Saf et al. 2023 — Agroecological valorization of olive mill wastewater UF/NF fractions

Saf and colleagues fractionated raw olive mill wastewater (OMW) from a three-phase Moroccan mill using sequential ultrafiltration (150 kDa) and nanofiltration (200 Da) and tested the four resulting streams (raw OMW, UF retentate, NF retentate, NF permeate) for their agronomic value: bioherbicide potential of the polyphenol- and salt-enriched NF retentate against wild mustard and flax weed competitors, and fertigation/irrigation potential of the depleted UF retentate and NF permeate on maize. The paper characterizes each fraction for bulk parameters (pH, EC, COD, dry matter, total polyphenols, total sugars) and mineral content (K, P, Ca, Mg, Fe) but does not measure regulated heavy metal contaminants (Pb, Cd, As, Hg, Ni, Al, Cr, Sn). Its contribution to the Heavy Metal Index corpus is upstream-supply-chain context for the olive-oil agroecology system in Mediterranean producing regions, not direct contamination measurement of any HMI-tracked product.

Key numbers

OMW source and feed characterization (Table 1, p. 2):

• Raw OMW pH 4.54 ± 0.01
• Electrical conductivity 16.17 ± 0.04 mS·cm⁻¹
• Total dry matter (DM) 80.57 ± 3.77 g·L⁻¹
• Mineral matter (MM) 17.27 ± 2.35 g·L⁻¹
• COD 192.85 ± 12.16 g·L⁻¹
• Total polyphenols (TPP) 5.63 ± 0.03 g·L⁻¹
• Total sugars 10.25 ± 1.19 g·L⁻¹

Chemical composition of OMW fractions after UF (150 kDa ceramic, TiO₂-ZrO₂) and NF (200 Da spiral-wound polyamide), Table 3, p. 3 — values are mean ± SD for the feed, UF retentate (UFR), NF retentate (NFR) and NF permeate (NFP):

• Total dry matter (g·L⁻¹): Feed 123.3 ± 3.2; UFR 83.4 ± 3.9; NFR 179.2 ± 3.6; NFP 9.2 ± 0.1
• Mineral matter (g·L⁻¹): Feed 37.6 ± 0.3; UFR 23.3 ± 1.8; NFR 89.8 ± 2.8; NFP 4.2 ± 0.8
• Total sugars (g·L⁻¹): Feed 6.5 ± 0.6; UFR 7.8 ± 0.3; NFR 8.9 ± 0.1; NFP 1.26 ± 0.06
• Total polyphenols (g·L⁻¹): Feed 4.2 ± 0.2; UFR 2.4 ± 0.1; NFR 5.6 ± 0.1; NFP 1.02 ± 0.05
• K (mg·L⁻¹): Feed 6168 ± 84; UFR 6113 ± 79; NFR 13,232 ± 135; NFP 911 ± 15
• P (mg·L⁻¹): Feed 516 ± 9; UFR 510 ± 11; NFR 1114 ± 14; NFP 67 ± 5
• Ca (mg·L⁻¹): Feed 250 ± 5; UFR 328 ± 4; NFR 521 ± 7; NFP 17.5 ± 0.6
• Mg (mg·L⁻¹): Feed 191 ± 3; UFR 186 ± 2; NFR 438 ± 6; NFP 4.4 ± 0.3
• Fe (mg·L⁻¹): Feed 49 ± 1; UFR 54.4 ± 0.6; NFR 105 ± 1; NFP <1

Phytotoxicity outcomes:

• NF retentate at 1/20, 1/15 and 1/10 dilutions (polyphenol content 0.3, 0.4, 0.6 g·L⁻¹ TPP) inhibited maize root growth by 50%, 64%, 78% respectively versus distilled-water control; flax root growth was suppressed to a similar but slightly attenuated extent (Section 3.2, p. 4-5).
• NF retentate at 1/20 produced 50% root-weight inhibition in maize and 27% in flax (Fig. 3b, p. 6).
• Chloroplast-pigment loss in maize seedlings was 60%, 72%, 95% (chlorophyll a+b) for 1/20, 1/15, 1/10 NF retentate dilutions; flax pigment loss was 41%, 47%, 73% (Table 5, p. 7).
• On Sinapis arvensis (wild mustard) in vitro, NF retentate at very low dilutions (1/125, 1/100, 1/75) produced germination inhibition of 15%, 40%, 73% and root-elongation reduction of 42%, 52%, 78% relative to control (Section 3.3, p. 5; Fig. 5, p. 9).
• On flax in vitro, the same low NF-retentate dilutions had no significant effect on germination, demonstrating species-selective bioherbicide action against the weed (Fig. 5, p. 9).
• GST (glutathione-S-transferase) activity in maize roots peaked at the 1/20 NF retentate dilution; polyphenol content in plant tissue increased with NF retentate concentration in both maize and flax leaves (Fig. 4, p. 8).
• UF retentate (1/8, 1/4) and NF permeate (1/6, 1/2) applied to maize in pot culture produced no phytotoxicity; shoot length and weight at 1/8 UF retentate and 1/6 NF permeate increased 14% and 30% over Murashige-Skoog control with no significant difference in chloroplast pigments (Section 3.4, p. 5-6; Fig. 7, p. 10).

Methods (brief)

OMW was collected from a semi-modern three-phase centrifugation mill in Marrakech, Morocco. Pretreatment was isoelectric-point acidification to pH 2 with 95-98% H₂SO₄ (8 mL·L⁻¹), 3 h decantation, 5 µm porosity filtration, then NaOH (5 M) elevation to pH 6. Filtration used a stainless-steel pilot-scale UF-NF system (Firmus, France) with a 70 L feed tank: UF was a 150 kDa ceramic tubular TiO₂-ZrO₂ membrane (19 channels, 0.24 m² area, Orelis Kleansep) at 1.5 bar TMP, 30 °C, 5 m·s⁻¹ cross-flow; NF was a 200 Da spiral-wound polyamide membrane (2.6 m² area, Suez DK) at 25 bar TMP, 30 °C, 1 m·s⁻¹. Filtrations ran in concentration mode under previously optimized conditions with continuous data acquisition (Ecograph T, Endress + Hauser).

Bulk OMW characterization: dry matter by oven drying at 40 °C to constant weight; mineral matter by ashing at 550 °C for 2 h; total polyphenols by Folin-Ciocalteu method in gallic acid equivalents (Singleton et al. 1999); total sugars by the Dubois colorimetric method. Plant culture: 8 cm × 6 cm × 6.5 cm pots filled with 30 g HAWITA potting soil (blond + brown peat, volcanic clay, perlite; 40% organic matter, 0.2% N, pH 6.1), 120 mL of treatment solution per pot, 21 ± 1 °C, randomized complete block design with three replications (n=9); 10-day culture. In vitro flax and wild mustard culture used Murashige-Skoog medium with 20 g·L⁻¹ sucrose, 8 g·L⁻¹ agar, pH 5.8, 25 °C, 16h/8h day/night cycle, 15 days. Growth measured with ImageJ 1.8.0. Chloroplast pigments quantified spectrophotometrically (A₆₆₃·₂ chlorophyll a, A₆₄₆·₈ chlorophyll b, A₄₇₀ carotenoids) per Lichtenthaler & Buschmann 2001. Protein by Bradford. Tissue total phenols by methanol:water extraction and Folin-Ciocalteu. GST activity by CDNB complexation with GSH (ε = 9.6 mM⁻¹·cm⁻¹) at 340 nm for 30 s in 0.1 M Tris-HCl pH 7.7 with 1 mM GSH, 1 mM CDNB, 100 µL protein extract, total volume 3 mL. ANOVA with Tukey’s multiple-range test (P < 0.05) in IBM SPSS Statistics v22. Phytotoxicity index PI = 1 − RLT/RLC where RLT and RLC are root lengths in treated and control seedlings (Rusan et al. 2015).

Limitations: the study reports K, P, Ca, Mg and Fe as nutrients but does not measure the regulated heavy metal contaminants (Pb, Cd, As, Hg, Ni, Cr, Al, Sn). Maize and flax tissue concentrations of any metal were not measured; only growth, photosynthetic pigments, GST activity and tissue polyphenols. The Fe values are therefore informative for OMW nutritional characterization but cannot be used to assess heavy-metal transfer from OMW-irrigated soils into food crops.

Implications

• Certification (HMTc): provides no direct occurrence data for any HMTc-regulated contaminant in any certifiable product category. Useful only as context that olive-mill effluent in Mediterranean producing regions is being actively repurposed as fertigation/irrigation water and as a bioherbicide, which creates a potential cropland exposure pathway worth tracking in future literature pulls that do measure heavy metals in OMW-amended soils and OMW-irrigated crops.
• Courses: relevant as a case study for the olive-oil supply-chain course module on byproduct management and circular-economy reuse, with the caveat that contamination data are absent here and must come from companion sources before any consumer-facing translation is attempted.
• App: contributes no values to any ingredient contamination_profile for HMI-tracked metals. Fe is measured in the OMW fractions but the matrix is industrial wastewater, not a food.
• Microbiome: not addressed by this paper.

Wiki pages this source may touch

• olive-oil — upstream supply-chain context only (OMW is the wet aqueous byproduct of three-phase centrifugation milling; this paper documents one Moroccan-French pilot-scale fractionation/valorization route).
• maize — fertigation context only (OMW UF retentate at 1/8 and NF permeate at 1/6 dilutions tested as irrigation source for maize; no heavy-metal transfer measured).
• iron — OMW nutrient-characterization Fe values; not contamination data.

Verification notes

• Paper does not measure Pb, Cd, As, Hg, Ni, Cr, Al, Sn or any other regulated heavy metal in OMW, soil, or plant tissue. The metals: [Fe] frontmatter entry captures the only HMI-vocabulary metal reported, and it is reported as a nutrient (mg·L⁻¹ in aqueous effluent), not as a contamination measurement in a food matrix.
• matrices uses bare-string descriptors per system-prompt guidance: olive-mill-wastewater for the effluent and its fractions, maize-seedling / flax-seedling for the experimental plant material. None of these are food matrices.
• Evidence tier set to C because the paper does not contribute direct food-contamination occurrence data for any HMI metal. It is a peer-reviewed agronomic study but its HMI-relevant content is upstream-context only.
• No brand names appear in this paper. Equipment vendors named per Part 12 Exception 2 (Firmus pilot system, Orelis Kleansep UF membrane, Suez DK NF membrane, Ecograph T data acquisition by Endress+Hauser, HAWITA Gruppe potting soil, Murashige-Skoog medium from Duchefa biochimie, Sigma-Aldrich, BIO-RAD Bradford reagent, IBM SPSS Statistics v22).
• DOI 10.1016/j.jenvman.2023.117467 verified against the journal-page banner on the first page of the PDF.
• Funding acknowledged: PHC TOUBKAL program grant 19/77: 41474 PK; GESTAD UMR1347 AGROECOLOGIE INRAE Centre de Dijon (weed seed supply).
• Audit subagent (2026-06-01) flagged P (UFR) SD as “510 ± 10”; verified against PDF Table 3 — corrected to “510 ± 11”.
• Audit subagent (2026-06-01) flagged a paper-internal contradiction: Section 3.3 (p. 5) states “no effect was observed for flax germination, highlighting the selectivity of the treatment” while Fig. 5b (captioned as flax seedlings) shows dose-dependent germination inhibition from ~87% (control) to ~73% / ~52% / ~23% at 1/125, 1/100, 1/75 NF retentate dilutions. The “15, 40, 73%” germination inhibition numbers in the same paragraph could plausibly be attributed to either S. arvensis (consistent with the “selectivity” claim) or flax (consistent with Fig. 5b). The Key numbers section attributes them to S. arvensis on the selectivity reading; any downstream synthesis that relies on this paper’s S. arvensis vs flax germination distinction should treat the attribution as ambiguous in the source.

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.