Rare-earth agromining has a downstream problem that rarely makes the headlines: what to do with the fern once you have harvested it. In the ion-adsorption clay deposits of southern China, Dicranopteris is the best-documented rare-earth-element hyperaccumulator, reaching total REE above 3,000 µg g⁻¹ in its fronds; our database records it as D. linearis, while this study uses the name D. pedata, a form long lumped with it. A new case study, Feng et al. (2026) in Waste Management, asks a blunt question about that biomass and answers it with a specific mechanism.
The result, stated plainly
The authors carbonise (pyrolyse) Dicranopteris clippings into a char they call REE/C, then test it as a soil amendment on contaminated orchard soil using passivation and leaching experiments. The char is mostly amorphous carbon with a small graphitic fraction, and — the load-bearing detail — agglomerated La-, Ce- and Y-containing particles are precipitated on its surface, giving a specific surface area of 439.48 m² g⁻¹. Applied to the soil, it cut the metals in the leachate: lead fell to 5.46 % of the control (a ~95 % reduction in leachable Pb), copper to 38.71 % (~61 %) and cadmium to 33.80 % (~66 %). In fractionation terms, REE/C pushed the metals out of the acid-soluble and reducible pools — the mobile, bioavailable ones — and into the oxidisable and residual pools that stay put.
Note the asymmetry the headline number hides: lead is nearly locked down, but cadmium — the most food-chain-mobile and toxicologically significant of the three in agricultural produce — is the least immobilised of the trio. On a farmland soil that is the result that should set the caution, not the 95 % on Pb.
Why this is non-obvious
The interesting move is which part of the plant does the work. The rare earths are usually treated either as the payload you want to recover from a phytomining crop, or as an awkward contaminant in the ash. Here the plant’s own accumulated La, Ce and Y appear to act as part of the reagent, not inert ballast: the abstract credits the REE-decorated surface with the high surface area and adsorption capacity, and says endogenous REEs “play an important role” in the remediation. It stops short of isolating those REE particles from the carbon matrix as the dominant sorbent, so read this as the authors’ framing and a working hypothesis rather than a settled apportionment — but even hedged, it points at a loop worth closing for REE mining regions, where the harvested hyperaccumulator biomass becomes an amendment for the lead, copper and cadmium that contaminate the same landscapes.
In this trial several specific soil-quality endpoints also improved — total nitrogen and alkaline-hydrolysable nitrogen rose 2.59- and 1.55-fold, two soil enzyme activities were higher (β-N-acetylglucosaminidase at 41.00 IU L⁻¹, leucine aminopeptidase at 95.61 IU L⁻¹), and microbial diversity increased. Those are the discrete measured endpoints, not a composite “soil health” score.
The trade-off nobody will headline
There are two competing fates for a tonne of REE-laden fern, and this is our read of the choice the approach forces rather than a claim in the paper. You can recover the rare earths — the agromining thesis, where the metal is the saleable product — or you can carbonise the biomass into an amendment, in which case the REEs are spent locking down someone else’s lead. The same accumulated metal cannot be both a rare-earth concentrate and an in-situ immobilising agent. That makes carbonisation the rational path only where the REE grade of the biomass is too low, or the recovery cost too high, to justify extraction — precisely the marginal biomass that would otherwise be a disposal liability.
The genuine risk
Two cautions matter more than the headline percentages. First, immobilisation is not removal. Shifting Pb, Cu and Cd from bioavailable into oxidisable and residual fractions is phytostabilisation, not phytoextraction: the metals remain in the soil, held by a speciation balance that can reverse — the “reducible” and “oxidisable” pools are each only conditionally stable, mobile again if pH drops or the redox state shifts — and a leaching test is a short-term proxy for a long-term question. Second, and less comfortably, spreading a La-, Ce- and Y-rich char onto farmland is itself a rare-earth loading pathway. Light rare earths have documented phytotoxicity at elevated soil levels, and an amendment designed to leave its active phase in place is, by construction, adding rare earths to an agricultural soil. In fairness, the study’s own short-term soil-biology endpoints show no acute harm at the tested dose and timeframe, so the concern is not that this char is acutely toxic; it is longer-term accumulation, repeated application, and transfer into the orchard crop — questions the abstract does not address, and that a practitioner should treat as open rather than settled.
Bound the claim accordingly: this is a single study, on orchard soils and one biomass feedstock, using passivation and leaching experiments rather than a multi-season field trial. We could not access the full text — ScienceDirect is paywalled — so every figure above is quoted from the publisher abstract, and we make no claim about the methods, controls, mass balance, or field durability we could not read. What the result establishes is a mechanism and a direction, not a deployment.
This analysis is grounded in the primary source’s publisher abstract (Feng et al. 2026, Waste Management 223:115704), which we retrieved and cross-checked against the source across two independent model vendors (Claude and Kimi) plus an independent web re-retrieval; two further coding vendors were unavailable at run time. Every figure is a verbatim quote from that abstract; the full text was not accessible, and claims are bounded to the abstract accordingly. The species’ REE figure follows the compilation in our database; see our methodology for how records are verified.