Phytoextraction budgets are usually written as if a hectare yields the same metal every year: multiply one good harvest by the number of croppings and read off the cleanup time. A new open-access study puts that assumption on a bench and watches it fail. Working on substrate from Argentina’s decommissioned Sierra Pintada uranium mine, Ubaldini, Castaño Gañán and colleagues (2026) in Applied Sciences grew sunflower (Helianthus annuus) through three successive three-month cycles in the same pot of soil — TRL-4 laboratory bioreactors — and measured how much Cu, Zn, nickel, Sr and P each cycle removed. The answer: almost all of it came out in cycle one, and the rest of the run tells us why.
What the trial did
Sunflower was cropped on mine soil blended 1:1 with volcanic ash and amended with 350 ppm ZnSO₄, at near-neutral pH (6.8–7.2) and mildly oxidising redox (+300 to +320 mV). At harvest the plants were washed, dried and analysed by X-ray fluorescence; bioconcentration (BCF) and translocation (TF) factors were computed against the soil of each cycle. In the first cycle whole-plant tissue carried, per kilogram of dry biomass, 544.73 mg Cu, 429.00 mg zinc, 230.01 mg Sr, 192.67 mg Ni and 2318.77 mg P (at ~198 g dry mass per plant). Scaling those single-cycle figures 1:100 to a projected TRL-6 “Vegetable Depuration Module” of 290 plants in a cubic metre of soil gave an estimated 16.15 g Cu, 20.81 g Zn, 11.39 g Sr, 7.43 g Ni and 126.87 g P removed. Those module numbers are a projection, not a measured field yield — no larger unit was actually built — and the point of the paper is what happens next.
The collapse, and the twist
Extraction did not hold. Shoot BCFs fell off a cliff after cycle one: zinc from 2.69 to 0.25 to 0.19, and nickel from 2.10 to 0.03 to 0.06 across the three cycles. The obvious reading — the plant mined the soil out — is the wrong one, and this is the study’s most useful contribution. Total metal in the soil did not fall; measured soil zinc actually rose (312.57 → 407.00 ppm) and phosphorus more than doubled (585 → 1750 ppm) from the first cycle to the third. If uptake collapsed while the total pool held or grew, then what the first crop exhausted was not the metal but its readily available (labile) fraction. The authors point to immobilisation in the rhizosphere: total glomalin-related soil protein roughly doubled over the run (T-GRSP 1.3 → 2.8 mg g⁻¹, a significant rise), consistent with metal being locked into recalcitrant organo-mineral and fungal phases rather than harvested away. That mechanistic attribution is the weakest link in the chain — mycorrhizal colonisation stayed low and did not differ significantly between cycles (p = 0.0944), and no mass balance or sequential-extraction assay was run — so read glomalin as a candidate sink alongside Fe/Mn-oxide sorption, not a proven one.
Why it matters for agromining
The practical lesson is sharp: year-one yield must not be annualised. A crop that skims the pre-existing labile pool in its first pass gives back steeply diminishing returns thereafter unless that pool is replenished. This is the same tension that runs through our recent coverage — the rhizosphere processes that make a site safer by fixing metal in place (phytostabilisation) are precisely what starve a repeated harvester of feedstock. It converges, from a completely different system, on the labile-pool ceiling we reported for nickel agromining on urban soil just one day earlier (our 29 July piece): two 2026 studies — one a pot trial with a true hyperaccumulator, this one a multi-cycle bioreactor run with a high-biomass generalist — reach the same conclusion that phytoavailability, not the total ore, governs the harvest.
It also puts sunflower’s economics in perspective. The paper’s own benchmark is Odontarrhena chalcidica, Europe’s Ni-agromining workhorse (our database records foliar nickel around 22,000 µg g⁻¹, roughly 2%): citing Bani et al. (2025), it can yield >150 kg Ni ha⁻¹, about 50 g Ni m⁻³ at 0.30 m depth, against sunflower’s projected 7.43 g m⁻³ — the paper calls this “one order of magnitude,” and even that flatters sunflower, whose whole-plant nickel sits roughly two orders below a real metal crop. High biomass does not buy back the concentration gap here.
Read the label carefully
Two caveats bound every number above. First, this is a small, early-stage experiment: two bioreactors, six plants per cycle, one substrate, laboratory scale, with the headline module figures obtained by ×100 projection to a unit that was never operated — and the raw data are embargoed in an unpublished doctoral thesis, so they cannot yet be re-derived independently. Second, the paper repeatedly calls H. annuus a “hyperaccumulator” and states its cycle-one tissue “exceeded the hyperaccumulator threshold for Ni and Cu.” By the paper’s own cited thresholds (Ni 1000, Cu 300 mg kg⁻¹) that is only true for copper; whole-plant nickel of 192.67 mg kg⁻¹ is about five times below the nickel line — an internal contradiction, and on a whole-plant rather than a foliar basis to boot. Sunflower is a useful high-biomass accumulator, not a hyperaccumulator in the sense used for Phyllanthus rufuschaneyi, Haumaniastrum robertii or Aeolanthus biformifolius, all cited in the paper’s own framing.
There is also a risk the study cannot see. The one lever the authors recommend for later cycles — amendments to restore bioavailability — is, in the general case, acidification or chelation to re-open the labile tap. On a uranium mine that is the wrong direction to push: the same chemistry that frees Ni, Cu and Zn will also mobilise uranium, radium and their daughters into leachate, and these bioreactors kept their outlet closed, so leachate chemistry under mobilisation was never tested. Treat that as a flagged hazard, not a finding — but it is exactly the interaction a field programme would need to measure before sustaining multi-cycle extraction on radiologically contaminated ground. (The paper also invokes Noccaea caerulescens — for its cadmium labile-pool depletion, not zinc — as the classic demonstration that hyperaccumulators draw down the same available fraction.)
Bottom line. The defensible takeaway is not a new metal crop; it is a clean measurement of a limit. Repeated phytoextraction is governed by how fast the labile fraction is replenished, not by the total metal in the soil — so on this uranium substrate the crop immobilised metal as much as it removed it, single-cycle yields cannot be projected forward, and any attempt to force yield by re-mobilising the labile pool carries an untested radionuclide-leaching risk.
Provenance: every figure here was read from the open-access full text (CC BY) and independently re-verified against the source; the interpretation and the flagged risks are ours. See our methodology for how we source and check.