A survey designed to find the best field, and what it actually found
Alves, da Silva and Novo (2026), in the Euro-Mediterranean Journal for Environmental Integration (doi:10.1007/s41207-026-01291-w), walked the Bragança ultramafic massif in north-eastern Portugal with a simple question: which of its serpentine sites could grow Odontarrhena serpyllifolia as a nickel crop? They sampled four sites (S1–S4, three plant–soil pairs each), measured soil chemistry and plant biomass, and computed the number agromining actually monetizes — the harvestable amount (HA), biomass × shoot Ni concentration.
The result inverts the intuitive ranking. Pseudo-total soil Ni ran from 862 mg kg⁻¹ at S1 to 2494 mg kg⁻¹ at S2 — S2 holds nearly three times the nickel. Yet the harvestable nickel per plant came out almost exactly backwards: 19.3 mg Ni plant⁻¹ at S3, 12.2 at S1, 8.1 at S4, and only 6.3 at S2. The richest soil grew the worst crop, by a factor of three. Shoot Ni told the same story: S1 (2133 mg kg⁻¹) and S3 (1870) significantly out-accumulated S2 (1217), with a single maximum of 2504 mg kg⁻¹ at S1.
The mechanism: what the plant actually experienced
Why? The soils differ less in what they contain than in what they let a plant do. S2, the nickel-rich site, is also the harshest: pH 7.50 (against 6.42 at S1), Mg at 129,013 mg kg⁻¹ (against ~45,000), and a soil Ca:Mg ratio of 0.01 — effectively zero calcium headroom. S1 and S3, the productive sites, combine slightly lower pH, more phosphorus (1085 and 1682 vs 449 mg kg⁻¹), more sodium, far less Mg, Fe, Co and Mn, and Ca:Mg ratios of 0.19–0.20.
The plant side confirms the edaphic story. O. serpyllifolia shoots averaged 15,451 mg kg⁻¹ Ca against 7,082 mg kg⁻¹ Mg — more than twice as much calcium as magnesium in tissue growing on a magnesium-saturated substrate. That Ca-for-Mg substitution is a textbook serpentine adaptation: raising internal Ca relieves Mg and Ni toxicity and, as Chaney and colleagues showed in solution culture, directly promotes Ni accumulation and translocation. Bioconcentration (shoot Ni ÷ soil Ni) was significantly highest at S1 — partly arithmetic, since S1’s soil is the poorest in Ni, but consistent with the correlation structure: across the 12 paired observations, harvestable Ni tracked soil P positively, shoot Ni and BCF tracked the Ca:Mg ratio and soil Na positively, and BCF fell as soil Fe, Co and Mn rose.
The DTPA lesson: extractable nickel is not plant-available nickel
The most quietly important finding is a negative one. DTPA-extractable Ni — the standard “bioavailable” assay — peaked at S4 (64.8 mg kg⁻¹), which produced middling shoots, and S2’s extractable Ni (34.7) was nearly half of S1’s (39.8) despite S2 holding 2.9× the total. If you had ranked these four sites by the agronomic soil test, you would have picked wrong twice. The authors’ reading is that hyperaccumulator roots do not passively sample a static pool: they deplete and re-equilibrate Ni locally, so a snapshot extraction misses the dynamic supply the root actually engineers. For site screening across the Mediterranean ultramafic belt — Albania to Anatolia — that argues for pairing soil assays with plant-based indicators rather than trusting DTPA alone.
Opportunity, risk, and the conservation shadow
The opportunity is practical: the difference between S2 and S3 (6.3 vs 19.3 mg Ni plant⁻¹) is a ~3× yield swing from site selection alone, before any agronomy, and the authors’ amendment program — add P, Ca and Na, nudge pH down, buffer Mg/Fe/Co/Mn stress — is a direct, testable follow-up. There is also a conservation dimension: Bragança’s serpentine flora is endemic-rich, and agromining proponents often assume ultramafic land is “wasteland” with no opportunity cost. A survey like this is the counter-instrument — it maps which patches are agronomically valuable, which is also a map of where disturbance should not happen.
The risks are real and the authors own most of them. Sampling was in October 2023, past flowering into senescence, so biomass and HA are conservative underestimates of peak-season potential; the same group’s earlier Bragança work (Morais et al. 2015) recorded 3405 mg kg⁻¹ in shoots here, and our species record for O. serpyllifolia reaches 15,000 mg kg⁻¹ — this October snapshot is a floor, not the species’ ceiling. Plant density was not measured, so no per-hectare extrapolation is legitimate. And with three plots per site and twelve paired observations, the correlation claims (Fig. 6) are directional hypotheses, not established drivers. One housekeeping caveat from the fact-check: the paper’s Ca:Mg row is reported without a stated basis; we quote the values as printed.
Bottom line
For nickel agromining in Mediterranean serpentine systems, soil grade is a weak predictor of yield; soil chemistry is the strong one. The best hectare is not the one with the most nickel but the one where pH, Ca:Mg balance and nutrient supply let a hyperaccumulator convert its soil into biomass. Site assay batteries that skip plant tissue — or lean on DTPA alone — will keep picking the wrong fields, and the same map that finds the good fields marks the endemic-rich ones worth leaving alone.
Provenance: based on the open-access full text (CC BY 4.0, Euro-Mediterranean Journal for Environmental Integration, published online 10 September 2026), read in full via Europe PMC (PMC13627269). Every load-bearing number was verified verbatim against the primary source by three independent reviewers on separate vendors (Claude / GPT / Kimi, each self-retrieving the full text); all 15 checked claims, including every site-to-value pairing and the mean-vs-maximum distinction, were confirmed by all three with zero numeric discrepancies. See /methodology/.