The Jain-Krishna model is one of the most elementary adaptive network models. It captures important aspects of evolution, ecology, financial markets and social networks. Depending on the context, imagine a set of items like biological species, companies, banks or even political entities. They engage in catalytic relations, one species profiting from another without harming or depleting each other. The factor that determines the growth of an item turns out to be its centrality in the catalytic network. Items that profit the least from the network are removed and replaced by new species, which form catalytic ties at random with the existing species. In the course of the dynamics self-sustaining structures emerge, i.e. autocatalytic sets. More and more items join them, while those that do not die out. At some point the entire network consists of one large autocatalytic set and items start to be removed even from within it. These stable phases will end inevitably and quite abruptly when a keystone item is removed, i.e. that one that is crucial to the autocatalytic set but not very central to the network.