Concise verdict

Shaposhnikov & Matveeva (2023) provide an updated bioinformatic survey arguing that Agrobacterium-to-plant horizontal gene transfer (HGT) is common: they report >50 newly detected naturally transgenic (nGMO) species in 2023 and a cumulative total >100 nGMO species (~7% of deposited terrestrial dicots). These claims are consistent with earlier systematic bioinformatic findings (Matveeva & Otten 2019) but depend critically on detection criteria, database composition, and contamination controls — which the paper states only qualitatively and does not provide as reproducible code or dataset links, lowering reproducibility and requiring validation by transparent pipelines and phylogenetic/synteny checks Shaposhnikov & Matveeva 2023 (nGMO update)Matveeva & Otten 2019 (foundational nGMO survey)

Run AI Scientist Analysis

Visual paper analysis — "New naturally transgenic crops" (10.17816/ecogen568608)

Data points: 2019 ≈ 30 species (Matveeva & Otten 2019); 2023 >100 species reported by Shaposhnikov & Matveeva (2023) after new database scans Matveeva & Otten 2019Shaposhnikov & Matveeva 2023

The paper lists several cultivated fruit and vegetable species among newly annotated nGMOs (carambola, persimmon, wasabi, raspberry, Luffa) — presence/absence is reported but functional impact of insertions is not experimentally shown Shaposhnikov & Matveeva 2023 (examples)

Critical appraisal (evidence-based, skeptical)

1. Strengths
   • Updates an active line of inquiry: extends the 2019 survey and reports a substantially larger set of candidate nGMO taxa, which is important for macroevolutionary and crop-trait origin questions Matveeva & Otten 2019 (methodological backbone).
   • Highlights cultivated species as nGMO candidates, prompting testable hypotheses about domestication/trait origins.
2. Primary limitations / red flags
   • No public code, accession list, or per-hit alignments are provided in the manuscript: this prevents reproducibility and independent reanalysis (authors note database-based methods but do not release data) Shaposhnikov & Matveeva 2023 (no data).
   • Distinguishing authentic HGT from contamination, assembly error, or symbiont DNA requires multiple orthogonal checks (phylogenetic incongruence, synteny, flanking-plant sequences, PCR validation) — the paper does not present these validations for the new hits, reducing confidence Matveeva & Otten 2019 (validation requirements).
   • Sampling bias: proportion (~7%) is given relative to 'deposited terrestrial dicotyledonous plants' — database composition (over-representation of some taxa) can inflate or deflate that figure; the authors note the stability of ~7% across database growth but without raw counts that can be re-evaluated.
   • Functional claims absent: listing species with T-DNA-like sequences is not the same as demonstrating selection or trait causation; no transcriptomics/protein expression or phenotypic association analyses are included for the new candidates.

What would increase confidence (actionable checklist)

• Release a table of candidate hits: accession IDs, coordinates, alignment statistics (percent identity/coverage), and flanking plant sequence context.
• Provide phylogenetic trees per candidate locus showing the Agrobacterium origin and incongruence with host gene trees.
• Show synteny and assembly validation: long-read mapping, PCR/amplicon sequencing from independent samples, and contamination controls.
• Report transcript evidence (RNA-seq) demonstrating expression, and where possible, functional annotation of inserted sequence ORFs.

Practical implications and next steps

If robust, the claim that Agrobacterium-mediated HGT contributed DNA to many cultivated species would reshape how we think about gene acquisition during plant evolution and domestication. However, the current paper is a survey announcement: it generates many useful leads but cannot by itself prove functional or adaptive impact of the insertions without the orthogonal validations above Shaposhnikov & Matveeva 2023 (interpretation).

Immediate reproducibility test (practical experiment)

Re-run the screening pipeline publicly: (1) BLAST/DIAMOND of Agrobacterium TDNA genes vs latest plant WGS/TSAs, (2) extract hit coordinates and flanking 5–10 kb, (3) build ML-based contamination filters (e.g., read-depth, GC-content), (4) construct per-locus phylogenies and synteny maps, (5) submit candidate loci and scripts to a public repository — this single pipeline would settle whether many new candidates survive strict filters.

Direct citations used in this review

• Shaposhnikov & Matveeva 2023
• Matveeva & Otten 2019

Confidence note: claims in the paper are plausible and build on prior high-quality work, but lack of released data/methods means independent verification is required; treat the reported counts as candidate-level, not yet fully validated HGT events.