Published today in the Anatolian Journal of Botany (10(2):173–193, online 2026-09-18), Ali Ünver’s source-audited synthesis does something the hyperaccumulation literature rarely does: instead of adding another name to the list of Turkish nickel plants, it audits the list itself. Working through 16 screened source units — reviews, primary studies, regional syntheses and taxonomic resources — the study compiled 66 source-linked candidate evidence entries for Ni hyperaccumulation, Ni accumulation or ultramafic-associated flora in Türkiye, and asked a brutally practical question of each: can this record actually be placed on a map, and does it carry physiological support?
The answer, in numbers, is uncomfortable. After a georeferencing audit, the spatial-confidence distribution is A=12, B=0, C=1, D=51, E=2 — twelve entries with source-supported coordinates, one representative-only locality, fifty-one confined to broad regional, provincial, mountain or grid-level descriptions, and two with insufficient or unresolved locality information. Fifty-three of 66 entries (80.3%) survive in the evidence inventory but cannot be honestly plotted as points. The paper is emphatic, in its abstract and three more times in the text, that an entry “is not equivalent to an independent occurrence record, a unique locality or a newly confirmed Ni-hyperaccumulator observation.”
Occurrence data is not physiology
The audit’s second act screens the country’s GBIF holdings and finds the same fragility in biodiversity-database form: of 88 GBIF occurrence records examined, only 11 (12.5%) contained usable geographic coordinates and 77 (87.5%) did not. Nine coordinate-bearing records were retained with spatial caution and two carried coordinate-related warning flags. None were treated as physiological evidence of Ni accumulation — correctly, because hyperaccumulation is a tissue chemistry property that cannot be inferred from a specimen in a database or a plant growing on serpentinite (the paper cites Baker & Brooks 1989; van der Ent et al. 2013; Krämer 2010). That distinction is the load-bearing wall between floristics and this field: an ultramafic endemic and a Ni hyperaccumulator are different claims, and a herbarium barcode satisfies neither by itself.
The mappable residue — and what it connects to here
Twelve A-class entries consolidate to 14 taxon–locality associations and 12 unique mapped localities, plus one C-class representative point for Noccaea camlikensis at the Çamlık–Derebucak–Kızıldağ complex (a named locality, not an exact collection coordinate) — 13 point symbols in all, built on the MTA 1:500,000 geological dataset in Google Earth Engine, overlaid on core ultramafic lithologies (serpentinite, peridotite) and a wider ophiolitic context layer. The provenance is striking: all twelve A-class localities trace to just two primary studies — six to Çelik et al. (2018) in the Yahyalı region of Kayseri (L01–L06) and six to Altınözlü et al. (2012) across southern and southwestern Anatolia (L07–L12). One of them, L03, is a georeferenced Turkish record of Odontarrhena muralis — the Alyssum murale of the European agromining literature — which this site’s species page currently maps only to the Balkans; the audit’s coordinate-supported Anatolian locality is exactly the kind of record worth adding, and worth verifying against vouchers. Other obligately ultramafine Ni hyperaccumulators in this database with Anatolian ranges, Alyssum lesbiacum and Odontarrhena chalcidica, sit in the same floristic complex the audit’s synthesis literature covers.
Why this matters beyond Türkiye
For agromining prospecting and phytoremediation planning alike, the takeaway is that “known from ultramafics” lists are not targeting datasets. Fifty-one records that cannot be placed more precisely than a province or a mountain are fifty-one reasons a site-selection model would be interpolating from publication geography, not plant geography — and the mapped pattern itself says so: the points cluster in central, southern and southwestern Türkiye because that is where detailed field studies happened, not because the country’s other ophiolite belts (the Kızıldağ, Fındıkpınarı and Amanos complexes the paper names) lack metallophyte flora. The conservation risk runs the same direction inverted: quarrying and nickel-mining pressure on Anatolian serpentines can extirpate populations nobody has ever chemically sampled, and an 80%-unmappable evidence base is a poor shield for a habitat in a permitting dispute. The audit’s own prescription — voucher examination, primary-source recovery, paired soil–plant sampling with organ, method and substrate documented — is the minimum metadata standard this site’s /methodology/ page argues for.
The caveats, including one the audit itself carries
This is a single-author descriptive synthesis with no new tissue chemistry and no inferential statistics — explicitly so; it claims to be a field-prioritization tool, not a checklist, and it refrains from any revised national richness estimate. Read it that way. And the audit inherits a little of the fragility it audits: the appendix maps entry C043 (Noccaea oxyceras) to localities “L09; L12”, yet Table 4’s L12 row lists only C003 (Odontarrhena carica), and C043’s second coordinate is byte-identical to C003’s — a copy-paste-level provenance defect inside the A-class set itself. The headline counts survive it (the 14th association exists in the appendix even if Table 4 under-reports the row), but it is a live demonstration of the paper’s core point: a coordinate is only as good as the chain of custody behind it, and most published chains have missing links.
Source: Ünver (2026), Anatolian Journal of Botany 10(2):173–193, DOI 10.30616/ajb.1960053 (received 2026-05-04, accepted 2026-08-22, online 2026-09-18; gold open access via Dergipark). Full text read from the publisher PDF. Every load-bearing number — the 66 candidate entries and 16 screened sources; the A=12/B=0/C=1/D=51/E=2 spatial-confidence distribution and the 53/66 (80.3%) unmappable fraction; the 12→14→12→13 mapping derivation; the GBIF 88/11/77/9/2 breakdown; the MTA 1:500,000 / Google Earth Engine mapping basis; and the Table 4 locality attributions — was verified verbatim by three independent reviewers on three different model vendors, each retrieving the primary source independently; the only correction among them concerned a coordinate envelope stated in a verification brief (Table 4’s actual minima are 28.60°E and 36.36°N), and the Table 4 / appendix inconsistency on the L12 row is disclosed above rather than smoothed over. See /methodology/ for how we source and check analysis pieces.