Hasan, Ashraphy, Akhter and colleagues at BCSIR’s Institute of National Analytical Research and Service (INARS) in Dhaka and Hajee Mohammad Danesh Science and Technology University published on 29 September 2026 in RSC Advances (open access, CC-BY) a field screening of twelve naturally growing plant species from the industrial belt of Gazipur, some 40 km north of Dhaka — an area dense with paint, textile, pharmaceutical and garment effluents. All tissue and soil digests ran in an ISO/IEC 17025-accredited laboratory by flame AAS, with NIST-traceable certified reference materials and spike recoveries of 96–104%. The analytical benchwork is a genuine strength. The interpretation is where a reviewer starts sharpening a pencil.
What the instruments actually measured
Across the twelve species, the highest tissue concentrations (Table 2, mg kg⁻¹ dry mass) were Pb 98.5 in Mimosa pudica, Cd 4.52 and Ni 112.5 in Mikania micrantha, Cr 61.1, Cu 239.0 and Zn 2339.0 in Stellaria media, and Co 31.7 with Mn 11,572.4 in Hygrophila polysperma. The paired soils (Table 3) are modest: averages of Pb 11.3, Cd 0.16, Cr 5.72, Ni 30.1, Cu 42.5 and Zn 19.6 mg kg⁻¹, and the authors state — correctly against their own tabulated comparison values — that soil toxic metals sit “well below” international soil-quality guideline thresholds. The plausible explanation they offer is that bioavailability, not total content, drives the tissue values; they concede pH, organic matter, CEC and metal speciation were not measured, so that mechanism remains a hypothesis, not a finding.
Ratios are not concentrations
Here is the load-bearing point for anyone citing this paper. A translocation factor of 8.15 (Pb, M. pudica) means shoot-to-root partitioning; a bioconcentration factor of ~9 (Pb, Colocasia esculenta) or 14.91 (Zn, Urena lobata) means tissue-to-soil enrichment against a low denominator. Neither says anything about absolute accumulation. Measured against the working shoot thresholds the field actually uses — van der Ent et al. (2013, Plant and Soil 362:319) and the literature it consolidates: roughly 100 µg g⁻¹ for Cd, 300 for Co and Cu, 1,000 for Ni, 3,000 for Zn, 10,000 for Mn — none of the twelve species qualifies, or comes close. The best Cd value is 22-fold below threshold; the best Ni value nine-fold; even the highest Zn (2,339 mg kg⁻¹) misses the 3,000 mark. The single value that crosses any canonical line is H. polysperma’s Mn at 11,572 mg kg⁻¹ — a 16% exceedance of the 10,000 µg g⁻¹ Mn threshold, if that lone Table 2 number is a shoot concentration, which the paper never states. Lead, tellingly, has no confirmed threshold at all: van der Ent et al. treat Pb hyperaccumulation as essentially undemonstrated because shoot surfaces adsorb soil dust that washing cannot fully remove — a caveat this field-collected, rinse-only dataset cannot exclude. The index scheme the paper cites for “hyperaccumulator” status (BCF > 10 plus TF > 1) comes from a 2023 review and a 2017 methods paper, not from Baker & Brooks or van der Ent; taken literally it would crown water hyacinth (Mn BCF 14.2) and Urena lobata (Zn BCF 14.91) as hyperaccumulators at shoot levels of 5,317 and 675 mg kg⁻¹ respectively. That is the reductio ad absurdum of ratio-only definitions.
The risks the paper underplays
Three stand out. Food safety: Colocasia esculenta is taro, a staple corm and leaf vegetable, and its tissues carry Pb 92.2 and Cd 1.59 mg kg⁻¹ — more than 18× and 5× the FAO/WHO comparison limits the paper itself prints — while M. micrantha carries Cd 4.52 (15×) and M. pudica Pb 98.5 (nearly 20×). On soils that officially “pass”, peri-urban foraging and leafy-vegetable cultivation can still move toxic metals onto plates; recommending a food crop as a Pb accumulator around battery industry without discussing the ingestion pathway is an omission worth naming. Invasives: two of the twelve “native plants” are on the IUCN ISSG list of the world’s hundred worst invasive alien species — Eichhornia crassipes (water hyacinth) and Mikania micrantha (the “mile-a-minute” vine, already a catastrophic invader across South and Southeast Asia). Proposing Mikania as a phytoextraction planting is proposing to cultivate a plant killer. Denominator sensitivity: every BCF and TF here was computed against one of the mildest soil suites in the phytoremediation literature; transfer these species to a genuinely contaminated site and the same ratios could invert.
The findings worth keeping
Strip the index framing and there is real, usable signal. A multi-metal, field-colonising flora has been characterised for one of the most polluted peri-urban corridors in Bangladesh, in a accredited lab with defensible QA — a biomonitoring baseline the region lacked. The PCA split is ecologically sensible: Brachiaria mutica and M. micrantha cluster as multi-metal translocators (TF ~4–8 across seven metals for the former — promising only as a sieve for species worth testing properly), while C. esculenta, E. crassipes, S. media cluster as root retainers, i.e. phytostabilisation candidates where keeping metals out of the shoot is the goal. For low-to-moderate contamination — which is what Gazipur’s soils mostly hold — tolerant, high-biomass, harvestable colonisers matched to stabilisation (rather than hyperaccumulation arithmetic) are the honest deployment path. See our hubs on /lead, /cadmium, /nickel, /zinc and /manganese for what those thresholds and strategies look like when the numbers are real.
Data-provenance problems the reader should know
Five items, none of which invalidates the instrument readings but all of which bound what can be claimed. First, Table 2 is captioned “roots and shoots” yet prints exactly one value per element per species; since TF requires shoot and root data, none of the headline Fig. 2 indices can be derived from the tabulated data — spot checks confirm it (the claimed Zn BCF of 14.91 would require a paired soil Zn of 45.3 mg kg⁻¹; the site measured 22.8, and the tabulated ratio is 29.6). Second, soil site S-12 reproduces the twelve-site average row to every printed digit across all nine elements — either a remarkable coincidence or a back-filled row; raw data are “available on request” only. Third, the paper’s “well below guidelines” claim is contradicted by its own Table 3, where average soil Cu (42.5) sits ~9× above the FAO value (4.5) the table prints — that Cu limit is itself implausible, as is the tissue Ni limit of 0.1 mg kg⁻¹ in Table 2, both traceable only to a vague grey citation. Fourth, the reference list contains entries irrelevant to their citing contexts (a SARS-CoV-2 antibody paper, an optical-switching materials paper). Fifth, the study-area coordinates span longitudes that put Gazipur ~250 km east of where it is. Screenings like this are how candidate species get shortlisted; they are not evidence of hyperaccumulation, and this one should be cited as twelve tolerant colonisers with a defensible elemental dataset — nothing more.
Provenance: full open-access text read (Europe PMC full-text XML of PMC13621246, all four tables extracted and independently recomputed; CC-BY). Every load-bearing number was independently re-retrieved from the primary source by three separate reviewers on different model vendors (Claude, GPT, Kimi) using a shared claim template, with zero cross-vendor discrepancies on any digit; external anchors (van der Ent et al. 2013 thresholds, IUCN ISSG-100 status of Mikania micrantha and Eichhornia crassipes, taro as a staple food crop) verified independently against the same sources. Paper-internal issues — the underivable Fig. 2 indices, the S-12/average-row identity, the misprinted regulatory rows, the uncited canonical threshold literature, the irrelevant references — are reported above rather than smoothed over; methods and limits at /methodology/.