Literature

Six legumes and a null result: Noccaea co-cropping moves the soil's DTPA-extractable metals, not the plant's

What was grown

Pierre Leglize, Julien Ancousture, Ya Brigitte Assoa and Emile Benizri (Université de Lorraine / INRAE, with CEA Grenoble) ran a five-month growth-chamber pot trial on soil from a brownfield at Homécourt, in northeastern France (doi:10.1007/s42729-026-03626-6, Journal of Soil Science and Plant Nutrition, accepted 3 October 2026). The soil is mildly contaminated: pH 8.32, total Zn 327 mg kg⁻¹ with 13.1 mg kg⁻¹ DTPA-extractable, total Ni 23.8 with 0.277 DTPA-extractable, and total Cd 0.347 with 0.056 DTPA-extractable mg kg⁻¹ — anthropic pollution above the regional background, but far below metalliferous-site grades. Noccaea caerulescens (see /species/noccaea-caerulescens-zn/) came from the Ganges calamine population in southern France, one of the strongest known Cd/Zn phytoextraction populations, and was grown alone for six weeks before one of six legumes — lentil, lucerne, fenugreek, white clover, faba bean or common vetch — was sown alongside it on 17 July 2023. The legumes were harvested at first flowering on 4 September, dried, crushed, and incorporated into the top 2 cm as green manure; Noccaea was harvested on 23 October. Four replicates per treatment, ANOVA with Tukey HSD at p ≤ 0.05. (The paper’s prose twice calls the legume phase “six weeks”; its own dates make it seven. We use the dates.)

The null on the plant side

This is the finding, and it is a non-finding: no co-cropping treatment changed anything in the plant. Every cell of Table 2 (TE concentrations, translocation factors, bioconcentration factors, yields) and Table 3 (ten functional traits) carries the same significance letter “a” — nothing differed from the Noccaea monocrop at p < 0.05. Shoot Zn stayed at 1,250–1,608 mg kg⁻¹ (monocrop 1,407 ± 201), Cd at 5.4–9.3 mg kg⁻¹, Ni at 22–31 mg kg⁻¹. The authors note, correctly, that the van der Ent et al. (2013) hyperaccumulation thresholds — Cd > 100, Ni > 1,000, Zn > 10,000 mg kg⁻¹ — were not reached on this soil in this cycle. Zn translocation factors ran 10–21 (a root-to-shoot pump working as designed; element hub /metals/zinc/), but the absolute concentrations and the ~10–15 mg Zn plant⁻¹ harvested reflect a mildly contaminated soil, not a calamine. Their own explanation for the flat plant response is the honest one: a ~7-week window between green-manure incorporation and final harvest was too short for residue mineralisation to release nitrogen, and the moderate contamination imposed too little stress for trait plasticity to show. The facilitative-N hypothesis — legume N boosts hyperaccumulator biomass and thus metal yield — had no time to fire.

The signal on the soil side

Here the experiment did move, significantly. DTPA-extractable Cd was lower in every co-cropping system than in both the legume monocultures and the unplanted bulk soil, and Noccaea + Trifolium repens carried the lowest DTPA-Cd and the only significant DTPA-Ni reduction versus bulk soil. All Noccaea-containing pots ran significantly higher pH than bare soil — and the authors’ mechanism is the interesting part: a Brassicaceae fed on nitrate alkalinises its rhizosphere (the classic Alyssum murale pattern), and at pH > 8 with a carbonate-buffered brownfield soil, that alkalinisation plus direct uptake, not a mysterious microbial synergy, is what shrinks the extractable pool. Aggregate stability pointed the same way, though not significantly: after ultrasonic dispersion the monocrop lost 43.0% of its mean weight diameter versus only 11.4% under white clover co-cropping. White clover was also recorded in phytosociological surveys of the brownfield itself (data not shown) — the authors’ pick as “most promising” is a native-already-there legume, which is a defensible reason.

The BCF trap

Table 2’s bioconcentration factors look spectacular — Cd BCF up to 165, Zn BCF ~108 — until you read the methods: the BCF denominator is the initial DTPA-extractable soil concentration, a labile pool measured in hundredths of a mg kg⁻¹, not total soil. Divide shoot Zn by total soil Zn and the enrichment is a sober ~4.3-fold. Both numbers are arithmetically faithful to their definitions (6.1/0.056 = 108.9 exactly), but only one of them means what a casual reader thinks “bioconcentration” means. A BCF of 165 against a 0.056 mg kg⁻¹ denominator is an artefact of dividing by a near-detection-limit pool, not evidence of a super-accumulating plant on this soil. We flag it because DTPA-denominator BCF inflation recurs in this literature.

An exploitative hyperaccumulator, and a trade-off

The trait data are the paper’s genuinely novel contribution: Noccaea SLA of 38–57 cm² g⁻¹ sits far above the TRY-database global mean of 16.6 cm² g⁻¹ (Kattge et al. 2011, 45,733 entries), with high leaf N — an exploitative fast-acquisitive strategy, matching its ruderal niche on disturbed metalliferous ground. That cuts against the TRY-based analysis of Gervais-Bergeron et al. (2023), who found hyperaccumulators as a group skew toward smaller leaves and weaker competitive nutrient acquisition; the authors reasonably argue this is a population-level effect (Ganges calamine versus serpentine floras), and one pot trial of one population cannot settle it. Within the experiment, every TE-accumulation index was negatively correlated with shoot and root biomass and positively correlated with leaf N and P — the authors read this as a growth-versus-accumulation trade-off, with the corollary that tissue-N/P management may matter more than neighbour identity. Caution: the paper’s Mantel test (distance-matrix level) reports the opposite sign for the same trait–accumulation relationship and never reconciles it, and the correlation sample is small; treat the trade-off as a hypothesis.

What it means for phytoextraction on brownfields

The practical reading: on mildly contaminated brownfield soils, the case for legume co-cropping is soil-side, not plant-side. If you expect a legume neighbour to lift the hyperaccumulator’s tissue metal, this trial says you will wait — the N has to mineralise first, and a single short season will not show it. What a clover companion did deliver within one season is a smaller labile Cd and Ni pool, better-structured soil, and a native species already adapted to the site — ecosystem services that matter if the goal is remediation plus restoration rather than maximum metal yield. The literature the authors cite runs both ways: co-cropping Sedum alfredii with Brassica oleracea nearly doubled shoot Cd (27.3 → 51.2 mg kg⁻¹; Ma et al. 2021), while Sedum plumbizincicola competed Noccaea’s phytoextraction efficiency down (Jacquet et al. 2025) and Brassica juncea suppressed its Zn uptake where Lupinus albus enhanced it (Martínez-Alcalá et al. 2020, as cited in the paper) — so the honest summary is that co-cropping outcomes are species-pair- and timing-specific, and “add a legume” is not a generic amplifier. The risk in overselling this paper’s trend-level plant numbers is real: with n = 4 and the largest single effect (Cd BCF under vetch, 165 ± 52) carrying a 30%-of-mean standard error, none of the plant-side differences would survive multiple-comparison scrutiny.


Provenance: full text read via the authors’ submitted version deposited at HAL (hal-05855845, CC BY-NC-ND), tables cross-checked against the published Springer version; every load-bearing number verified verbatim by three independent reviewers. The paper prints soil organic C and N and the green-manure quantities with mg kg⁻¹ units that its own C/N ratio shows must be g kg⁻¹; we quote none of those. See /methodology/ for how sources are chosen and checked.

Primary source: https://doi.org/10.1007/s42729-026-03626-6

← All news & analysis