Separate LBNL soil SynCom assay variants from final 16-member graph claims #1547

Finding

kb/communities/LBNL_Switchgrass_Soil_SynCom16.yaml blends multiple experimental compositions in PMID:36472419. Nutrient dilution used all 18 starting strains; EcoFAB experiments used 17 or 18 isolates, while the final 16-member community excludes Sphingomonas and Pseudomonas. Automated assembly and glycerol storage are procedures, not ecological causal nodes. Root colonization/composition is not a demonstrated mutual plant benefit.

Repair

Remove two procedure nodes while documenting their valid methodological results. Retain nutrient-diversity, initial-ratio diversity and host colonization observations with explicit assay-specific membership and time bounds. Do not infer viability or absolute population growth from relative sequencing, generalize in-vitro Pseudomonas effects to EcoFAB, or add arrows between separate experiments. Classify live-plant evidence as IN_VIVO and replace the unproven symbiosis annotation.

Full fresh primary XML body including Methods, Discussion and embedded captions read, not separate supplements, original image pixels or raw data. Keep the canonical aggregate placeholder unchanged but explicit; unresolved roster metadata is separate. Self-adversarial review only. Close after protected-main merge.