An individual-based model of camelid social structure starts from a simple premise: each female joins the group that maximizes her fitness, given local conditions — forage quality, predation risk, group size. No centralized coordination. No blueprint for harem structure. Each animal makes a locally optimal choice about which male's territory to occupy.
The result is population-level structure — characteristic group sizes, predictable fitness distributions, spatial organization — that looks designed. Rebaudo and Cuzol show that these emergent properties aren't additive combinations of individual decisions. Change the sex ratio, and group structure shifts non-linearly. Change habitat heterogeneity, and the distribution of group sizes reorganizes in ways that no single female intended.
The standard reading: bottom-up emergence, the usual surprise that complex patterns arise from simple rules.
The reading that interests me: the structure is real but no one is maintaining it. Each female is optimizing for herself. The population structure is a side effect of those optimizations happening in a shared, heterogeneous landscape. If you showed the resulting harem distribution to an ecologist without the model, they might infer a selection pressure favoring that particular group size, or a social instinct driving females toward particular configurations. They would be wrong. The structure is caused by local optimization in a variable environment, not by selection for the structure itself.
This is the inverse of the diagnostic error in “The Slowing Door.” There, an observed property (low turnover) was mistakenly attributed to the wrong mechanism (resilience instead of depletion). Here, an observed property (social structure) is mistakenly attributed to the wrong level (population-level selection instead of individual optimization). The error in both cases: inferring mechanism from pattern. The pattern is real. The inferred mechanism is wrong.
The general form: when agents optimize locally in a heterogeneous landscape, the global pattern that emerges carries no information about whether the pattern was selected for. You can't distinguish “this structure exists because it was favored” from “this structure exists because individual choices in variable terrain produce it as a side effect.” The observation of structure tells you nothing about whether there is a structuring force.
This matters beyond camelids. Any system of agents making local choices in a complex landscape will generate apparent structure. The structure is not illusory — the harem sizes are real, the spatial distribution is measurable — but the explanation of the structure is wrong if it posits a force maintaining it. The maintenance is an artifact of continued local optimization, not a property of the structure itself.