Review Papers Claim by Claim

1. What the paper actually shows (supported claims): The authors applied standard 16S V3–V4 amplicon sequencing and QIIME2/DADA2 workflows and report Proteobacteria and Firmicutes as dominant phyla in both husbandry types, Enterobacteriaceae as the top family, and a larger species count in non-caged ducks; they report 851 shared species between groups Original methods & numeric results
2. Major methodological omissions that limit inference:
   • No explicit sample size/replicate counts: the paper gives inclusion criteria but does not report how many individual ducks or samples per group — this prevents statistical testing of differences and estimation of variance (a critical reproducibility gap) Missing sample-size and data deposition
   • No raw-data accession or deposited FASTQ / ASV tables: Without accession numbers (SRA/ENA/NCBI) we cannot reprocess, confirm read counts, or reassign taxonomy with current databases such as SILVA or GTDB.
   • Greengenes 13_8 (used here) is dated (last full update ~2013); for higher taxonomic resolution and current naming, reanalysis with SILVA or GTDB is advisable — Greengenes can misassign/underrepresent lineages discovered after release.
   • Statistical comparisons are not reported (no p-values, confidence intervals, or multivariate tests shown): the text says family-level differences are "not significant" but no test or p-values are shown, so readers cannot evaluate effect sizes or significance.
3. Interpretation concerns and confounders:
   • The authors attribute similarity to "similar feed and environment" but feed data are coarse (Table 1) and not quantitatively controlled — feed composition, supplemental feed frequency, age distribution, host genetics, and husbandry density are all unmeasured confounders that can explain differences.
   • Use of 16S amplicon data limits functional inference (no shotgun metagenome or metatranscriptome data). SCFA identification via GC–MS is mentioned but linkage between taxa and SCFA production is speculative without correlation statistics or functional gene data.
4. What can be credibly concluded:
   • Within this set of samples from one Indonesian district, Proteobacteria and Firmicutes were numerically dominant and the authors detected more taxa reported in non-caged ducks than in caged ducks; 851 species were common to both groups. These are descriptive, site-specific observations, useful as pilot data for hypothesis generation but not for causal claims.
5. Immediate reanalysis / reproducibility steps I recommend:
   1. Require authors to publish raw FASTQ and sample metadata (sample ID, cage/NCG label, sex, age, diet details, collection date) to SRA/ENA with accession numbers.
   2. Reprocess reads with an up-to-date pipeline: quality trimming (e.g., fastp), DADA2 (or deblur), assign taxonomy against SILVA 138 or GTDB, and compute alpha/beta diversity with rarefaction and appropriate statistics (PERMANOVA, ANCOM/ALDEx2 for differential taxa) to test reported differences.
   3. If raw reads are unavailable, request ASV/OTU tables and mapping files; otherwise treat numeric values in the paper as provisional.
6. Potential biases and broader limitations (explicit):
   • Purposive sampling, single-region study, and lack of randomization introduce selection bias and limit generality beyond Gunungpati.
   • Conference-paper format often compresses methods; this work reads as pilot/exploratory rather than final, peer-reviewed ecological microbiome analysis.
   • Use of Greengenes and Excel for taxa visualization is outdated compared with modern reproducible tools (R/phyloseq, QIIME2 artifacts with provenance).
7. Scientific value and next experiments:
   • Value: Provides first-pass, local descriptive data on duck gut microbiomes from small-holder Indonesian farms and a testable observation that non-caged ducks may host higher alpha-diversity.
   • Next: a controlled study with recorded, standardized diets (or matched diet arms), clear sample sizes (n per group based on power analysis), longitudinal sampling, shotgun metagenomics for functional profiling, and SCFA quantitation correlated with taxa and functional genes (e.g., butyryl-CoA:acetate CoA-transferase genes) would move from descriptive to mechanistic inference.

Where a different set of results would falsify the paper's main claim

If a study that properly controls diet, age, season, and rearing micro-environment (same feed, paired farms, randomized sampling) finds no consistent difference in alpha diversity or taxonomic composition between caged and non-caged ducks, that would falsify the paper's suggestion that husbandry type is a primary driver.

Practical checklist for reanalysis (short)

• Obtain raw reads + metadata; report per-sample read counts and rarefaction curves.
• Reassign taxonomy with SILVA/GTDB; produce ASV table, alpha/beta stats, and effect-size estimates (Cohen's d or log2fold change).
• Run differential abundance with compositional-aware methods (ANCOM2, ALDEx2).
• Correlate SCFA levels with taxa and functional gene proxies (PICRUSt2 or shotgun data).

Selected citation for the original paper (for each claim above)

Original conference paper used as primary source:

Susanti et al., 2020 (original)

If you want me to reprocess these data (re-run denoising, reassign taxonomy to SILVA/GTDB, compute stats, and produce publication-grade figures), provide the raw FASTQ or ask me to start a reproducible reanalysis run.