Academic Fraud Review: Soil sterilization accumulates higher content of antibacterial compound and recruits beneficial fungus for konjac soft rot disease management by multi-omics analyses (DOI: 10.1186/s12870-025-07985-7)

Academic fraud report · Geng Detector

Summary

Verdict: Highly suspicious. This review of a 2025 BMC Plant Biology paper by Nong et al. identifies multiple serious flaws that undermine the validity of its core conclusions. The most critical issue is a fatal logical error in the metabolomics design: diseased soils from three distinct fields (D1, D2, D3) were pooled before autoclaving into a single sterilized sample (S), yet each individual field sample was subsequently compared against S. The resulting 'differences' cannot be attributed to sterilization versus soil heterogeneity, invalidating downstream metabolomics claims. A second confirmed issue involves a 'ghost pathogen': the methods describe inoculation with P. carotovorum only, yet results (Fig. 4a–b) present inhibition data for P. aroidearum with no corresponding methodology. Additional concerns include misuse of Student's t-test for multi-group comparisons in Fig. 4, inconsistent solvent controls between pot and transcriptomic experiments, and inability to verify image integrity in plain-text format. Confidence is high for findings 1–3; image-based checks were not possible. Final determination of misconduct requires institutional investigation.

Verdict

Highly suspicious. The paper contains at least one confirmed fatal flaw in experimental design and one confirmed data inconsistency. The core metabolomics conclusions are unreliable due to a logical error in how disease soils (D1, D2, D3) and sterilized soil (S) were generated and compared.

Key findings

Fatal design flaw in metabolomics: Diseased soils from three distinct fields were mixed and autoclaved together to produce a single pooled sterilized soil (S). Each individual field soil (D1, D2, D3) was then compared against this pooled sample, conflating sterilization effects with between-field soil heterogeneity.
Ghost pathogen (P. aroidearum): Methods describe inoculation with *P. carotovorum* (Pcc) only at ~1.4 × 10⁹ CFU/mL, but results (Fig. 4a, 4b) present inhibitory activity data for *P. aroidearum* (Pa) without any corresponding methods description.
Statistical misuse: Figure 4 compares multiple concentration groups (1, 10, 100 µg/mL plus control) but the legend states "Significant differences were calculated by Student's t-test," which is inappropriate for ≥3 groups and inflates Type I error. The authors correctly use ANOVA with Duncan's test in Figure 8, indicating awareness of the correct method.
Inconsistent solvent controls: Pot experiments with 1,4-naphthoquinone omit any DMSO control, whereas transcriptomic experiments later acknowledge DMSO. 1,4-naphthoquinone is poorly water-soluble and typically requires an organic co-solvent.
Image-based checks not possible: Plain-text format precluded evaluation of image reuse or manipulation.
Timeline plausibility: Submission and acceptance timeline (~8 months) is consistent with normal BMC Plant Biology practice; instruments described (Thermo UHPLC-Q Exactive HF-X, Illumina NovaSeq 6000) are standard.

Evidence highlights

The methods explicitly state that diseased soils "from the three distinct fields were mixed together and underwent autoclaving" to produce the sterilized soil. The results section describes comparing "2,206 metabolites … in all disease soils … (D1, D2, and D3) and the pooled sterilized soil (S)."
Methods (Inhibitory experiments) mention only Pcc inoculation at ~1.4 × 10⁹ CFU/mL; Fig. 4a–b present data for Pa with no methodological basis provided.
Fig. 4 legend states "Significant differences were calculated by Student's t-test" for a multi-group comparison.
Methods for the pot experiment specify a control lacking only the 1,4-naphthoquinone additive, without mention of the solvent carrier used elsewhere in the study.

Notes

The reviewer's assessment is constrained to text-level evidence; image forensics, raw data inspection, and replication checks are not possible here.
Concerns 1 and 2 are flagged as confirmed (实锤); concerns 3 and 4 are flagged as highly suspicious; concern 6 is flagged as no anomaly.
Definitive attribution of misconduct requires institutional review of original lab records, raw data, and image files. Recommended follow-up: request raw experimental records (especially Pa inoculation details), post concerns on PubPeer, and report the core design defect to the journal editorial office.
DOI of the paper under review: 10.1186/s12870-025-07985-7.
This summary is for academic discussion only and does not constitute a formal finding of misconduct.