella et al. 2024 — Heavy metals in Maltese oil-olive cultivars and mill waste via FAAS

This study measured iron, copper, cobalt, nickel, zinc, cadmium, and strontium in four olive cultivars (Carolea, Cipressina, Leccino, Bidni) grown in Malta and sampled from a single cultivator’s October 2023 harvest, using Flame Atomic Absorption Spectrometry (FAAS) with the method of standard additions. Analytes were measured across four fractions — skin, pit, solid waste, and liquid waste from a two-phase continuous-centrifugation mill — to characterise distribution within the olive and its processing byproducts. The authors emphasise that the olives sampled were grown for olive-oil production, not as table olives, which constrains the direct applicability of EU Regulation 2023/915 (a table-olive cadmium limit of 0.02 mg/kg) and the Codex Alimentarius vegetable cadmium limit of 0.2 mg/kg. The cadmium finding is methodologically compromised: FAAS was insufficiently sensitive at the low concentrations regulated for table olives, producing standard errors equal to or larger than the measured values (0.2–1.1 mg/kg SE on 0.4–2.1 mg/kg means), so no defensible conclusion on regulatory compliance could be drawn for cadmium. For all other metals, results were treated as valid; statistical testing (Shapiro–Wilk, one-way ANOVA, Kruskal–Wallis at α = 0.05) found no significant difference in concentration between cultivars but significant differences between fractions for several metals, particularly between the liquid waste fraction and the solid fractions.

Key numbers

All concentrations reported as mean ± SE in ppm (mg/kg) on dry weight for the solid fractions (after freeze-drying) and as ppm on a per-volume basis for the liquid waste (15 mL aliquots digested). Table 4 study-overall averages across all four cultivars and all four media:

Skin fraction by cultivar (Table 3), the edible-fraction values most relevant for human exposure:

• Carolea skin: Fe 30.0 ± 3.3, Cu 14.9 ± 4.1, Co 6.9 ± 2.7, Ni 5.7 ± 3.0, Zn 22.9 ± 4.7, Cd 1.2 ± 0.4, Sr 6.5 ± 2.7 ppm
• Leccino skin: Fe 3.7 ± 2.6, Cu 15.0 ± 2.0, Co 5.5 ± 2.8, Ni 5.2 ± 2.8, Zn 25.2 ± 4.8, Cd 1.7 ± 1.0, Sr 0.8 ± 0.7 ppm
• Cipressina skin: Fe 5.2 ± 2.8, Cu 51.0 ± 1.5, Co 5.7 ± 2.8, Ni 12.8 ± 3.7, Zn 21.2 ± 4.2, Cd 1.8 ± 1.0, Sr 1.9 ± 0.7 ppm
• Bidni skin: Fe 1.3 ± 0.6, Cu 12.7 ± 1.3, Co 4.8 ± 2.6, Ni 5.3 ± 3.0, Zn 23.3 ± 5.0, Cd 1.6 ± 0.9, Sr 1.6 ± 1.4 ppm

Pit fraction by cultivar (Table 3):

• Carolea pit: Fe 24.1 ± 2.9, Cu 8.9 ± 3.9, Co 5.4 ± 3.5, Ni 6.4 ± 1.6, Zn 15.8 ± 2.7, Cd 1.4 ± 0.5, Sr 9.4 ± 1.7 ppm
• Leccino pit: Fe 6.4 ± 2.4, Cu 10.4 ± 2.5, Co 6.4 ± 2.7, Ni 5.9 ± 2.8, Zn 27.5 ± 8.0, Cd 1.2 ± 0.7, Sr 1.3 ± 0.9 ppm
• Cipressina pit: Fe 4.4 ± 2.6, Cu 10.8 ± 3.2, Co 7.5 ± 3.3, Ni 2.5 ± 1.3, Zn 24.3 ± 6.5, Cd 1.3 ± 0.6, Sr 3.1 ± 1.9 ppm
• Bidni pit: Fe 5.0 ± 4.1, Cu 11.0 ± 3.4, Co 4.5 ± 2.4, Ni 4.7 ± 2.7, Zn 17.0 ± 3.9, Cd 0.8 ± 1.1*, Sr 2.9 ± 2.4 ppm (* SE > value; reading rejected)

Liquid waste by cultivar (Table 3) — consistently the lowest-concentration fraction except for one anomalous Cipressina zinc reading:

• Carolea liquid waste: Fe 3.0 ± 1.5, Cu 2.0 ± 0.9, Co 2.0 ± 1.1, Ni 2.4 ± 1.1, Zn 8.5 ± 2.1, Cd 0.5 ± 0.3, Sr 0.7 ± 0.3 ppm
• Leccino liquid waste: Fe 1.6 ± 1.0, Cu 1.9 ± 0.6, Co 1.4 ± 0.8, Ni 1.6 ± 0.9, Zn 6.3 ± 0.6, Cd 0.7 ± 0.4, Sr 0.8 ± 0.5 ppm
• Cipressina liquid waste: Fe 0.8 ± 0.2, Cu 2.3 ± 0.7, Co 1.6 ± 0.8, Ni 2.2 ± 0.9, Zn 62.6 ± 8.7 (anomalous; authors flag as possible cleaning/contamination artefact), Cd 0.4 ± 0.2, Sr 0.9 ± 0.6 ppm
• Bidni liquid waste: Fe 0.4 ± 0.2, Cu 2.5 ± 1.0, Co 1.0 ± 0.9, Ni 1.8 ± 0.9, Zn 6.3 ± 1.6, Cd 0.7 ± 0.4, Sr 0.3 ± 0.2 ppm

Note on cadmium reliability: all cadmium results carry SE of 0.2–1.1 mg/kg, equal to or larger than both the EU table-olive limit (0.02 mg/kg, Regulation 2023/915) and the Codex Alimentarius vegetable limit (0.2 mg/kg), making regulatory comparisons impossible with this method. The Bidni pit cadmium reading (0.8 ± 1.1 ppm) was explicitly rejected by the authors as the SE exceeded the value.

Statistical results: Shapiro–Wilk normality test (Table 5) showed Fe, Cu, Co, Zn, Sr non-parametric (p < 0.05) and Ni (p = 0.093) and Cd (p = 0.459) parametric. Kruskal–Wallis and one-way ANOVA between media (Table 6, Skin–Pit through SW–LW) found significant differences in heavy-metal concentration between liquid waste and the solid fractions for most metals, with cadmium showing significant differences across nearly all media pairs. Between cultivars (Table 7), no significant differences were found for any metal: H0 was retained in all cases.

Comparison with prior literature (Table 4): the FAAS values from this study were generally higher than ICP-MS values reported by Sahan et al. 2007 (Bursa, Turkey) for Fe, Cu, Co, Ni, Cd; broadly comparable to ICP-OES averages reported by Wilson & Pyatt 2007 (Cyprus) for Cu and Zn; and substantially lower than XRF values reported by Nedjimi 2020 for Fe and Cu. The authors attribute the spread to differing cultivars, cultivation practices, and analytical methods rather than instrument bias per se, since the method of standard additions corrects for matrix effects.

Methods

Sampling: olives and mill waste from a single Maltese cultivator, harvested in October at ripe stage (the recommended harvest window of late August to late November). Four cultivar types (Carolea, Cipressina, Leccino, Bidni). The cultivator used a two-phase continuous-centrifugation mill, which produces a thick sludge containing skin and pit fragments as the solid waste and a thin water-dominated liquid waste; this mill geometry concentrates metals into the solid waste fraction. The mill was run briefly to flush any prior residue before sample collection.

Sample masses (Table 2): solid fractions digested at ~5.0 g (skin, pit, solid waste each ~5.00–5.05 g per cultivar); liquid waste digested at 15 mL.

Pre-treatment: glassware acid-washed for 24 hours in 3:1:1 HCl:HNO3:H2O. Solid samples manually split into skin and pit, filtered to separate solid from liquid mill waste, frozen, then freeze-dried for 72 hours and pulverised to a fine powder. Liquid waste was refrigerated and analysed as-is without freeze-drying.

Digestion: hot-plate aqua regia (37% HCl + 69.5% HNO3, 3:1 v/v). Solid samples digested in 50 mL aqua regia in a 250 mL conical flask; skin and liquid waste digested ~1 hour; pit and solid waste required ~2 hours with an additional 30 mL aqua regia added partway. Digestion vessels covered with watch glasses.

Standard additions: a seven-point calibration curve per element per medium per cultivar (112 total calibration curves), using zero-addition and six additions ranging from 0.25 to 12.50 ppm final concentration via a 100 ppm working standard in aqua regia. Calibration curves were extrapolated to y = 0 to obtain analyte concentration; r² ≥ 0.990 was the acceptance criterion (Zn and Cd curves required dropping the two highest standards to reach this in some cultivars). The analyte concentration was multiplied by the 100× dilution factor (100 mL volumetric / sample mass or volume) to obtain reported ppm.

Instrument: continuous-flame AAS, air–acetylene flame for all elements, burner height 7 mm, burner lateral and angle 0, acetylene flow 1.6–2.2 L/min, air-support flow 15.0 L/min. Lamp BGC-D2 mode for all elements except Sr (NON-BGC). Element-specific wavelengths (nm): Cu 324.8, Fe 248.3, Co 240.6, Ni 231.9, Zn 213.8, Cd 228.8, Sr 460.6.

Standard error per reading was calculated from the standard-additions calibration as sxE = (sy/x / m) × √(1/nc + ȳ² / (m² Σ(xi - x̄)²)), where sy/x is the standard error of the y-estimate, m is the calibration gradient, and nc is the number of calibration points.

Statistics: α = 0.05 throughout. Shapiro–Wilk for normality; one-way ANOVA for parametric data (Ni, Cd); Kruskal–Wallis independent-samples for non-parametric data (Fe, Cu, Co, Zn, Sr). Two paired hypothesis tests: media-pairwise (Skin-Pit through SW-LW) and cultivar-pairwise (Carolea-Leccino through Cipressina-Bidni).

Key methodological limitation: FAAS is designed for ppm-range detection. Cadmium in table olives is regulated at 0.02 mg/kg (EU) to 0.2 mg/kg (Codex vegetables), well below FAAS sensitivity. Authors recommend GFAAS, ICP-MS, or ICP-OES with microwave aqua regia + HF digestion (Chen & Ma 2001) for cadmium quantification at these concentrations.

Evidence tier assigned as B due to: single-cultivator sampling (no geographic diversity within Malta, no inter-cultivator variability), small number of distinct biological samples (one tree-row equivalent per cultivar), journal is a practitioner/student-oriented applied-research publication from a vocational college rather than a primary peer-reviewed food-science journal, and the cadmium data — the most policy-relevant analyte for olives — is methodologically unusable for regulatory comparison.

Implications

Certification: the olives sampled were grown for oil production, not as table olives, so the values do not bear directly on table-olive cadmium thresholds, and no olive oil was analysed in this study. The skin-fraction Ni values (5–13 ppm dry weight, with Cipressina notably elevated at 12.8 ppm) and the Cu values (with Cipressina skin at 51.0 ppm, ~3× the other cultivars) are informative as context for olive-oil precursor material. The study underlines the method-selection problem: FAAS is unsuitable for regulatory cadmium monitoring in olive material; ICP-MS or GFAAS is required.

Courses: useful illustration of method-selection consequences — a study can produce valid calibration curves and still fail to answer the regulatory question because the analytical technique’s LOQ is above the regulatory limit of interest. The Bidni-pit cadmium rejection (SE > mean) is a clean teaching example of when to discard a measurement on internal-consistency grounds rather than report it.

App: not suitable as a primary data source for contamination_profile values due to B-tier evidence, single-cultivator sampling, and compromised cadmium data. Values may inform a broad olive-fruit context band but should not anchor a typical_ppb or p95_ppb estimate.

Verification notes

Page was first ingested under the legacy raw_handle: manual-fetch-kimi and last updated: 2026-05-14. Merge-enhanced on 2026-05-28 via the v2 autonomous manual-fetch ingest skill against the source PDF. Changes:

• Critical data correction: the previously-published “Cipressina skin” and “Leccino skin” rows in Key numbers were transposed. Verified against Table 3 (page 11 of the PDF): row ordering in the table is Carolea → Leccino → Cipressina → Bidni. The values previously labelled “Cipressina skin” (Fe 3.7 ± 2.6, Cu 15.0 ± 2.0, Sr 0.8 ± 0.7) are in fact Leccino skin, and the values previously labelled “Leccino skin” (Fe 5.2 ± 2.8, Cu 51.0 ± 1.5, Ni 12.8 ± 3.7, Sr 1.9 ± 0.7) are in fact Cipressina skin. Rows have been corrected to match the source table.
• Removed table-olives from matrices because the source explicitly states “the olives tested in this study weren’t table olives, but olives which are used in the production of olive oil” (page 9) and “no olive oil was tested in this study” (page 16). Page title updated to reflect “oil-olive cultivars” to prevent the same misclassification on re-read.
• Changed products from non-root-vegetables to olive-oil to reflect the actual product class of the sampled material (oil-production olives). The skin/pit fractions are precursors entering the olive-oil supply chain, not vegetable-category food.
• Replaced legacy raw_handle: manual-fetch-kimi with the canonical MFK_investigation-on-the-concentration-of-heavy-metals handle derived from the PDF filename.
• Corrected raw_path filename from the truncated “Investigation on the Concentration of Heavy Metals found in Olive.pdf” to the actual on-disk name “Investigation on the Concentration of Heavy Metals found in Olives and Olive Mill Waste.pdf”.
• Added the journal volume/issue (“Vol. 8, Issue 2”) to the publication field.
• Expanded Key numbers to include pit and liquid-waste rows by cultivar (previously only skin was published, and the Bidni-pit cadmium rejection asterisk was not surfaced).
• Expanded Methods to include sample masses (Table 2), instrument-specific operating parameters (burner geometry, gas flows, lamp modes), the explicit standard-error calculation equation, and the statistical-testing pipeline (Shapiro–Wilk → ANOVA or Kruskal–Wallis at α = 0.05 with paired-hypothesis structure).
• Added Statistical results paragraph distinguishing media-pairwise from cultivar-pairwise outcomes (no cultivar-pairwise significance for any metal; significant media-pairwise differences concentrated on the SW–LW and Pit–LW boundaries, with Cd showing significance across nearly all media pairs).
• Added comparison-with-prior-literature paragraph summarising Table 4.

Preserved unchanged: cite_key, doi: null / no_doi_assigned: true, license: unknown, near_duplicates: [], evidence tier B, sample_n: 4, jurisdictions: [MT], and the metals list (Sr was reported in the source but is not in the wiki metals taxonomy, so it is mentioned in Key numbers without being declared in frontmatter).

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.