Marine microbiomes: what if statistical analyses were inventing randomness?

The hypothesis is formulated in a falsifiable manner with precise quantitative bounds (a reduction of 20–40 percentage points, a variance ratio ≥1.5, Δm ≤ −0.05, and so forth), which permits an objective evaluation of the results.

The hypothesis identifies a genuine and underexploited statistical inconsistency in the microbial ecology literature: the application of βNTI and Sloan neutral models to relative abundances without prior transformation does indeed violate simplex geometry, and the transfer of CoDA reasoning (Aitchison, 1982; Gloor et al., 2017) to the inference of assembly mechanisms is conceptually legitimate and non-trivial.

The issue of compositional bias in amplicon data is genuine and well documented; the hypothesis addresses an authentic source of bias in assembly-based inferences.

The addressable market is real: metagenomics and environmental surveillance platforms (Qiagen, Illumina via DRAGEN, Zymo Research, as well as CROs such as CosmosID or Resphera) already sell microbial assembly analysis pipelines to academic clients, environmental agencies (NOAA, Ifremer) and industrial clients (aquaculture, marine biotech). A methodological fix validated on global marine data (Tara Oceans, Ocean Sampling Day) could become a premium module within these pipelines, priced at €5–15k per annual licence or €2–5k per project.

High-impact cross-cutting methodological hypothesis: the correction of the closure effect by CLR potentially affects thousands of studies in marine microbial ecology, which provides a highly marketable 'meta-science' narrative for methodological and open science calls.