Application of principal component analysis in the pollution assessment with heavy metals of vegetable food chain in the old mining areas

Gergen and Harmanescu analyze metal contamination patterns across two historic Romanian mining areas and a reference area. Their PCA study reproduces vegetable concentrations and male hazard quotients from the earlier Harmanescu 2011 study; these remain linked secondary observations rather than additional independent samples.

Key numbers

Table 3 fresh-matter means, mg/kg; R=Ruschita, M=Moldova Noua, Ref=Borlova:

Vegetable/area	Code	Fe	Mn	Zn	Cu	Ni	Cd	Pb
Root parsley R	R1	221.80	37.83	45.82	6.88	1.28	0.20	15.78
Root parsley M	M1	99.02	8.10	7.74	6.66	0.49	0.04	0.66
Root parsley Ref	Ref1	48.78	3.27	3.52	1.22	0.19	0.01	0.08
Root carrot R	R2	29.97	3.06	4.93	1.77	0.18	0.08	2.11
Root carrot M	M2	31.89	2.22	3.18	1.54	0.08	0.03	0.09
Root carrot Ref	Ref2	17.31	1.43	2.05	0.74	0.04	0.01	0.06
Onion R	R3	15.26	4.07	10.91	1.37	0.21	0.06	0.50
Onion M	M3	1.56	0.32	0.78	0.25	0.01	0.01	0.03
Onion Ref	Ref3	4.65	1.34	2.01	0.43	0.03	0.01	0.04
Leaf parsley R	R4	73.17	5.88	9.39	1.77	0.30	0.09	1.97
Leaf parsley M	M4	106.75	9.72	10.44	4.79	1.87	0.05	0.50
Leaf parsley Ref	Ref4	104.03	7.32	9.13	1.03	0.38	0.03	0.28
Leaf carrot R	R5	14.25	1.38	1.36	0.29	0.11	0.01	0.20
Leaf carrot M	M5	51.45	5.57	5.61	2.12	0.38	0.04	0.12
Leaf carrot Ref	Ref5	31.59	3.13	3.77	0.40	0.17	0.01	0.03
Cabbage R	R6	60.11	10.47	16.30	1.36	0.70	0.12	0.90
Cabbage M	M6	31.53	9.15	8.51	2.77	0.33	0.06	0.25
Cabbage Ref	Ref6	16.06	3.85	3.28	0.45	0.13	0.01	0.05
Lettuce R	R7	35.88	8.33	14.46	1.86	0.28	0.09	0.62
Lettuce M	M7	13.60	4.12	5.14	2.22	0.18	0.09	0.21
Lettuce Ref	Ref7	16.90	3.46	5.32	0.76	0.10	0.02	0.08
Cucumber R	R8	2.39	6.73	8.96	0.98	0.28	0.13	0.72
Cucumber M	M8	2.38	6.01	1.39	2.42	0.54	0.15	0.37
Cucumber Ref	Ref8	1.70	0.45	0.95	0.45	0.23	0.03	0.16
Green Bean R	R9	58.85	7.06	13.22	1.05	0.49	0.06	0.35
Green Bean M	M9	21.65	4.77	10.17	1.45	0.52	0.07	0.19
Green Bean Ref	Ref9	26.19	6.45	9.68	0.29	0.22	0.03	0.07

All 27 soil mean±SD rows (Table 1) and 27 male THQ rows (Table 4) are separately preserved in structured evidence. A four-metal soil PCA explains 98.6% of variance; the plant model including soil properties explains 84.4%.

Methods (brief)

Subsistence farms in Ruschita and Moldova Noua mining areas and Borlova reference. Nine soil composites per area, each 10 grabs, 0–20 cm. Soil aqua-regia extraction; plant ash at 550°C 8 h dissolved in 0.5 N nitric acid. FAAS, certified reference materials; recoveries 92–105%, variation coefficients<10%. PCA on logged values using PAST. Detailed sampling and vegetable data refer to prior study 61; the preserved paper does not give collection dates.

Implications

Supports legacy mining and tissue/region differences. Parsley root and leaf remain distinct, as do carrot root and leaf. The fresh-matter food table is retained with cohort linkage so that a reanalysis cannot inflate the evidence pool. Source THQ rankings are model results, not clinical outcomes.

Evidence Fitness: EF-2 for qualified region/tissue summaries and soil measurements; EF-3 for secondary hazard models. Prior-cohort duplication, LOD context and missing dates must be resolved for pooling.

Wiki pages this source may touch

Verification notes

Broad vegetables matrix is a vocabulary extension; structured rows specify tissue. Table 3 explicitly identifies secondary 2011 data; Table 4 also says compiled from 2011. Several rounded vegetable entries are below the paper’s listed analyticalLOD; retain printed values with qualification. No source safety ranking is adopted. SupplementaryPCA plots are referenced but not embedded; source plot loadings are model context, not additional food measurements.

The preserved Library PDF is identified by RPHMI_17; SHA-256 103e0af2d12590d269ed680134230440ae2a03862c362826f843aad176eac228. The byline, year and publication were verified in the PDF. DOI provenance and any publication-date differences are described above. Original access location: http://link.springer.com/content/pdf/10.1186/1752-153X-6-156.pdf. Source-specific numeric observations, context dispositions, row-fit decisions and remaining gaps are preserved in the structured evidence record. No source value was converted across product, soil, water or experimental matrices during ingest.

Update history

No substantive edit history is available in this build. The full commit record is available in git.