Predicting cascading extinctions and efficient restoration strategies in ecological networks

The ecologically important task of predicting the severity of cascading extinctions is made challenging by the complexity of ecological networks. In this work, we study an ensemble of network models that describe mutualistic inter-species interactions by Boolean threshold functions. We demonstrate that identifying generalized positive feedback loops (stable motifs) helps pinpoint the species whose extinction leads to catastrophic damage to the whole community. We compare stable motif-based results with previously studied network structural measures and show that stable motifs can identify certain crucial species that the other measures fail to find. We also use the stable motifs of the Boolean model to propose mitigation measures to 1. prevent the damage to the community by protecting a subset of the species, 2. restore the community after the damage by restoring a subset of species. The analysis in this work indicates that the stable motifs predict the most fruitful strategies to manage ecological systems. This approach can also be implemented in other complex systems to achieve the desired outcome.