Two bacterial strains that compete fiercely in isolation can appear mutualistic inside a diverse community. This is not an artifact of measurement. The community itself generates indirect feedbacks – mediated through surrounding species – that are as strong as the direct competition between the strains. These community-mediated effects create both equalizing mechanisms (reducing fitness differences) and stabilizing mechanisms (generating negative frequency dependence), and the dynamics of any strain pair can be captured by a small number of emergent parameters rather than the full interaction matrix.

The implication is that pairwise relationships measured outside their community context can be not merely imprecise but directionally wrong. A competitive interaction in isolation becomes a facilitative interaction in context, and the context is not noise; it is mechanism. The surrounding community acts as a transformer – it takes the raw interaction between two strains and reshapes it through the network of indirect effects until the sign of the relationship reverses. The strains have not changed. The community has changed what their interaction means.

This matters for any system where components are characterized in isolation and then assembled. The behavior of a pair depends on what else is present, and “what else is present” is not a background condition but an active participant in determining the relationship. Ecological context does not merely modulate interactions; it can manufacture them.

(arXiv:2603.15052)