Phytolacca americana — American pokeweed — holds an odd record on this site’s rare-earth element roster: it is, the authors of a new study state, “the only reported rare earth element (REE) hyperaccumulator plant capable of selectively accumulating heavy REEs (HREEs, Gd to Lu plus Y)”. Field populations on the ion-adsorption deposits of Dingnan, Jiangxi reach 1,040 mg kg⁻¹ total REE in leaves (Yuan et al. 2018, Int. J. Phytoremediation 20:415–423), with preferential HREE translocation into foliage already visible at field level. What nobody has shown is where inside the plant the selectivity happens — the tissue and organelle at which the lanthanide series is sorted. Liu, Huang, Zhang, Xie, van der Ent, Morel, Tang and Qiu (Sun Yat-sen University, with van der Ent and Morel) now supply that picture, using laser ablation ICP time-of-flight mass spectrometry for in situ spatial mapping plus chemical-extraction subcellular fractionation (Industrial Crops and Products 252:124360, CC BY; online 11 September 2026).
A gradient indexed by atomic number
The headline pattern is a smooth chemical fractionation inside a living leaf. Root-to-shoot translocation of REEs increases with atomic number along the lanthanide series — the heavier the ion, the more readily it moves upward. All REEs sit predominantly in leaf veins and mesophyll, but the distribution between those two compartments diverges: LREEs are preferentially retained in the vein vascular tissues, while HREEs are unloaded into mesophyll cells more efficiently. And the subcellular bookkeeping moves in step — as atomic number rises, the share of each element held in cell walls falls from 64.5% to 56.2% of total content, while the share compartmentalised into vacuoles rises from 5.5% to 14.5%. The authors’ reading: HREEs cross the bundle-to-mesophyll boundary and the membranes into vacuoles more easily; the plant is effectively handing its heavier lanthanides across to the mesophyll’s detoxification storage.
Two cautions on the numbers. The abstract itemises only the wall and vacuole fractions, which sum to ~70% at both ends of the series — the residual ~30% (organelle and soluble fractions of the extraction scheme) is not broken out in the abstract. And the endpoints are described as “increasing atomic number along the lanthanide series” rather than explicitly assigned to La and Lu, though that reading is the natural one (Y, counted as an HREE here, is not a lanthanide).
The non-obvious implication: a self-fractionating feedstock
For phytomining the point is not the total REE tonnage — it is that HREEs (terbium, dysprosium, yttrium) are the scarce, high-value fraction of the lanthanide market, and the deposits where plant-based recovery is most plausible, the South-China ion-adsorption clays, are precisely the world’s principal HREE source — yet winning those HREEs today means ammonium-sulfate leaching with a well-documented environmental bill. A crop that pre-fractionates — loading a shoot biomass whose HREE/LREE ratio already beats the soil’s — hands a hydrometallurgist a feedstock that has done part of the separation chemistry in the field, in sunlight, without solvent — the economic argument we set out in our August piece on REE phytomining and biomass upcycling. The deeper mechanism also rhymes with a long-standing principle of REE plant physiology: across the lanthanide series, ionic radius shrinks (“lanthanide contraction”), HREEs complex organic ligands more strongly and are generally taken up by roots less readily than LREEs, so a plant whose shoot ends up HREE-enriched must be winning the selectivity in translocation and storage rather than uptake — exactly what the vein-to-mesophyll handoff shows.
The risks, stated plainly
First, vacuolar storage is a double-edged asset. A metal locked in mesophyll vacuoles is detoxified for the plant but harder to leach for the processor: the tonoplast must be broken and the vacuolar milieu (organic acids, phenolics) handled before the HREEs are accessible, adding a recovery step that eats margin. Second, the gradient is a shift in partition, not a clean separation — even at the heavy end, 56% of the element is still in the cell-wall fraction, so expect modest, not dramatic, separation factors between biomass and substrate. Third, and most important, the abstract does not state the experimental system — hydroponics, sand culture or soil, exposure levels, plant age — and in REE work the answer matters: solution speciation (carbonate, organic ligands) strongly drives lanthanide fractionation, and a gradient measured in hydroponics may flatten or shift in real mine soils. Finally, the “only reported” framing is the authors’ priority claim; other REE accumulators such as the fern Dicranopteris dichotoma exist but are LREE-enriched, which is consistent with — though not proof of — the claim. Deployment also has an awkward wrinkle: P. americana is a vigorous weed across much of its naturalised range, so field cultivation for phytomining would mean managing an invasive hyperaccumulator on purpose.
What to watch next
The authors name the obvious next step: identify the HREE-selective transporters in P. americana — candidates would include tonoplast and plasmalemma transporters whose binding-pocket geometry discriminates on ionic radius, the same trick that metal-transport systems elsewhere in the hyperaccumulation literature perform for Ni and Zn. A transporter, not just a pattern, is what would make the trait portable into a higher-biomass or non-weedy chassis. Until the full text is accessible and the growth system is known, the honest summary is: the first spatial map of HREE-selective hyperaccumulation shows a vascular-handoff mechanism sorted by atomic number, verified numbers in hand, with the agronomy still unwritten.
Source: Liu, Huang, Zhang, Xie, van der Ent, Morel, Tang & Qiu (2026), Industrial Crops and Products 252:124360, DOI 10.1016/j.indcrop.2026.124360 (CC BY; online 11 September 2026, print October 2026). All claims are bounded to the publisher abstract: the gold-OA full text sits behind a bot wall that blocked every retrieval route from our host, so the experimental system and the residual subcellular fractions could not be checked in the body text. Every load-bearing number (64.5% → 56.2% cell walls; 5.5% → 14.5% vacuoles; the atomic-number translocation gradient) and the bibliographic record were verified verbatim by three independent reviewers via independent retrieval paths (OpenAlex abstract, Crossref, Unpaywall), with no disagreements. See /methodology/ for how we source and check analysis pieces.