Two species help each other. Classic mutualism — bees and flowers, clownfish and anemones. The standard model treats the interaction as fixed: mutualistic at all densities, with the benefit saturating as populations grow. Mian et al. drop this assumption.
In their model, the sign of the interaction depends on density. At low density, species A helps species B. At high density, the same interaction becomes parasitic — A harms B. The relationship isn't defined by the species identities but by how crowded the neighborhood is.
The central result: limit cycles — sustained population oscillations — arise when the interaction includes a parasitic phase. In purely mutualistic regimes, no matter how the parameters are tuned, oscillations don't occur. The system settles to a fixed point or diverges. Oscillation requires the relationship to flip.
This is a structural result, not a parameter-sensitivity result. You can't tune a purely mutualistic system into oscillations. The qualitative behavior — cycling vs equilibrium — depends on whether the interaction ever turns negative. The boundary between mutualism and parasitism is a boundary between dynamical regimes.
The biological intuition makes sense once you see it. Pure mutualism is self-reinforcing: more of species A helps species B, which helps species A. This positive feedback drives the system toward a stable state or toward runaway growth. There's no mechanism to reverse direction. But when crowding flips the interaction to parasitism, the positive feedback becomes negative feedback at high density. Growth triggers suppression. Suppression reduces density. Low density restores mutualism. The system cycles.
The implication is that the oscillations we observe in mutualistic systems — pollinator population fluctuations, coral-algae cycles — may not be noise or external forcing. They may be structural signatures that the relationship turns parasitic at high density. The oscillation is diagnostic: if you see it, the interaction flipped.
This reframes a question. Instead of asking “is this relationship mutualistic or parasitic?” — asking “at what density does it switch?” The relationship is both, at different population states. The identity of the interaction is not a property of the species but a property of the configuration.