Academic Fraud Investigation Report: Optimization of Clostridium butyricum Fermentation Medium and Degradation of Aflatoxin B1 by Fermentation Products

Academic fraud report · Geng Detector

Summary

Verdict: Highly Suspicious (Orange). This 2023 paper published in 微生物学通报 (DOI: 10.13344/j.microbiol.china.230086) by Liu Yamei, Cong Lina, and Chen Ming presents multiple serious methodological and data-integrity issues. The most critical flaw is a fatal logical contradiction in the degradation rate calculation: the methods state the test group received AFB1 at a final concentration of 40 ng/mL while the control contained 400 ng/mL, which, if applied to the reported formula, would yield an apparent 90% degradation rate at zero time even with no actual degradation. Additional concerns include implausibly smooth data in Figure 4 (three time points across 24 hours showing near-identical degradation rates with negligible error bars), visually overlapping error bars in Figure 6 between fermentation broth and supernatant groups, and self-contradictory descriptions of response surface contour plots. Confidence is moderate-to-high based on the unambiguous arithmetic error and visual anomalies, though final determination requires access to raw OD values and laboratory notebooks from the authors.

Verdict

Highly Suspicious. The paper contains at least one clear methodological/arithmetical error of a magnitude that any peer reviewer should have caught, combined with data presentation patterns highly atypical of genuine biological replicates.

Key findings

Critical calculation logic error (40 vs 400 ng/mL): The methods section states test samples received AFB1 to a final concentration of 40 ng/mL while the control contained 400 ng/mL. Substituting these into the reported degradation rate formula yields an apparent 90% baseline degradation with zero actual activity. Either the formula's premises are violated or the methods text contains a transcription/copy-paste error of serious magnitude.
Implausibly perfect plateau in Figure 4: Degradation rates at 24 h, 36 h, and 48 h are described as "essentially identical" with error bars too small to be visible across a 24-hour biological span, violating expected stochastic variability.
Overlapping error bars in Figure 6: Whole-fermentation broth (54.56%) and supernatant (53.50%) data points overlap visually, lacking the normal dispersion expected between two independent treatment groups.
Self-contradictory statistical description: The text defines "slightly elliptical" contours as indicating non-significant interaction, then applies this description to the X1×X4 interaction (P = 0.687, indeed non-significant), inverting the conventional response-surface methodology convention (elliptical = significant, circular = non-significant).

Evidence highlights

DOI: 10.13344/j.microbiol.china.230086
Reported concentrations: test 40 ng/mL vs. control 400 ng/mL (Sections 1.2.6 / 1.2.7)
Formula (1): Degradation rate = (1 − test content / control content) × 100%
Figure 4 time points: 12, 24, 36, 48 h; reported identical degradation at 24, 36, 48 h
Figure 6 values: fermentation broth 54.56%, supernatant 53.50%
X1×X4 interaction P-value: 0.687

Notes

The 40 vs 400 ng/mL discrepancy is the single most damaging finding: if the methods text is correct, the entire degradation dataset is mathematically invalidated; if it is a typo, it represents a careless editing error that nonetheless escaped peer review.
Visual analysis summaries cited in the source report were generated by automated tooling and should be confirmed by direct figure inspection.
Final determination of misconduct requires the authors' raw ELISA absorbance values, standard curves, and original laboratory records.
Confidence in the calculation-error finding: high. Confidence in the data-fabrication suspicion (Findings 2–3): moderate, pending raw-data review.