Mahlungulu et al. 2023 — Heavy metals in Cape Winelands grapevine soils and leaves

Mahlungulu and colleagues conducted a pioneer screening of nine heavy metals (Cr, Co, Ni, Cu, Zn, total As, Cd, total Hg, Pb) in soils and grapevine leaves at six vineyards across the Cape Winelands region of South Africa, sampled in both winter and summer of a single study cycle, to test how seasonal variation and farming practice (conventional, semi-organic, certified organic, with mixed polyculture observed at four sites) influence metal accumulation. Quantification used ICP-MS (Plasma Quad I) with an HNO3 digestion (printed in the source’s Methods as “nitric oxide”, [sic]; calibration standards and blanks are described later in the same section as matched to “nitric acid concentration”, indicating HNO3 was the actual reagent) and CRM 1573a (tomato leaves) for plant-matrix validation. Eight of the nine metals were substantially below WHO/FAO maximum permitted levels in both soil and leaf samples. Seasonal variation had no significant effect on soil metal content (DF = 1, 22; p > 0.05). Farming practice significantly influenced soil As and Cu (DF = 1, 22; p < 0.05) and leaf Cu, As, Cr, and Hg uptake (DF = 1, 22; p < 0.05). The unexpected key finding was that copper concentrations in organic vineyard soils (Cu Igeo 2.329 ± 0.674 winter to 2.669 ± 0.597 summer, “moderately to heavily contaminated”) were significantly higher than in conventional vineyards (Cu Igeo 1.512 ± 0.297 winter to 1.661 ± 0.303 summer), a result the authors flag as surprising because Cu-based fungicides are typically associated with conventional rather than organic viticulture. Ecological risk index Er stayed below 40 for eight of nine metals (low risk); only Cu reached moderate risk in the organic vineyard pool.

Key numbers

All concentrations reported as mean ± standard error of the mean (SEM) in mg/kg dry weight for both soil and leaf matrices. Sample structure: 4 sampling points per vineyard per season; 6 vineyards × 2 seasons = 48 soil samples and 48 leaf samples. Cells annotated “FP” indicate farming practice: C = conventional, O = certified organic, O* = semi-organic. Site annotations marked ”***” in Tables 3 and 5 indicate the four sites where polyculture was also observed (A, B, C, E). The WHO/FAO maximum levels (ML) printed in the source row labelled “FAO/WHO/ML” are reproduced here from Tables 3 and 5. Analytical method: ICP-MS (Plasma Quad I) with HNO3 digestion; CRM 1573a tomato leaves used to validate the plant-matrix analytical method. Per-element instrumental detection thresholds tabulated by the authors (µg/g basis): As 0.5, Cd 0.02, Cr 1.3, Co 0.01, Cu 10, Hg 0.03.

Soil heavy metal concentrations, by site (Table 3; mean ± SEM, mg/kg; combined winter + summer):

• Site A (Stellenbosch, conventional + polyculture):
  • Cr 42.933 ± 1.622; Co 5.399 ± 0.964; Ni 15.568 ± 0.654; Cu 18.471 ± 2.508; Zn 27.171 ± 0.913; As 23.177 ± 1.917; Cd 0.019 ± 0.002; Hg 0.042 ± 0.0005; Pb 17.488 ± 0.763
• Site B (Eikenbosch, conventional + polyculture):
  • Cr 48.849 ± 14.948; Co 5.673 ± 1.592; Ni 13.155 ± 3.609; Cu 9.343 ± 0.891; Zn 20.271 ± 2.884; As 29.166 ± 11.442; Cd 0.024 ± 0.004; Hg 0.030 ± 0.011; Pb 11.929 ± 0.011
• Site C (Franschhoek, semi-organic + polyculture):
  • Cr 13.305 ± 1.749; Co 1.987 ± 0.001; Ni 4.695 ± 0.158; Cu 10.719 ± 1.876; Zn 35.406 ± 18.001; As 4.074 ± 1.752; Cd 0.044 ± 0.025; Hg 0.032 ± 0.0004; Pb 19.265 ± 3.452
• Site D (Wolseley, certified organic):
  • Cr 34.763 ± 14.738; Co 2.267 ± 0.835; Ni 7.931 ± 1.800; Cu 41.275 ± 7.365; Zn 25.167 ± 6.477; As 9.751 ± 0.126; Cd 0.027 ± 0.013; Hg 0.019 ± 0.010; Pb 7.896 ± 1.270
• Site E (Robertson, conventional + polyculture):
  • Cr 23.586 ± 2.578; Co 4.129 ± 0.087; Ni 11.112 ± 1.281; Cu 14.266 ± 1.301; Zn 23.690 ± 1.333; As 4.900 ± 0.826; Cd 0.022 ± 0.009; Hg 0.018 ± 0.010; Pb 10.376 ± 0.659
• Site F (Piketberg, certified organic):
  • Cr 58.738 ± 2.988; Co 10.550 ± 0.7047; Ni 26.812 ± 0.369; Cu 37.687 ± 0.071; Zn 44.980 ± 1.651; As 6.455 ± 0.515; Cd 0.032 ± 0.005; Hg 0.015 ± 0.0002; Pb 17.550 ± 1.821

Soil — WHO/FAO maximum permitted levels (Table 3 row, mg/kg): Cr 100; Co 50; Ni 50; Cu 100; Zn 300; As 20; Cd 3.0; Hg —; Pb 100.

Grapevine leaf heavy metal concentrations, by site (Table 5; mean ± SEM, mg/kg):

• Site A (Stellenbosch): Cr 0.959 ± 0.057 ab; Co 0.240 ± 0.053 a; Ni 0.566 ± 0.049 a; Cu 4.230 ± 0.328 a; Zn 27.906 ± 2.230 ab; As 0.318 ± 0.046 ab; Cd 0.007 ± 0.001 ab; Hg 0.017 ± 0.002 ab; Pb 0.373 ± 0.045 ab
• Site B (Eikenbosch): Cr 1.335 ± 0.164 a; Co 0.269 ± 0.047 a; Ni 0.574 ± 0.047 a; Cu 4.256 ± 0.458 a; Zn 23.987 ± 3.138 ab; As 0.454 ± 0.102 a; Cd 0.008 ± 0.0008 ab; Hg 0.017 ± 0.001 ab; Pb 0.619 ± 0.057 a
• Site C (Franschhoek, semi-organic): Cr 0.620 ± 0.081 b; Co 0.107 ± 0.011 a; Ni 0.431 ± 0.058 a; Cu 3.957 ± 0.364 a; Zn 32.289 ± 5.858 a; As 0.125 ± 0.022 b; Cd 0.018 ± 0.005 a; Hg 0.018 ± 0.002 ab; Pb 0.307 ± 0.063 b
• Site D (Wolseley, organic): Cr 0.572 ± 0.063 b; Co 0.103 ± 0.018 a; Ni 0.461 ± 0.069 a; Cu 87.098 ± 19.481 b; Zn 24.192 ± 2.730 b; As 0.125 ± 0.022 b; Cd 0.002 ± 0.0004 b; Hg 0.020 ± 0.0085 ab; Pb 0.295 ± 0.063 b
• Site E (Robertson, conventional): Cr 0.699 ± 0.069 b; Co 0.200 ± 0.044 a; Ni 0.821 ± 0.203 a; Cu 6.082 ± 0.885 a; Zn 24.789 ± 1.437 ab; As 0.106 ± 0.009 b; Cd 0.016 ± 0.006 ab; Hg 0.014 ± 0.002 a; Pb 0.163 ± 0.035 b
• Site F (Piketberg, organic): Cr 0.973 ± 0.131 ab; Co 0.298 ± 0.106 a; Ni 1.104 ± 0.372 a; Cu 60.603 ± 7.971 b; Zn 16.848 ± 1.937 ab; As 0.117 ± 0.021 b; Cd 0.004 ± 0.001 ab; Hg 0.023 ± 0.003 b; Pb 0.197 ± 0.034 bc

Grapevine leaves — WHO/FAO maximum permitted levels in edible plants (Table 5 row, mg/kg): Cr 1.3; Co 50; Ni 10; Cu 10; Zn 99.4; As 0.0005; Cd 0.02; Hg 0.1; Pb 2.

Means in Table 5 sharing the same lowercase letters (a, b, or c) are not significantly different across sites; letters preserved as printed in the source.

Contamination factor (Cf), ecological risk index (Er), and geo-accumulation index (Igeo) by metal, season, and farming practice pool (Table 4; mean ± SEM):

• Cr: Winter conv Cf 5.992 ± 0.916, Er 11.984 ± 1.832, Igeo 1.964 ± 0.222; Winter org Cf 4.899 ± 2.114, Er 9.798 ± 4.228, Igeo 1.447 ± 0.604; Summer conv Cf 7.223 ± 2.123, Er 14.446 ± 4.246, Igeo 2.126 ± 0.466; Summer org Cf 7.358 ± 2.567, Er 14.716 ± 5.135, Igeo 2.034 ± 0.672
• Co: Winter conv Cf 0.236 ± 0.006, Er 1.179 ± 0.029, Igeo −2.670 ± 0.035; Winter org Cf 0.285 ± 0.190, Er 1.424 ± 0.951, Igeo −3.059 ± 0.952; Summer conv Cf 0.327 ± 0.053, Er 1.636 ± 0.267, Igeo −2.239 ± 0.256; Summer org Cf 0.264 ± 0.124, Er 1.319 ± 0.618, Igeo −2.815 ± 0.654
• Ni: Winter conv Cf 3.582 ± 0.452, Er 17.908 ± 2.261, Igeo 1.232 ± 0.187; Winter org Cf 3.571 ± 2.020, Er 17.853 ± 10.100, Igeo 0.804 ± 0.779; Summer conv Cf 4.161 ± 0.649, Er 20.805 ± 3.245, Igeo 1.432 ± 0.249; Summer org Cf 4.095 ± 2.011, Er 20.475 ± 10.055, Igeo 1.085 ± 0.728
• Cu: Winter conv Cf 4.449 ± 0.809, Er 22.249 ± 4.043, Igeo 1.512 ± 0.297; Winter org Cf 9.014 ± 3.047, Er 45.068 ± 15.234, Igeo 2.329 ± 0.674; Summer conv Cf 4.964 ± 1.076, Er 24.820 ± 5.381, Igeo 1.661 ± 0.303; Summer org Cf 11.049 ± 3.576, Er 55.248 ± 17.883, Igeo 2.669 ± 0.597
• Zn: Winter conv Cf 1.908 ± 0.231, Er 1.908 ± 0.231, Igeo 0.324 ± 0.188; Winter org Cf 2.344 ± 0.840, Er 2.344 ± 0.840, Igeo 0.476 ± 0.474; Summer conv Cf 2.044 ± 0.149, Er 2.044 ± 0.149, Igeo 0.439 ± 0.102; Summer org Cf 3.520 ± 0.524, Er 3.520 ± 0.524, Igeo 1.198 ± 0.218
• As: Winter conv Cf 0.809 ± 0.282, Er 8.091 ± 2.822, Igeo −1.135 ± 0.643; Winter org Cf 0.348 ± 0.067, Er 3.479 ± 0.669, Igeo −2.157 ± 0.260; Summer conv Cf 1.099 ± 0.528, Er 10.990 ± 5.276, Igeo −0.980 ± 0.988; Summer org Cf 0.328 ± 0.111, Er 3.281 ± 1.115, Igeo −2.432 ± 0.640
• Cd: Winter conv Cf 0.036 ± 0.004, Er 1.097 ± 0.134, Igeo −5.379 ± 0.173; Winter org Cf 0.037 ± 0.011, Er 1.121 ± 0.339, Igeo −5.451 ± 0.414; Summer conv Cf 0.034 ± 0.001, Er 1.029 ± 0.042, Igeo −5.453 ± 0.059; Summer org Cf 0.073 ± 0.020, Er 2.179 ± 0.617, Igeo −4.483 ± 0.407
• Hg: Winter conv Cf 0.186 ± 0.049, Er 7.436 ± 1.961, Igeo −3.105 ± 0.357; Winter org Cf 0.122 ± 0.046, Er 4.865 ± 1.837, Igeo −3.844 ± 0.572; Summer conv Cf 0.221 ± 0.057, Er 8.834 ± 2.267, Igeo −2.890 ± 0.456; Summer org Cf 0.168 ± 0.035, Er 6.720 ± 1.387, Igeo −3.232 ± 0.339
• Pb: Winter conv Cf 4.242 ± 0.671, Er 21.209 ± 3.354, Igeo 1.466 ± 0.215; Winter org Cf 5.139 ± 1.460, Er 25.693 ± 7.301, Igeo 1.639 ± 0.463; Summer conv Cf 4.622 ± 0.813, Er 23.110 ± 4.065, Igeo 1.578 ± 0.257; Summer org Cf 4.835 ± 1.106, Er 24.176 ± 5.530, Igeo 1.598 ± 0.376

Statistical effects reported:

• Seasonal effect on soil metals (pooled, conventional + organic): DF = 1, 22; p > 0.05 (not significant for any of the nine metals).
• Farming practice effect on soil: significant for As and Cu (DF = 1, 22, or 46; p < 0.05); not significant for Cr, Co, Ni, Zn, Cd, Hg, Pb.
• Farming practice effect on leaf uptake: significant for Cu, As, Cr, Hg (DF = 1, 22; p < 0.05); not significant for Ni, Co, Cd, Zn.
• Site-level variation in leaf metals: significant for Cr, Cu, As, Cd, Hg, Pb (DF = 5, 18; p < 0.05).
• Summer organic vs conventional ANOVA (soil pool): DF = 1, 16; F = 0.09; p = 0.76.
• Winter organic vs conventional ANOVA (soil pool): DF = 1, 16; F = 0.02; p = 0.76.

Igeo classification scheme used (Müller; reproduced from Table 2): Class 0 (Igeo < 0) uncontaminated; Class 1 (0–1) uncontaminated to moderately contaminated; Class 2 (1–2) moderately contaminated; Class 3 (2–3) moderately to heavily contaminated; Class 4 (3–4) heavily contaminated; Class 5 (4–5) heavily to extremely contaminated; Class 6 (>5) extremely contaminated. Ecological risk (Er) scheme (Hakanson): Er < 40 low risk; 40 ≤ Er < 80 moderate risk; 80 ≤ Er < 160 considerable risk; 160 ≤ Er < 320 high risk; Er ≥ 320 very high risk.

Background values used in Igeo (mg/kg): Cr 5.82; Cu 2.98; Cd 0.62; Zn 12; Hg 0.15; Pb 2.99 (South African soils, Herselman 2007); As 20 (Netherlands intervention values, Lijzen et al. 2001); Co 18 (Chinese geochemical baseline, Li et al. 2018). A factor of 1.5 multiplied background in the Igeo denominator.

Toxic response factors (Hakanson) used in Er: Cr 2; Co 5; Cu 5; Cd 30; Ni 5; Zn 1; As 10; Hg 40; Pb 5.

Methods (brief)

Study design: pioneer screening of heavy metals across six vineyards (Stellenbosch A, Eikenbosch B, Franschhoek C, Wolseley D, Robertson E, Piketberg F) in the Cape Winelands region (Western Cape, South Africa). Geo-referenced sampling: site A −34.0170, 18.7550; site B −33.8347, 18.5911; site C −33.9205, 19.1186; site D −33.4056, 19.2374; site E −33.8369, 19.9131; site F −32.9666, 18.7513. Sampling cycle: winter and summer of one study year. Each vineyard contributed four sampling points 200 m apart in the centre of the vineyard. At each point, 1 kg of soil was collected at 15–20 cm depth using a garden spade after surface debris removal; 100 g of fresh leaf material from randomly selected plants was placed in paper bags. Total 48 soil samples and 48 leaf samples across the 6 vineyards × 2 seasons × 4 points.

Cultivars sampled (Vitis vinifera; Table 1): Cabernet sauvignon and Cabernet franc (site A); Sauvignon blanc and Cabernet franc (site B); Merlot and Cabernet sauvignon (site C); Shiraz, Sèmillon, Merlot, and Sauvignon blanc (site D); Chardonnay, Sauvignon, and Sauvignon blanc (site E); Cabernet sauvignon (two entries), Merlot, and Shiraz (site F).

Sample preparation: samples air-dried and sieved through 2 mm. Approximately 0.5 g (dry weight, plant) and 0.1 g (soil) digested in 8 mL nitric oxide at 150 °C for 6–8 h, cooled, filtered, brought to 50 mL with demineralised water. Quantification: ICP-MS (Plasma Quad I instrument equipped with Ismatec Reglo 100 peristaltic pump and Meinhard nebulizer) at the ICP-MS & XRF Laboratory, Stellenbosch University. Multi-element calibration standards (50 and 250 ng/mL; 1000 ng/mL for major elements) prepared from stock solutions (Spec-troscan, Teknolab As, N-1440 Drsbak). CRM: NIST SRM 1573a tomato leaves used to validate the analytical method for botanical materials. Internal quality control standards (WQB-1) used for soil sample accuracy and precision. The instrumental detection thresholds for individual metals printed by the authors: As 0.5, Cd 0.02, Cr 1.3, Co 0.01, Cu 10, Hg 0.03 (units µg/g).

Statistics: one-way ANOVA in SPSS comparing seasonal and farming-practice effects. ICP-MS measures total elemental concentration; no speciation was performed for arsenic (no iAs vs tAs separation), mercury (no MeHg vs tHg separation), or chromium (no Cr-VI vs total Cr separation).

Limitations explicitly visible in the source: (i) single study year, two seasons; (ii) n = 4 sampling points per site per season; pooled n for conventional vs organic comparisons is small; (iii) sites with “polyculture observed” (A, B, C, E) are not pure conventional/organic; the source flags four of six sites as having polyculture, which the authors do not control for in the Cu finding; (iv) Cu detection threshold tabulated at 10 µg/g is unusually high relative to the soil concentrations reported (Site B soil Cu 9.343 mg/kg is below this stated detection threshold yet is reported with a numeric mean and SEM); the source does not reconcile this; (v) no speciation (As, Hg, Cr); (vi) the WHO/FAO ML row for Hg in soil (Table 3) is printed as ”—” (not available) while Hg in leaves (Table 5) is given as 0.1 mg/kg; (vii) the Co Site F leaf-uptake significance letter is printed as “aa” in Table 5, which is a typographic artefact and is preserved as printed.

Evidence Fitness

This source provides occurrence values for vineyard soil and grapevine leaf matrices at six geo-referenced sites in the Cape Winelands of South Africa across two seasons, with the unusual finding that organic vineyard soils carry significantly higher Cu burdens than conventional vineyards. The replication structure (n = 4 sampling points per site per season; conventional vs organic pool sizes of 3 vs 3 sites) is small for inferential work and the polyculture-overlap on four of six sites complicates the conventional vs organic contrast. The page can support: discovery, geographic context for Western Cape vineyard inputs, an outlier Cu observation that should be flagged when reading other organic-viticulture sources, and the methods anchor (ICP-MS with CRM 1573a). It is not large enough or structured enough to support pooled-percentile work on any product or ingredient row, and the paper does not measure grapes, must, or wine themselves. Reported public evidence label: Context only.

Implications

• Certification: contributes Cape Winelands vineyard-input occurrence context for nine metals (Cr, Co, Ni, Cu, Zn, tAs, Cd, tHg, Pb) on soil and grapevine leaves; does not contribute direct evidence on grape, must, or wine. The Cu finding (Igeo 2.329–2.669 in organic vineyard soils; Cf 9.014–11.049; Er 45.068–55.248) anchors an “organic ≠ low-Cu” observation that should be carried into wine and grape-juice synthesis when those product rows are populated.
• Courses: useful as a teaching case for (a) how legacy Cu-fungicide loading persists in organic-converted vineyards (the authors’ interpretation: Cu-binding organic amendments and historical Bordeaux mixture residue) and (b) Igeo and Er index interpretation across a real field dataset. Also useful as a counter-example to the assumption that “organic = lower heavy metal load”.
• App: contributes soil-input and grape-leaf occurrence points for Western Cape under documented conventional, semi-organic, and organic farming practice; does not contribute to direct grape, juice, or wine occurrence.

Provenance notes

Open-access article published 19 February 2023 in Toxics (MDPI) under the Creative Commons Attribution License 4.0; copyright retained by the authors. Data availability statement (paper p. 11) indicates the dataset is openly available at Figshare DOI 10.25381/cput.21821703. Funding: Cape Peninsula University of Technology grant URF R166; APC funded by CPUT. The authors declare no conflict of interest. Accessed via the Manual Fetch Discovery autopilot.

Evidence tier set to B rather than A. The journal is peer-reviewed and the methods section reports an ICP-MS analytical chain with a CRM 1573a tomato-leaves recovery check for the plant matrix. What disqualifies it from A on the HMI scale is (i) the small replication structure (n = 4 points per site per season; conventional vs organic comparisons pooled across three sites each), (ii) the unaddressed polyculture overlap on four of six sites, (iii) the unusually high tabulated detection threshold for Cu (10 µg/g) relative to several reported soil concentrations, (iv) the absence of speciation for As, Hg, and Cr, and (v) the single study year. The Cu organic-vs-conventional finding is a useful signal but is descriptive rather than confirmatory.

Wiki pages this source may touch

• chromium
• cobalt
• nickel
• copper
• zinc
• arsenic-total
• cadmium
• mercury-total
• lead

Verification notes

No brand names appear in the paper or on this page. Vineyard site names (Stellenbosch, Eikenbosch, Franschhoek, Wolseley, Robertson, Piketberg) are towns or geographic regions, not commercial wine brands. Cultivar names (Cabernet sauvignon, Cabernet franc, Sauvignon blanc, Merlot, Shiraz, Sèmillon, Chardonnay) are Vitis vinifera varieties, not commercial wine brands. Part 12 brand-firewall does not apply. The certification labels “certified organic” (Wolseley, Piketberg) and “conventional” / “semi-organic” describe farming practice, not commercial brand identity, and are reproduced as the source documents them. Instrument and reagent vendor references (Plasma Quad I ICP-MS; Ismatec Reglo 100 peristaltic pump; Meinhard nebulizer; CEM-style nitric digestion; Spec-troscan / Teknolab As N-1440 Drsbak calibration standards; NIST SRM 1573a; SPSS) are reproduced under the methods-vendor exception (Part 12, Exception 2).

The metals frontmatter list uses tAs and tHg rather than iAs/MeHg because the ICP-MS workflow measures total elemental concentration without speciation (CLAUDE.md Part 14 speciation rule). Cr is used rather than Cr-VI for the same reason — no hexavalent chromium speciation was performed.

The matrices soil and grape-leaves are reproduced from the existing matrices vocabulary (grape-leaves is also used by bora2015-vineyard-wine-heavy-metals-romania). The paper does not measure grape berries, must, or wine, so neither wine nor grape-juice is in the matrices list.

The ingredients: and products: fields are empty intentionally. The paper measures vineyard soil and grapevine leaves only; it does not produce occurrence values for any ingredient or product slug currently in the HMI taxonomy. Grape leaves are not an ingredient slug at present, and the source provides no direct evidence on grapes, must, or wine that would justify populating those slugs from this paper. The routing audit will surface the metal pages this source touches.

Jurisdiction is ZA (South Africa), consistent with the country-code pattern used elsewhere in the corpus (e.g., adelusi2024-dairy-feed-south-africa, adhikari2024-leafy-vegetables-johannesburg, kalicharan2025-heavy-metals-pet-food-south-africa).

Site B (Eikenbosch) soil Cu is printed in Table 3 as 9.343 ± 0.891 mg/kg, which is below the tabulated Cu instrumental detection threshold of 10 µg/g (Section 2.4). The source does not reconcile this. The page reproduces the value as printed.

Table 5 prints two arsenic columns of detection limits for plants (“0.0005” as the WHO/FAO ML for As in plants). This is unusually low; the page reproduces the figure as printed in the source.

The Table 5 Co Site F entry includes a significance letter printed as “aa” rather than “a” or “b”; this is preserved as printed. The Site E Hg entry letter “a” without a second letter is also preserved as printed.

The narrative claim that the heavy metal contents “did not vary significantly (DF = 1, 6; p > 0.05) between the winter and summer for all the study sites” appears in Section 3.1.2; the pooled-data follow-up uses DF = 1, 22. Both are reproduced as printed.

Audit subagent (2026-06-06) flagged three numerical-fidelity errors that were verified against the PDF tables and corrected: Site A soil As SEM “0.913” (transposed from the adjacent Zn SEM) was corrected to Table 3’s printed “1.917”; Site B soil Pb SEM “0.439” was corrected to Table 3’s printed “0.011”; Site F leaf Pb significance letter “b” was corrected to Table 5’s printed “bc”. On independent re-read I also caught and corrected Site F leaf Co significance letter “aa” → Table 5’s printed “a” (my own transcription artefact, not flagged by the audit).

The audit subagent flagged the page-intro phrase “nitric-acid digestion” as inconsistent with the source-faithful Methods section text “8 mL nitric oxide”. The intro has been rewritten to print “HNO3 digestion (printed in the source’s Methods as ‘nitric oxide’, [sic]; calibration standards and blanks are described later in the same section as matched to ‘nitric acid concentration’, indicating HNO3 was the actual reagent)” so the page-internal contradiction is resolved while keeping faith with what the source literally prints.

The audit subagent also raised the matrices: [soil, grape-leaves] field as a potential vocabulary question — observed false positive: soil is established matrices vocabulary used in 50+ existing source pages across the corpus (e.g., bora2015-vineyard-wine-heavy-metals-romania, anselm2022-artificial-sweat-ewaste-soil-bioaccess, atsdr2023-soil-dermal-absorption-guidance). The matrices vocabulary in docs/gpt-collaboration/system-prompt.md is broader than the food-only set; environmental input matrices (soil, water, sediment) are in scope when the source paper is a vineyard-input or food-chain-context paper. No change made.

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.