Artisanal and small-scale gold mining is the largest single anthropogenic source of mercury on Earth — an estimated 838 t emitted to air in 2015 alone, 38% of global anthropogenic emissions (UN Environment’s Global Mercury Assessment, cited in the paper), and in Ghana gold was 48% of total exports in 2022. Revegetating abandoned ASGM land means choosing trees that survive mercury while doing something useful with it. Sommer, Guo, Rasche, Hagemann and Zörb (Environmental Science and Pollution Research 33:11665–11678, 2026) ran two greenhouse pot experiments on four legume trees widespread in West Africa — Acacia mangium, Gliricidia sepium, Leucaena leucocephala and Senna siamea — dosing a sand:vermiculite substrate with HgCl₂ at 12.5 and 25.0 mg Hg kg⁻¹ for eight weeks, then adding an arbuscular mycorrhizal inoculum (Rhizophagus irregularis QS81) in a second experiment.
The winner concentrates mercury in its wood
At 25 mg Hg kg⁻¹, above-ground accumulation ranked A. mangium > G. sepium > S. siamea > L. leucocephala, with striking tissue numbers: 445 mg/kg Hg in A. mangium stems and 360.3 mg/kg in G. sepium stems, against 77.0 and 95.8 mg/kg respectively in leaves. In the AMF experiment, AMF-treated A. mangium reached 202 mg/kg in leaves and 203 mg/kg in stems — the highest above-ground values measured. A. mangium also carried the highest mean tolerance index (TI = 1.07): its assimilation rate and stem height fell, but total dry weight did not — mercury stunted it without starving it. G. sepium told the opposite story: high uptake but real toxicity, with four of ten mercury-treated plants dying — the authors disqualify it despite its numbers. L. leucocephala was a clean excluder (TI mean 0.86, lowest uptake, physiology untouched), and S. siamea sat in between with TI ≈ 1 and a bioaccumulation factor above 2 in both experiments.
The most interesting number is a ratio. Doubling the substrate dose from 12.5 to 25 mg/kg raised stem Hg in A. mangium 49-fold and leaf Hg in G. sepium sevenfold, while root Hg merely doubled across all four species. Whatever gate keeps mercury out of above-ground tissue saturates or opens sharply with external dose — accumulation in stems is threshold-like, not linear. For field design that cuts both ways: modest soil Hg may barely move into harvestable wood, and the plants that accumulate most are those pushed closest to their toxicity limit.
The mycorrhiza does not deliver
Here is the result practitioners should sit with. R. irregularis colonised successfully — mycorrhizal frequency was high in A. mangium and G. sepium roots, and in L. leucocephala colonisation actually rose under mercury — yet inoculation had no significant effect on mercury accumulation in any of the four species, and no significant effect on the bioaccumulation factor either. What AMF did show was a subtler “buffer”: the mercury-induced drop in assimilation was less pronounced in inoculated accumulator species, and it significantly relieved the assimilation hit in S. siamea. The fungi support plant health under stress; they do not, in this system, move more mercury into biomass. That is a meaningful null for a literature in which AMF is routinely sold as an uptake-enhancing inoculant for phytoremediation — mostly on cadmium, nickel and zinc evidence. Mercury’s chemistry (tight binding to fungal walls, hyphae and vesicles) plausibly immobilises it in the root zone instead of trafficking it upward — consistent with the finding that roots held the highest Hg concentration in every species, and no species achieved a translocation factor above 1.0.
The species this rhymes with
The pattern — root retention, TF < 1, shoot values that are real but sub-hyperaccumulator — is the same one our recent piece on Pteris vittata described from the other direction: that fern locks mercury up as β-HgS and glutathione complexes, and its 83 mg/kg shoot maximum left the “mercury hyperaccumulator” label deliberately unclaimed. There is still no accepted hyperaccumulation threshold for mercury (the canonical cutoffs are 10,000 mg/kg for Zn, 1,000 for Ni, 100 for Cd), and the paper cites the field’s own tally: more than 200 plant species screened in two decades without a confirmed mercury hyperaccumulator — Erato polymnioides and a few other Neotropical rainforest species are the only ones so labelled, and none of them belongs to West Africa. Sommer et al.'s trees are therefore best read as phytostabilisation and phyto-management candidates, not phytoextractors: the mercury is held, mostly below ground, in living tissue.
What to be skeptical about
The bounds matter, and the authors’ own text supplies some. This is a pot study in sand:vermiculite with soluble HgCl₂ spikes, eight weeks long, on young trees — not tailings chemistry, where sulfide speciation, organic matter and microbial methylation decide what mercury actually does. The AMF experiment also differed in exposure protocol (5 mg kg⁻¹ weekly for five weeks), and A. mangium accumulated more there than at the nominal same dose, which the authors attribute to plant age and exposure duration — a reminder that these numbers move with experimental details. One internal inconsistency is worth flagging: the text calls G. sepium’s TF “high, 35.4” while also stating no species exceeded 1.0; the table supports the latter, and 35.4 most plausibly belongs to a different column (likely the increase factor). And A. mangium itself is a non-native plantation species — the paper notes the invasive risk of introduced species even as it recommends planting it, a tension any Ghanaian deployment would have to manage. Finally, timber is a removal pathway only if the mercury-laden wood is handled as contaminated material; harvesting 445 mg/kg stems into fuelwood or sawn products would simply relocate the problem. The defensible design the authors sketch is a mixed stand: A. mangium as accumulator, S. siamea as tolerant companion, L. leucocephala as excluder that keeps mercury out of fodder and food webs while rebuilding soil — monitored over decades, with the trees treated as a managed, eventually harvestable sink rather than a quick fix.
Source: Sommer, Guo, Rasche, Hagemann & Zörb (2026), Environmental Science and Pollution Research 33:11665–11678, DOI 10.1007/s11356-026-38009-y (published 2026-07-11; open access, CC-BY 4.0). Every load-bearing number — the 12.5/25.0 mg kg⁻¹ HgCl₂ doses, the 445 and 360.3 mg/kg stem maxima, the 202/203 mg/kg AMF-experiment values, the 49-fold and sevenfold increase factors, TI means of 1.07 and 0.86, the four-of-ten G. sepium mortality, the universal TF < 1 and the AMF null result — was verified verbatim by three independent reviewers who each retrieved the primary source separately (Springer full text, PubMed Central mirror, and Crossref metadata); no numeric discrepancies were found. The reviewers also caught, and this text follows the paper’s own wording on, “Erato polymnioides and a few other species” rather than a singular claim, and the TF-35.4 internal inconsistency is flagged above as a likely column mix-up rather than a real translocation factor. See /methodology/ for how we source and check analysis pieces.