Correct seawater-ion graph typing and treatment-specific causal scope #1342

Source-Based Graph Findings

CommunityMech:000258, Coastal_Forested_Wetland_Seawater_Ion_Microcosm_Community.yaml.

Primary full text, Methods, main table and seven captions reviewed: https://doi.org/10.1093/ismeco/ycae040.

• A manipulated ion treatment and its measured responses are not NICHE_PARTITIONING; the sulfate-only null response is not COMPETITION.
• Retain both treatment-to-response arrows and the positive ion controls, but separate abiotic intervention effects from untested biological mediators.
• Functional genes and guilds are relative DNA-abundance measurements, not expression, absolute growth, or process activity. Community-level measurements have no causal donor taxon.
• Bound the sulfate null conclusion to the tested Week-12 endpoints; the Day-14 sulfide observation prevents a blanket claim that sulfate reduction was inactive.
• Distinguish factorial effects from specific pairwise comparisons for methane, carbon dioxide, nitrous oxide and diversity. Do not infer a measured gene-to-emission causal link.

All four nodes and both arrows require explicit decisions. No new structure is proposed. #644 already fixed different incubation-atmosphere/control assertions and must not be reopened for this graph issue.

Deduplication includes 678 all-state issue titles/bodies and ignored/hidden local files; #644, #642, #524 and #312 read. No duplicate exact graph finding in that scope. Comments not exhaustive.